add

work_dir_h/PWOOD/a_noise/0.1/20250307_082134.log

@@ -0,0 +1,558 @@
+2025-03-07 08:21:34,598 - mmrotate - INFO - Environment info:
+------------------------------------------------------------
+sys.platform: linux
+Python: 3.8.0 (default, Nov  6 2019, 21:49:08) [GCC 7.3.0]
+CUDA available: True
+GPU 0,1: NVIDIA GeForce RTX 3090
+CUDA_HOME: /usr/local/cuda
+NVCC: Cuda compilation tools, release 11.3, V11.3.58
+GCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0
+PyTorch: 1.13.1+cu116
+PyTorch compiling details: PyTorch built with:
+  - GCC 9.3
+  - C++ Version: 201402
+  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications
+  - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)
+  - OpenMP 201511 (a.k.a. OpenMP 4.5)
+  - LAPACK is enabled (usually provided by MKL)
+  - NNPACK is enabled
+  - CPU capability usage: AVX2
+  - CUDA Runtime 11.6
+  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86
+  - CuDNN 8.3.2  (built against CUDA 11.5)
+  - Magma 2.6.1
+  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.6, CUDNN_VERSION=8.3.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, 
+
+TorchVision: 0.14.1+cu116
+OpenCV: 4.11.0
+MMCV: 1.7.2
+MMCV Compiler: GCC 9.3
+MMCV CUDA Compiler: 11.6
+MMRotate: 0.3.4+7833b87
+------------------------------------------------------------
+
+2025-03-07 08:21:35,870 - mmrotate - INFO - Distributed training: True
+2025-03-07 08:21:37,280 - mmrotate - INFO - Config:
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='SemiGMMLoss', cls_channels=16, policy='high'),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='AddNoise', p=0.1),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='AddNoise', p=0.1),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+custom_imports = dict(imports=['semi_mmrotate'], allow_failed_imports=False)
+work_dir = 'work_dir/h2rv2_mcl/a_noise/0.1/'
+auto_resume = False
+gpu_ids = range(0, 2)
+
+2025-03-07 08:21:37,281 - mmrotate - INFO - Set random seed to 42, deterministic: True
+2025-03-07 08:21:56,941 - mmrotate - INFO - Start running, host: yangxue@think4proj-105-234, work_dir: /mnt/nas2/home/yangxue/lmx/sood-mcl/work_dir/h2rv2_mcl/a_noise/0.1
+2025-03-07 08:21:56,942 - mmrotate - INFO - Hooks will be executed in the following order:
+before_run:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(NORMAL      ) CheckpointHook                     
+(NORMAL      ) MeanTeacher                        
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_train_epoch:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(NORMAL      ) NumClassCheckHook                  
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_train_iter:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+ -------------------- 
+after_train_iter:
+(ABOVE_NORMAL) OptimizerHook                      
+(NORMAL      ) CheckpointHook                     
+(NORMAL      ) MeanTeacher                        
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+after_train_epoch:
+(NORMAL      ) CheckpointHook                     
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_val_epoch:
+(NORMAL      ) NumClassCheckHook                  
+(LOW         ) IterTimerHook                      
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_val_iter:
+(LOW         ) IterTimerHook                      
+ -------------------- 
+after_val_iter:
+(LOW         ) IterTimerHook                      
+ -------------------- 
+after_val_epoch:
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+after_run:
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+2025-03-07 08:21:56,942 - mmrotate - INFO - workflow: [('train', 1)], max: 120000 iters
+2025-03-07 08:21:56,942 - mmrotate - INFO - Checkpoints will be saved to /mnt/nas2/home/yangxue/lmx/sood-mcl/work_dir/h2rv2_mcl/a_noise/0.1 by HardDiskBackend.

work_dir_h/PWOOD/a_noise/0.1/20250307_082134.log.json

@@ -0,0 +1 @@
+{"env_info": "sys.platform: linux\nPython: 3.8.0 (default, Nov  6 2019, 21:49:08) [GCC 7.3.0]\nCUDA available: True\nGPU 0,1: NVIDIA GeForce RTX 3090\nCUDA_HOME: /usr/local/cuda\nNVCC: Cuda compilation tools, release 11.3, V11.3.58\nGCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0\nPyTorch: 1.13.1+cu116\nPyTorch compiling details: PyTorch built with:\n  - GCC 9.3\n  - C++ Version: 201402\n  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications\n  - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)\n  - OpenMP 201511 (a.k.a. OpenMP 4.5)\n  - LAPACK is enabled (usually provided by MKL)\n  - NNPACK is enabled\n  - CPU capability usage: AVX2\n  - CUDA Runtime 11.6\n  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86\n  - CuDNN 8.3.2  (built against CUDA 11.5)\n  - Magma 2.6.1\n  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.6, CUDNN_VERSION=8.3.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, \n\nTorchVision: 0.14.1+cu116\nOpenCV: 4.11.0\nMMCV: 1.7.2\nMMCV Compiler: GCC 9.3\nMMCV CUDA Compiler: 11.6\nMMRotate: 0.3.4+7833b87", "config": "angle_version = 'le90'\ndetector = dict(\n    type='SemiRotatedFCOS',\n    backbone=dict(\n        type='ResNet',\n        depth=50,\n        num_stages=4,\n        out_indices=(0, 1, 2, 3),\n        frozen_stages=1,\n        zero_init_residual=False,\n        norm_cfg=dict(type='BN', requires_grad=True),\n        norm_eval=True,\n        style='pytorch',\n        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),\n    neck=dict(\n        type='FPN',\n        in_channels=[256, 512, 1024, 2048],\n        out_channels=256,\n        start_level=1,\n        add_extra_convs='on_output',\n        num_outs=5,\n        relu_before_extra_convs=True),\n    bbox_head=dict(\n        type='SemiRotatedFCOSHeadH2RV2MCL',\n        num_classes=16,\n        in_channels=256,\n        stacked_convs=4,\n        feat_channels=256,\n        strides=[8, 16, 32, 64, 128],\n        center_sampling=True,\n        center_sample_radius=1.5,\n        norm_on_bbox=True,\n        centerness_on_reg=True,\n        square_cls=[1, 9, 11],\n        resize_cls=[1],\n        scale_angle=False,\n        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),\n        loss_cls=dict(\n            type='FocalLoss',\n            use_sigmoid=True,\n            gamma=2.0,\n            alpha=0.25,\n            loss_weight=1.0),\n        loss_bbox=dict(type='IoULoss', loss_weight=1.0),\n        loss_centerness=dict(\n            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),\n        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),\n    train_cfg=None,\n    test_cfg=dict(\n        nms_pre=2000,\n        min_bbox_size=0,\n        score_thr=0.05,\n        nms=dict(iou_thr=0.1),\n        max_per_img=2000))\nmodel = dict(\n    type='H2RV2MCLTeacher',\n    model=dict(\n        type='SemiRotatedFCOS',\n        backbone=dict(\n            type='ResNet',\n            depth=50,\n            num_stages=4,\n            out_indices=(0, 1, 2, 3),\n            frozen_stages=1,\n            zero_init_residual=False,\n            norm_cfg=dict(type='BN', requires_grad=True),\n            norm_eval=True,\n            style='pytorch',\n            init_cfg=dict(\n                type='Pretrained', checkpoint='torchvision://resnet50')),\n        neck=dict(\n            type='FPN',\n            in_channels=[256, 512, 1024, 2048],\n            out_channels=256,\n            start_level=1,\n            add_extra_convs='on_output',\n            num_outs=5,\n            relu_before_extra_convs=True),\n        bbox_head=dict(\n            type='SemiRotatedFCOSHeadH2RV2MCL',\n            num_classes=16,\n            in_channels=256,\n            stacked_convs=4,\n            feat_channels=256,\n            strides=[8, 16, 32, 64, 128],\n            center_sampling=True,\n            center_sample_radius=1.5,\n            norm_on_bbox=True,\n            centerness_on_reg=True,\n            square_cls=[1, 9, 11],\n            resize_cls=[1],\n            scale_angle=False,\n            bbox_coder=dict(\n                type='DistanceAnglePointCoder', angle_version='le90'),\n            loss_cls=dict(\n                type='FocalLoss',\n                use_sigmoid=True,\n                gamma=2.0,\n                alpha=0.25,\n                loss_weight=1.0),\n            loss_bbox=dict(type='IoULoss', loss_weight=1.0),\n            loss_centerness=dict(\n                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),\n            loss_ss_symmetry=dict(\n                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),\n        train_cfg=None,\n        test_cfg=dict(\n            nms_pre=2000,\n            min_bbox_size=0,\n            score_thr=0.05,\n            nms=dict(iou_thr=0.1),\n            max_per_img=2000)),\n    semi_loss=dict(type='SemiGMMLoss', cls_channels=16, policy='high'),\n    train_cfg=dict(\n        iter_count=0,\n        burn_in_steps=12800,\n        sup_weight=1.0,\n        unsup_weight=1.0,\n        weight_suppress='exp',\n        logit_specific_weights=dict(),\n        cls_channels=16),\n    test_cfg=dict(inference_on='teacher'))\nimg_norm_cfg = dict(\n    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ncommon_pipeline = [\n    dict(\n        type='Normalize',\n        mean=[123.675, 116.28, 103.53],\n        std=[58.395, 57.12, 57.375],\n        to_rgb=True),\n    dict(type='Pad', size_divisor=32),\n    dict(type='DefaultFormatBundle'),\n    dict(\n        type='Collect',\n        keys=['img', 'gt_bboxes', 'gt_labels'],\n        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',\n                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',\n                   'img_norm_cfg', 'tag'))\n]\nstrong_pipeline = [\n    dict(type='DTToPILImage'),\n    dict(\n        type='DTRandomApply',\n        operations=[\n            ColorJitter(\n                brightness=[0.6, 1.4],\n                contrast=[0.6, 1.4],\n                saturation=[0.6, 1.4],\n                hue=[-0.1, 0.1])\n        ],\n        p=0.8),\n    dict(type='DTRandomGrayscale', p=0.2),\n    dict(\n        type='DTRandomApply',\n        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),\n    dict(type='DTToNumpy'),\n    dict(type='ExtraAttrs', tag='unsup_strong')\n]\nweak_pipeline = [\n    dict(type='RResize', img_scale=(1024, 1024)),\n    dict(\n        type='RRandomFlip',\n        flip_ratio=[0.25, 0.25, 0.25],\n        direction=['horizontal', 'vertical', 'diagonal'],\n        version='le90'),\n    dict(type='ExtraAttrs', tag='unsup_weak')\n]\nunsup_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(type='LoadEmptyAnnotations', with_bbox=True),\n    dict(\n        type='STMultiBranch',\n        unsup_strong=[\n            dict(type='DTToPILImage'),\n            dict(\n                type='DTRandomApply',\n                operations=[\n                    ColorJitter(\n                        brightness=[0.6, 1.4],\n                        contrast=[0.6, 1.4],\n                        saturation=[0.6, 1.4],\n                        hue=[-0.1, 0.1])\n                ],\n                p=0.8),\n            dict(type='DTRandomGrayscale', p=0.2),\n            dict(\n                type='DTRandomApply',\n                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,\n                                                                   2.0])]),\n            dict(type='DTToNumpy'),\n            dict(type='ExtraAttrs', tag='unsup_strong')\n        ],\n        unsup_weak=[\n            dict(type='RResize', img_scale=(1024, 1024)),\n            dict(\n                type='RRandomFlip',\n                flip_ratio=[0.25, 0.25, 0.25],\n                direction=['horizontal', 'vertical', 'diagonal'],\n                version='le90'),\n            dict(type='ExtraAttrs', tag='unsup_weak')\n        ],\n        common_pipeline=[\n            dict(\n                type='Normalize',\n                mean=[123.675, 116.28, 103.53],\n                std=[58.395, 57.12, 57.375],\n                to_rgb=True),\n            dict(type='Pad', size_divisor=32),\n            dict(type='DefaultFormatBundle'),\n            dict(\n                type='Collect',\n                keys=['img', 'gt_bboxes', 'gt_labels'],\n                meta_keys=('filename', 'ori_filename', 'ori_shape',\n                           'img_shape', 'pad_shape', 'scale_factor', 'flip',\n                           'flip_direction', 'img_norm_cfg', 'tag'))\n        ],\n        is_seq=True)\n]\nsup_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(type='LoadAnnotations', with_bbox=True),\n    dict(type='AddNoise', p=0.1),\n    dict(type='RResize', img_scale=(1024, 1024)),\n    dict(\n        type='RRandomFlip',\n        flip_ratio=[0.25, 0.25, 0.25],\n        direction=['horizontal', 'vertical', 'diagonal'],\n        version='le90'),\n    dict(type='ExtraAttrs', tag='sup_weak'),\n    dict(\n        type='Normalize',\n        mean=[123.675, 116.28, 103.53],\n        std=[58.395, 57.12, 57.375],\n        to_rgb=True),\n    dict(type='Pad', size_divisor=32),\n    dict(type='DefaultFormatBundle'),\n    dict(\n        type='Collect',\n        keys=['img', 'gt_bboxes', 'gt_labels'],\n        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',\n                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',\n                   'img_norm_cfg', 'tag'))\n]\ntest_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(\n        type='MultiScaleFlipAug',\n        img_scale=(1024, 1024),\n        flip=False,\n        transforms=[\n            dict(type='RResize'),\n            dict(\n                type='Normalize',\n                mean=[123.675, 116.28, 103.53],\n                std=[58.395, 57.12, 57.375],\n                to_rgb=True),\n            dict(type='Pad', size_divisor=32),\n            dict(type='DefaultFormatBundle'),\n            dict(type='Collect', keys=['img'])\n        ])\n]\ndataset_type = 'DOTAv15WSOODDataset'\nclasses = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n           'basketball-court', 'storage-tank', 'soccer-ball-field',\n           'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n           'container-crane')\ndata = dict(\n    samples_per_gpu=3,\n    workers_per_gpu=5,\n    train=dict(\n        type='SemiDataset',\n        sup=dict(\n            type='DOTAv15WSOODDataset',\n            pipeline=[\n                dict(type='LoadImageFromFile'),\n                dict(type='LoadAnnotations', with_bbox=True),\n                dict(type='AddNoise', p=0.1),\n                dict(type='RResize', img_scale=(1024, 1024)),\n                dict(\n                    type='RRandomFlip',\n                    flip_ratio=[0.25, 0.25, 0.25],\n                    direction=['horizontal', 'vertical', 'diagonal'],\n                    version='le90'),\n                dict(type='ExtraAttrs', tag='sup_weak'),\n                dict(\n                    type='Normalize',\n                    mean=[123.675, 116.28, 103.53],\n                    std=[58.395, 57.12, 57.375],\n                    to_rgb=True),\n                dict(type='Pad', size_divisor=32),\n                dict(type='DefaultFormatBundle'),\n                dict(\n                    type='Collect',\n                    keys=['img', 'gt_bboxes', 'gt_labels'],\n                    meta_keys=('filename', 'ori_filename', 'ori_shape',\n                               'img_shape', 'pad_shape', 'scale_factor',\n                               'flip', 'flip_direction', 'img_norm_cfg',\n                               'tag'))\n            ],\n            ann_file='data/train_20p_labeled/annfiles/',\n            img_prefix='data/train_20p_labeled/images/',\n            version='le90',\n            classes=('plane', 'baseball-diamond', 'bridge',\n                     'ground-track-field', 'small-vehicle', 'large-vehicle',\n                     'ship', 'tennis-court', 'basketball-court',\n                     'storage-tank', 'soccer-ball-field', 'roundabout',\n                     'harbor', 'swimming-pool', 'helicopter',\n                     'container-crane')),\n        unsup=dict(\n            type='DOTAv15WSOODDataset',\n            pipeline=[\n                dict(type='LoadImageFromFile'),\n                dict(type='LoadEmptyAnnotations', with_bbox=True),\n                dict(\n                    type='STMultiBranch',\n                    unsup_strong=[\n                        dict(type='DTToPILImage'),\n                        dict(\n                            type='DTRandomApply',\n                            operations=[\n                                ColorJitter(\n                                    brightness=[0.6, 1.4],\n                                    contrast=[0.6, 1.4],\n                                    saturation=[0.6, 1.4],\n                                    hue=[-0.1, 0.1])\n                            ],\n                            p=0.8),\n                        dict(type='DTRandomGrayscale', p=0.2),\n                        dict(\n                            type='DTRandomApply',\n                            operations=[\n                                dict(\n                                    type='DTGaussianBlur',\n                                    rad_range=[0.1, 2.0])\n                            ]),\n                        dict(type='DTToNumpy'),\n                        dict(type='ExtraAttrs', tag='unsup_strong')\n                    ],\n                    unsup_weak=[\n                        dict(type='RResize', img_scale=(1024, 1024)),\n                        dict(\n                            type='RRandomFlip',\n                            flip_ratio=[0.25, 0.25, 0.25],\n                            direction=['horizontal', 'vertical', 'diagonal'],\n                            version='le90'),\n                        dict(type='ExtraAttrs', tag='unsup_weak')\n                    ],\n                    common_pipeline=[\n                        dict(\n                            type='Normalize',\n                            mean=[123.675, 116.28, 103.53],\n                            std=[58.395, 57.12, 57.375],\n                            to_rgb=True),\n                        dict(type='Pad', size_divisor=32),\n                        dict(type='DefaultFormatBundle'),\n                        dict(\n                            type='Collect',\n                            keys=['img', 'gt_bboxes', 'gt_labels'],\n                            meta_keys=('filename', 'ori_filename', 'ori_shape',\n                                       'img_shape', 'pad_shape',\n                                       'scale_factor', 'flip',\n                                       'flip_direction', 'img_norm_cfg',\n                                       'tag'))\n                    ],\n                    is_seq=True)\n            ],\n            ann_file='data/train_20p_unlabeled/empty_annfiles/',\n            img_prefix='data/train_20p_unlabeled/images/',\n            version='le90',\n            classes=('plane', 'baseball-diamond', 'bridge',\n                     'ground-track-field', 'small-vehicle', 'large-vehicle',\n                     'ship', 'tennis-court', 'basketball-court',\n                     'storage-tank', 'soccer-ball-field', 'roundabout',\n                     'harbor', 'swimming-pool', 'helicopter',\n                     'container-crane'),\n            filter_empty_gt=False)),\n    val=dict(\n        type='DOTAv15WSOODDataset',\n        pipeline=[\n            dict(type='LoadImageFromFile'),\n            dict(\n                type='MultiScaleFlipAug',\n                img_scale=(1024, 1024),\n                flip=False,\n                transforms=[\n                    dict(type='RResize'),\n                    dict(\n                        type='Normalize',\n                        mean=[123.675, 116.28, 103.53],\n                        std=[58.395, 57.12, 57.375],\n                        to_rgb=True),\n                    dict(type='Pad', size_divisor=32),\n                    dict(type='DefaultFormatBundle'),\n                    dict(type='Collect', keys=['img'])\n                ])\n        ],\n        img_prefix='data/val/images/',\n        ann_file='data/val/annfiles/',\n        version='le90',\n        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n                 'basketball-court', 'storage-tank', 'soccer-ball-field',\n                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n                 'container-crane')),\n    test=dict(\n        type='DOTAv15WSOODDataset',\n        pipeline=[\n            dict(type='LoadImageFromFile'),\n            dict(\n                type='MultiScaleFlipAug',\n                img_scale=(1024, 1024),\n                flip=False,\n                transforms=[\n                    dict(type='RResize'),\n                    dict(\n                        type='Normalize',\n                        mean=[123.675, 116.28, 103.53],\n                        std=[58.395, 57.12, 57.375],\n                        to_rgb=True),\n                    dict(type='Pad', size_divisor=32),\n                    dict(type='DefaultFormatBundle'),\n                    dict(type='Collect', keys=['img'])\n                ])\n        ],\n        img_prefix='data/vis_val/images/',\n        ann_file='data/vis_val/annfiles/',\n        version='le90',\n        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n                 'basketball-court', 'storage-tank', 'soccer-ball-field',\n                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n                 'container-crane')),\n    sampler=dict(\n        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))\ncustom_hooks = [\n    dict(type='NumClassCheckHook'),\n    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)\n]\nevaluation = dict(\n    type='SubModulesDistEvalHook',\n    interval=3200,\n    metric='mAP',\n    save_best='mAP')\noptimizer = dict(\n    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)\noptimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))\nlr_config = dict(\n    policy='step',\n    warmup='linear',\n    warmup_iters=500,\n    warmup_ratio=0.3333333333333333,\n    step=120000)\nrunner = dict(type='IterBasedRunner', max_iters=120000)\ncheckpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)\nlog_config = dict(\n    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nopencv_num_threads = 0\nmp_start_method = 'fork'\ncustom_imports = dict(imports=['semi_mmrotate'], allow_failed_imports=False)\nwork_dir = 'work_dir/h2rv2_mcl/a_noise/0.1/'\nauto_resume = False\ngpu_ids = range(0, 2)\n", "seed": 42, "exp_name": "gmm.py"}

work_dir_h/PWOOD/a_noise/0.1/20250307_082733.log

@@ -0,0 +1,558 @@
+2025-03-07 08:27:33,797 - mmrotate - INFO - Environment info:
+------------------------------------------------------------
+sys.platform: linux
+Python: 3.8.0 (default, Nov  6 2019, 21:49:08) [GCC 7.3.0]
+CUDA available: True
+GPU 0,1: NVIDIA GeForce RTX 4090
+CUDA_HOME: /usr/local/cuda
+NVCC: Cuda compilation tools, release 11.8, V11.8.89
+GCC: gcc (Ubuntu 9.4.0-1ubuntu1~18.04) 9.4.0
+PyTorch: 1.13.1+cu116
+PyTorch compiling details: PyTorch built with:
+  - GCC 9.3
+  - C++ Version: 201402
+  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications
+  - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)
+  - OpenMP 201511 (a.k.a. OpenMP 4.5)
+  - LAPACK is enabled (usually provided by MKL)
+  - NNPACK is enabled
+  - CPU capability usage: AVX2
+  - CUDA Runtime 11.6
+  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86
+  - CuDNN 8.3.2  (built against CUDA 11.5)
+  - Magma 2.6.1
+  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.6, CUDNN_VERSION=8.3.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, 
+
+TorchVision: 0.14.1+cu116
+OpenCV: 4.11.0
+MMCV: 1.7.2
+MMCV Compiler: GCC 9.3
+MMCV CUDA Compiler: 11.6
+MMRotate: 0.3.4+7833b87
+------------------------------------------------------------
+
+2025-03-07 08:27:35,019 - mmrotate - INFO - Distributed training: True
+2025-03-07 08:27:36,388 - mmrotate - INFO - Config:
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='SemiGMMLoss', cls_channels=16, policy='high'),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='AddNoise', p=0.1),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='AddNoise', p=0.1),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+custom_imports = dict(imports=['semi_mmrotate'], allow_failed_imports=False)
+work_dir = 'work_dir/h2rv2_mcl/a_noise/0.1/'
+auto_resume = False
+gpu_ids = range(0, 2)
+
+2025-03-07 08:27:36,388 - mmrotate - INFO - Set random seed to 42, deterministic: True
+2025-03-07 08:28:00,761 - mmrotate - INFO - Start running, host: yangxue@think4proj-105-232, work_dir: /mnt/nas2/home/yangxue/lmx/sood-mcl/work_dir/h2rv2_mcl/a_noise/0.1
+2025-03-07 08:28:00,761 - mmrotate - INFO - Hooks will be executed in the following order:
+before_run:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(NORMAL      ) CheckpointHook                     
+(NORMAL      ) MeanTeacher                        
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_train_epoch:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(NORMAL      ) NumClassCheckHook                  
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_train_iter:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+ -------------------- 
+after_train_iter:
+(ABOVE_NORMAL) OptimizerHook                      
+(NORMAL      ) CheckpointHook                     
+(NORMAL      ) MeanTeacher                        
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+after_train_epoch:
+(NORMAL      ) CheckpointHook                     
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_val_epoch:
+(NORMAL      ) NumClassCheckHook                  
+(LOW         ) IterTimerHook                      
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_val_iter:
+(LOW         ) IterTimerHook                      
+ -------------------- 
+after_val_iter:
+(LOW         ) IterTimerHook                      
+ -------------------- 
+after_val_epoch:
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+after_run:
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+2025-03-07 08:28:00,761 - mmrotate - INFO - workflow: [('train', 1)], max: 120000 iters
+2025-03-07 08:28:00,762 - mmrotate - INFO - Checkpoints will be saved to /mnt/nas2/home/yangxue/lmx/sood-mcl/work_dir/h2rv2_mcl/a_noise/0.1 by HardDiskBackend.

work_dir_h/PWOOD/a_noise/0.1/20250307_082733.log.json

@@ -0,0 +1 @@
+{"env_info": "sys.platform: linux\nPython: 3.8.0 (default, Nov  6 2019, 21:49:08) [GCC 7.3.0]\nCUDA available: True\nGPU 0,1: NVIDIA GeForce RTX 4090\nCUDA_HOME: /usr/local/cuda\nNVCC: Cuda compilation tools, release 11.8, V11.8.89\nGCC: gcc (Ubuntu 9.4.0-1ubuntu1~18.04) 9.4.0\nPyTorch: 1.13.1+cu116\nPyTorch compiling details: PyTorch built with:\n  - GCC 9.3\n  - C++ Version: 201402\n  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications\n  - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)\n  - OpenMP 201511 (a.k.a. OpenMP 4.5)\n  - LAPACK is enabled (usually provided by MKL)\n  - NNPACK is enabled\n  - CPU capability usage: AVX2\n  - CUDA Runtime 11.6\n  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86\n  - CuDNN 8.3.2  (built against CUDA 11.5)\n  - Magma 2.6.1\n  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.6, CUDNN_VERSION=8.3.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, \n\nTorchVision: 0.14.1+cu116\nOpenCV: 4.11.0\nMMCV: 1.7.2\nMMCV Compiler: GCC 9.3\nMMCV CUDA Compiler: 11.6\nMMRotate: 0.3.4+7833b87", "config": "angle_version = 'le90'\ndetector = dict(\n    type='SemiRotatedFCOS',\n    backbone=dict(\n        type='ResNet',\n        depth=50,\n        num_stages=4,\n        out_indices=(0, 1, 2, 3),\n        frozen_stages=1,\n        zero_init_residual=False,\n        norm_cfg=dict(type='BN', requires_grad=True),\n        norm_eval=True,\n        style='pytorch',\n        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),\n    neck=dict(\n        type='FPN',\n        in_channels=[256, 512, 1024, 2048],\n        out_channels=256,\n        start_level=1,\n        add_extra_convs='on_output',\n        num_outs=5,\n        relu_before_extra_convs=True),\n    bbox_head=dict(\n        type='SemiRotatedFCOSHeadH2RV2MCL',\n        num_classes=16,\n        in_channels=256,\n        stacked_convs=4,\n        feat_channels=256,\n        strides=[8, 16, 32, 64, 128],\n        center_sampling=True,\n        center_sample_radius=1.5,\n        norm_on_bbox=True,\n        centerness_on_reg=True,\n        square_cls=[1, 9, 11],\n        resize_cls=[1],\n        scale_angle=False,\n        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),\n        loss_cls=dict(\n            type='FocalLoss',\n            use_sigmoid=True,\n            gamma=2.0,\n            alpha=0.25,\n            loss_weight=1.0),\n        loss_bbox=dict(type='IoULoss', loss_weight=1.0),\n        loss_centerness=dict(\n            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),\n        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),\n    train_cfg=None,\n    test_cfg=dict(\n        nms_pre=2000,\n        min_bbox_size=0,\n        score_thr=0.05,\n        nms=dict(iou_thr=0.1),\n        max_per_img=2000))\nmodel = dict(\n    type='H2RV2MCLTeacher',\n    model=dict(\n        type='SemiRotatedFCOS',\n        backbone=dict(\n            type='ResNet',\n            depth=50,\n            num_stages=4,\n            out_indices=(0, 1, 2, 3),\n            frozen_stages=1,\n            zero_init_residual=False,\n            norm_cfg=dict(type='BN', requires_grad=True),\n            norm_eval=True,\n            style='pytorch',\n            init_cfg=dict(\n                type='Pretrained', checkpoint='torchvision://resnet50')),\n        neck=dict(\n            type='FPN',\n            in_channels=[256, 512, 1024, 2048],\n            out_channels=256,\n            start_level=1,\n            add_extra_convs='on_output',\n            num_outs=5,\n            relu_before_extra_convs=True),\n        bbox_head=dict(\n            type='SemiRotatedFCOSHeadH2RV2MCL',\n            num_classes=16,\n            in_channels=256,\n            stacked_convs=4,\n            feat_channels=256,\n            strides=[8, 16, 32, 64, 128],\n            center_sampling=True,\n            center_sample_radius=1.5,\n            norm_on_bbox=True,\n            centerness_on_reg=True,\n            square_cls=[1, 9, 11],\n            resize_cls=[1],\n            scale_angle=False,\n            bbox_coder=dict(\n                type='DistanceAnglePointCoder', angle_version='le90'),\n            loss_cls=dict(\n                type='FocalLoss',\n                use_sigmoid=True,\n                gamma=2.0,\n                alpha=0.25,\n                loss_weight=1.0),\n            loss_bbox=dict(type='IoULoss', loss_weight=1.0),\n            loss_centerness=dict(\n                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),\n            loss_ss_symmetry=dict(\n                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),\n        train_cfg=None,\n        test_cfg=dict(\n            nms_pre=2000,\n            min_bbox_size=0,\n            score_thr=0.05,\n            nms=dict(iou_thr=0.1),\n            max_per_img=2000)),\n    semi_loss=dict(type='SemiGMMLoss', cls_channels=16, policy='high'),\n    train_cfg=dict(\n        iter_count=0,\n        burn_in_steps=12800,\n        sup_weight=1.0,\n        unsup_weight=1.0,\n        weight_suppress='exp',\n        logit_specific_weights=dict(),\n        cls_channels=16),\n    test_cfg=dict(inference_on='teacher'))\nimg_norm_cfg = dict(\n    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ncommon_pipeline = [\n    dict(\n        type='Normalize',\n        mean=[123.675, 116.28, 103.53],\n        std=[58.395, 57.12, 57.375],\n        to_rgb=True),\n    dict(type='Pad', size_divisor=32),\n    dict(type='DefaultFormatBundle'),\n    dict(\n        type='Collect',\n        keys=['img', 'gt_bboxes', 'gt_labels'],\n        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',\n                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',\n                   'img_norm_cfg', 'tag'))\n]\nstrong_pipeline = [\n    dict(type='DTToPILImage'),\n    dict(\n        type='DTRandomApply',\n        operations=[\n            ColorJitter(\n                brightness=[0.6, 1.4],\n                contrast=[0.6, 1.4],\n                saturation=[0.6, 1.4],\n                hue=[-0.1, 0.1])\n        ],\n        p=0.8),\n    dict(type='DTRandomGrayscale', p=0.2),\n    dict(\n        type='DTRandomApply',\n        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),\n    dict(type='DTToNumpy'),\n    dict(type='ExtraAttrs', tag='unsup_strong')\n]\nweak_pipeline = [\n    dict(type='RResize', img_scale=(1024, 1024)),\n    dict(\n        type='RRandomFlip',\n        flip_ratio=[0.25, 0.25, 0.25],\n        direction=['horizontal', 'vertical', 'diagonal'],\n        version='le90'),\n    dict(type='ExtraAttrs', tag='unsup_weak')\n]\nunsup_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(type='LoadEmptyAnnotations', with_bbox=True),\n    dict(\n        type='STMultiBranch',\n        unsup_strong=[\n            dict(type='DTToPILImage'),\n            dict(\n                type='DTRandomApply',\n                operations=[\n                    ColorJitter(\n                        brightness=[0.6, 1.4],\n                        contrast=[0.6, 1.4],\n                        saturation=[0.6, 1.4],\n                        hue=[-0.1, 0.1])\n                ],\n                p=0.8),\n            dict(type='DTRandomGrayscale', p=0.2),\n            dict(\n                type='DTRandomApply',\n                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,\n                                                                   2.0])]),\n            dict(type='DTToNumpy'),\n            dict(type='ExtraAttrs', tag='unsup_strong')\n        ],\n        unsup_weak=[\n            dict(type='RResize', img_scale=(1024, 1024)),\n            dict(\n                type='RRandomFlip',\n                flip_ratio=[0.25, 0.25, 0.25],\n                direction=['horizontal', 'vertical', 'diagonal'],\n                version='le90'),\n            dict(type='ExtraAttrs', tag='unsup_weak')\n        ],\n        common_pipeline=[\n            dict(\n                type='Normalize',\n                mean=[123.675, 116.28, 103.53],\n                std=[58.395, 57.12, 57.375],\n                to_rgb=True),\n            dict(type='Pad', size_divisor=32),\n            dict(type='DefaultFormatBundle'),\n            dict(\n                type='Collect',\n                keys=['img', 'gt_bboxes', 'gt_labels'],\n                meta_keys=('filename', 'ori_filename', 'ori_shape',\n                           'img_shape', 'pad_shape', 'scale_factor', 'flip',\n                           'flip_direction', 'img_norm_cfg', 'tag'))\n        ],\n        is_seq=True)\n]\nsup_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(type='LoadAnnotations', with_bbox=True),\n    dict(type='AddNoise', p=0.1),\n    dict(type='RResize', img_scale=(1024, 1024)),\n    dict(\n        type='RRandomFlip',\n        flip_ratio=[0.25, 0.25, 0.25],\n        direction=['horizontal', 'vertical', 'diagonal'],\n        version='le90'),\n    dict(type='ExtraAttrs', tag='sup_weak'),\n    dict(\n        type='Normalize',\n        mean=[123.675, 116.28, 103.53],\n        std=[58.395, 57.12, 57.375],\n        to_rgb=True),\n    dict(type='Pad', size_divisor=32),\n    dict(type='DefaultFormatBundle'),\n    dict(\n        type='Collect',\n        keys=['img', 'gt_bboxes', 'gt_labels'],\n        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',\n                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',\n                   'img_norm_cfg', 'tag'))\n]\ntest_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(\n        type='MultiScaleFlipAug',\n        img_scale=(1024, 1024),\n        flip=False,\n        transforms=[\n            dict(type='RResize'),\n            dict(\n                type='Normalize',\n                mean=[123.675, 116.28, 103.53],\n                std=[58.395, 57.12, 57.375],\n                to_rgb=True),\n            dict(type='Pad', size_divisor=32),\n            dict(type='DefaultFormatBundle'),\n            dict(type='Collect', keys=['img'])\n        ])\n]\ndataset_type = 'DOTAv15WSOODDataset'\nclasses = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n           'basketball-court', 'storage-tank', 'soccer-ball-field',\n           'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n           'container-crane')\ndata = dict(\n    samples_per_gpu=3,\n    workers_per_gpu=5,\n    train=dict(\n        type='SemiDataset',\n        sup=dict(\n            type='DOTAv15WSOODDataset',\n            pipeline=[\n                dict(type='LoadImageFromFile'),\n                dict(type='LoadAnnotations', with_bbox=True),\n                dict(type='AddNoise', p=0.1),\n                dict(type='RResize', img_scale=(1024, 1024)),\n                dict(\n                    type='RRandomFlip',\n                    flip_ratio=[0.25, 0.25, 0.25],\n                    direction=['horizontal', 'vertical', 'diagonal'],\n                    version='le90'),\n                dict(type='ExtraAttrs', tag='sup_weak'),\n                dict(\n                    type='Normalize',\n                    mean=[123.675, 116.28, 103.53],\n                    std=[58.395, 57.12, 57.375],\n                    to_rgb=True),\n                dict(type='Pad', size_divisor=32),\n                dict(type='DefaultFormatBundle'),\n                dict(\n                    type='Collect',\n                    keys=['img', 'gt_bboxes', 'gt_labels'],\n                    meta_keys=('filename', 'ori_filename', 'ori_shape',\n                               'img_shape', 'pad_shape', 'scale_factor',\n                               'flip', 'flip_direction', 'img_norm_cfg',\n                               'tag'))\n            ],\n            ann_file='data/train_20p_labeled/annfiles/',\n            img_prefix='data/train_20p_labeled/images/',\n            version='le90',\n            classes=('plane', 'baseball-diamond', 'bridge',\n                     'ground-track-field', 'small-vehicle', 'large-vehicle',\n                     'ship', 'tennis-court', 'basketball-court',\n                     'storage-tank', 'soccer-ball-field', 'roundabout',\n                     'harbor', 'swimming-pool', 'helicopter',\n                     'container-crane')),\n        unsup=dict(\n            type='DOTAv15WSOODDataset',\n            pipeline=[\n                dict(type='LoadImageFromFile'),\n                dict(type='LoadEmptyAnnotations', with_bbox=True),\n                dict(\n                    type='STMultiBranch',\n                    unsup_strong=[\n                        dict(type='DTToPILImage'),\n                        dict(\n                            type='DTRandomApply',\n                            operations=[\n                                ColorJitter(\n                                    brightness=[0.6, 1.4],\n                                    contrast=[0.6, 1.4],\n                                    saturation=[0.6, 1.4],\n                                    hue=[-0.1, 0.1])\n                            ],\n                            p=0.8),\n                        dict(type='DTRandomGrayscale', p=0.2),\n                        dict(\n                            type='DTRandomApply',\n                            operations=[\n                                dict(\n                                    type='DTGaussianBlur',\n                                    rad_range=[0.1, 2.0])\n                            ]),\n                        dict(type='DTToNumpy'),\n                        dict(type='ExtraAttrs', tag='unsup_strong')\n                    ],\n                    unsup_weak=[\n                        dict(type='RResize', img_scale=(1024, 1024)),\n                        dict(\n                            type='RRandomFlip',\n                            flip_ratio=[0.25, 0.25, 0.25],\n                            direction=['horizontal', 'vertical', 'diagonal'],\n                            version='le90'),\n                        dict(type='ExtraAttrs', tag='unsup_weak')\n                    ],\n                    common_pipeline=[\n                        dict(\n                            type='Normalize',\n                            mean=[123.675, 116.28, 103.53],\n                            std=[58.395, 57.12, 57.375],\n                            to_rgb=True),\n                        dict(type='Pad', size_divisor=32),\n                        dict(type='DefaultFormatBundle'),\n                        dict(\n                            type='Collect',\n                            keys=['img', 'gt_bboxes', 'gt_labels'],\n                            meta_keys=('filename', 'ori_filename', 'ori_shape',\n                                       'img_shape', 'pad_shape',\n                                       'scale_factor', 'flip',\n                                       'flip_direction', 'img_norm_cfg',\n                                       'tag'))\n                    ],\n                    is_seq=True)\n            ],\n            ann_file='data/train_20p_unlabeled/empty_annfiles/',\n            img_prefix='data/train_20p_unlabeled/images/',\n            version='le90',\n            classes=('plane', 'baseball-diamond', 'bridge',\n                     'ground-track-field', 'small-vehicle', 'large-vehicle',\n                     'ship', 'tennis-court', 'basketball-court',\n                     'storage-tank', 'soccer-ball-field', 'roundabout',\n                     'harbor', 'swimming-pool', 'helicopter',\n                     'container-crane'),\n            filter_empty_gt=False)),\n    val=dict(\n        type='DOTAv15WSOODDataset',\n        pipeline=[\n            dict(type='LoadImageFromFile'),\n            dict(\n                type='MultiScaleFlipAug',\n                img_scale=(1024, 1024),\n                flip=False,\n                transforms=[\n                    dict(type='RResize'),\n                    dict(\n                        type='Normalize',\n                        mean=[123.675, 116.28, 103.53],\n                        std=[58.395, 57.12, 57.375],\n                        to_rgb=True),\n                    dict(type='Pad', size_divisor=32),\n                    dict(type='DefaultFormatBundle'),\n                    dict(type='Collect', keys=['img'])\n                ])\n        ],\n        img_prefix='data/val/images/',\n        ann_file='data/val/annfiles/',\n        version='le90',\n        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n                 'basketball-court', 'storage-tank', 'soccer-ball-field',\n                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n                 'container-crane')),\n    test=dict(\n        type='DOTAv15WSOODDataset',\n        pipeline=[\n            dict(type='LoadImageFromFile'),\n            dict(\n                type='MultiScaleFlipAug',\n                img_scale=(1024, 1024),\n                flip=False,\n                transforms=[\n                    dict(type='RResize'),\n                    dict(\n                        type='Normalize',\n                        mean=[123.675, 116.28, 103.53],\n                        std=[58.395, 57.12, 57.375],\n                        to_rgb=True),\n                    dict(type='Pad', size_divisor=32),\n                    dict(type='DefaultFormatBundle'),\n                    dict(type='Collect', keys=['img'])\n                ])\n        ],\n        img_prefix='data/vis_val/images/',\n        ann_file='data/vis_val/annfiles/',\n        version='le90',\n        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n                 'basketball-court', 'storage-tank', 'soccer-ball-field',\n                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n                 'container-crane')),\n    sampler=dict(\n        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))\ncustom_hooks = [\n    dict(type='NumClassCheckHook'),\n    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)\n]\nevaluation = dict(\n    type='SubModulesDistEvalHook',\n    interval=3200,\n    metric='mAP',\n    save_best='mAP')\noptimizer = dict(\n    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)\noptimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))\nlr_config = dict(\n    policy='step',\n    warmup='linear',\n    warmup_iters=500,\n    warmup_ratio=0.3333333333333333,\n    step=120000)\nrunner = dict(type='IterBasedRunner', max_iters=120000)\ncheckpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)\nlog_config = dict(\n    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nopencv_num_threads = 0\nmp_start_method = 'fork'\ncustom_imports = dict(imports=['semi_mmrotate'], allow_failed_imports=False)\nwork_dir = 'work_dir/h2rv2_mcl/a_noise/0.1/'\nauto_resume = False\ngpu_ids = range(0, 2)\n", "seed": 42, "exp_name": "gmm.py"}

work_dir_h/PWOOD/a_noise/0.1/gmm.py

@@ -0,0 +1,468 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='SemiGMMLoss', cls_channels=16, policy='high'),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='AddNoise', p=0.1),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='AddNoise', p=0.1),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+custom_imports = dict(imports=['semi_mmrotate'], allow_failed_imports=False)
+work_dir = 'work_dir/h2rv2_mcl/a_noise/0.1/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/a_noise/0.3/20250307_115752.log

@@ -0,0 +1,558 @@
+2025-03-07 11:57:52,177 - mmrotate - INFO - Environment info:
+------------------------------------------------------------
+sys.platform: linux
+Python: 3.8.0 (default, Nov  6 2019, 21:49:08) [GCC 7.3.0]
+CUDA available: True
+GPU 0,1: NVIDIA GeForce RTX 3090
+CUDA_HOME: /usr
+NVCC: Cuda compilation tools, release 10.1, V10.1.24
+GCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0
+PyTorch: 1.13.1+cu116
+PyTorch compiling details: PyTorch built with:
+  - GCC 9.3
+  - C++ Version: 201402
+  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications
+  - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)
+  - OpenMP 201511 (a.k.a. OpenMP 4.5)
+  - LAPACK is enabled (usually provided by MKL)
+  - NNPACK is enabled
+  - CPU capability usage: AVX2
+  - CUDA Runtime 11.6
+  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86
+  - CuDNN 8.3.2  (built against CUDA 11.5)
+  - Magma 2.6.1
+  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.6, CUDNN_VERSION=8.3.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, 
+
+TorchVision: 0.14.1+cu116
+OpenCV: 4.11.0
+MMCV: 1.7.2
+MMCV Compiler: GCC 9.3
+MMCV CUDA Compiler: 11.6
+MMRotate: 0.3.4+7833b87
+------------------------------------------------------------
+
+2025-03-07 11:57:52,864 - mmrotate - INFO - Distributed training: True
+2025-03-07 11:57:53,655 - mmrotate - INFO - Config:
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='SemiGMMLoss', cls_channels=16, policy='high'),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='AddNoise', p=0.3),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='AddNoise', p=0.3),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+custom_imports = dict(imports=['semi_mmrotate'], allow_failed_imports=False)
+work_dir = 'work_dir/h2rv2_mcl/a_noise/0.3/'
+auto_resume = False
+gpu_ids = range(0, 2)
+
+2025-03-07 11:57:53,655 - mmrotate - INFO - Set random seed to 42, deterministic: True
+2025-03-07 11:58:11,686 - mmrotate - INFO - Start running, host: yangxue@thinklab-105-226, work_dir: /mnt/nas2/home/yangxue/lmx/sood-mcl/work_dir/h2rv2_mcl/a_noise/0.3
+2025-03-07 11:58:11,687 - mmrotate - INFO - Hooks will be executed in the following order:
+before_run:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(NORMAL      ) CheckpointHook                     
+(NORMAL      ) MeanTeacher                        
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_train_epoch:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(NORMAL      ) NumClassCheckHook                  
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_train_iter:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+ -------------------- 
+after_train_iter:
+(ABOVE_NORMAL) OptimizerHook                      
+(NORMAL      ) CheckpointHook                     
+(NORMAL      ) MeanTeacher                        
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+after_train_epoch:
+(NORMAL      ) CheckpointHook                     
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_val_epoch:
+(NORMAL      ) NumClassCheckHook                  
+(LOW         ) IterTimerHook                      
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_val_iter:
+(LOW         ) IterTimerHook                      
+ -------------------- 
+after_val_iter:
+(LOW         ) IterTimerHook                      
+ -------------------- 
+after_val_epoch:
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+after_run:
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+2025-03-07 11:58:11,687 - mmrotate - INFO - workflow: [('train', 1)], max: 120000 iters
+2025-03-07 11:58:11,687 - mmrotate - INFO - Checkpoints will be saved to /mnt/nas2/home/yangxue/lmx/sood-mcl/work_dir/h2rv2_mcl/a_noise/0.3 by HardDiskBackend.

work_dir_h/PWOOD/a_noise/0.3/20250307_115752.log.json

@@ -0,0 +1 @@
+{"env_info": "sys.platform: linux\nPython: 3.8.0 (default, Nov  6 2019, 21:49:08) [GCC 7.3.0]\nCUDA available: True\nGPU 0,1: NVIDIA GeForce RTX 3090\nCUDA_HOME: /usr\nNVCC: Cuda compilation tools, release 10.1, V10.1.24\nGCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0\nPyTorch: 1.13.1+cu116\nPyTorch compiling details: PyTorch built with:\n  - GCC 9.3\n  - C++ Version: 201402\n  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications\n  - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)\n  - OpenMP 201511 (a.k.a. OpenMP 4.5)\n  - LAPACK is enabled (usually provided by MKL)\n  - NNPACK is enabled\n  - CPU capability usage: AVX2\n  - CUDA Runtime 11.6\n  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86\n  - CuDNN 8.3.2  (built against CUDA 11.5)\n  - Magma 2.6.1\n  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.6, CUDNN_VERSION=8.3.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, \n\nTorchVision: 0.14.1+cu116\nOpenCV: 4.11.0\nMMCV: 1.7.2\nMMCV Compiler: GCC 9.3\nMMCV CUDA Compiler: 11.6\nMMRotate: 0.3.4+7833b87", "config": "angle_version = 'le90'\ndetector = dict(\n    type='SemiRotatedFCOS',\n    backbone=dict(\n        type='ResNet',\n        depth=50,\n        num_stages=4,\n        out_indices=(0, 1, 2, 3),\n        frozen_stages=1,\n        zero_init_residual=False,\n        norm_cfg=dict(type='BN', requires_grad=True),\n        norm_eval=True,\n        style='pytorch',\n        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),\n    neck=dict(\n        type='FPN',\n        in_channels=[256, 512, 1024, 2048],\n        out_channels=256,\n        start_level=1,\n        add_extra_convs='on_output',\n        num_outs=5,\n        relu_before_extra_convs=True),\n    bbox_head=dict(\n        type='SemiRotatedFCOSHeadH2RV2MCL',\n        num_classes=16,\n        in_channels=256,\n        stacked_convs=4,\n        feat_channels=256,\n        strides=[8, 16, 32, 64, 128],\n        center_sampling=True,\n        center_sample_radius=1.5,\n        norm_on_bbox=True,\n        centerness_on_reg=True,\n        square_cls=[1, 9, 11],\n        resize_cls=[1],\n        scale_angle=False,\n        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),\n        loss_cls=dict(\n            type='FocalLoss',\n            use_sigmoid=True,\n            gamma=2.0,\n            alpha=0.25,\n            loss_weight=1.0),\n        loss_bbox=dict(type='IoULoss', loss_weight=1.0),\n        loss_centerness=dict(\n            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),\n        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),\n    train_cfg=None,\n    test_cfg=dict(\n        nms_pre=2000,\n        min_bbox_size=0,\n        score_thr=0.05,\n        nms=dict(iou_thr=0.1),\n        max_per_img=2000))\nmodel = dict(\n    type='H2RV2MCLTeacher',\n    model=dict(\n        type='SemiRotatedFCOS',\n        backbone=dict(\n            type='ResNet',\n            depth=50,\n            num_stages=4,\n            out_indices=(0, 1, 2, 3),\n            frozen_stages=1,\n            zero_init_residual=False,\n            norm_cfg=dict(type='BN', requires_grad=True),\n            norm_eval=True,\n            style='pytorch',\n            init_cfg=dict(\n                type='Pretrained', checkpoint='torchvision://resnet50')),\n        neck=dict(\n            type='FPN',\n            in_channels=[256, 512, 1024, 2048],\n            out_channels=256,\n            start_level=1,\n            add_extra_convs='on_output',\n            num_outs=5,\n            relu_before_extra_convs=True),\n        bbox_head=dict(\n            type='SemiRotatedFCOSHeadH2RV2MCL',\n            num_classes=16,\n            in_channels=256,\n            stacked_convs=4,\n            feat_channels=256,\n            strides=[8, 16, 32, 64, 128],\n            center_sampling=True,\n            center_sample_radius=1.5,\n            norm_on_bbox=True,\n            centerness_on_reg=True,\n            square_cls=[1, 9, 11],\n            resize_cls=[1],\n            scale_angle=False,\n            bbox_coder=dict(\n                type='DistanceAnglePointCoder', angle_version='le90'),\n            loss_cls=dict(\n                type='FocalLoss',\n                use_sigmoid=True,\n                gamma=2.0,\n                alpha=0.25,\n                loss_weight=1.0),\n            loss_bbox=dict(type='IoULoss', loss_weight=1.0),\n            loss_centerness=dict(\n                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),\n            loss_ss_symmetry=dict(\n                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),\n        train_cfg=None,\n        test_cfg=dict(\n            nms_pre=2000,\n            min_bbox_size=0,\n            score_thr=0.05,\n            nms=dict(iou_thr=0.1),\n            max_per_img=2000)),\n    semi_loss=dict(type='SemiGMMLoss', cls_channels=16, policy='high'),\n    train_cfg=dict(\n        iter_count=0,\n        burn_in_steps=12800,\n        sup_weight=1.0,\n        unsup_weight=1.0,\n        weight_suppress='exp',\n        logit_specific_weights=dict(),\n        cls_channels=16),\n    test_cfg=dict(inference_on='teacher'))\nimg_norm_cfg = dict(\n    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ncommon_pipeline = [\n    dict(\n        type='Normalize',\n        mean=[123.675, 116.28, 103.53],\n        std=[58.395, 57.12, 57.375],\n        to_rgb=True),\n    dict(type='Pad', size_divisor=32),\n    dict(type='DefaultFormatBundle'),\n    dict(\n        type='Collect',\n        keys=['img', 'gt_bboxes', 'gt_labels'],\n        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',\n                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',\n                   'img_norm_cfg', 'tag'))\n]\nstrong_pipeline = [\n    dict(type='DTToPILImage'),\n    dict(\n        type='DTRandomApply',\n        operations=[\n            ColorJitter(\n                brightness=[0.6, 1.4],\n                contrast=[0.6, 1.4],\n                saturation=[0.6, 1.4],\n                hue=[-0.1, 0.1])\n        ],\n        p=0.8),\n    dict(type='DTRandomGrayscale', p=0.2),\n    dict(\n        type='DTRandomApply',\n        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),\n    dict(type='DTToNumpy'),\n    dict(type='ExtraAttrs', tag='unsup_strong')\n]\nweak_pipeline = [\n    dict(type='RResize', img_scale=(1024, 1024)),\n    dict(\n        type='RRandomFlip',\n        flip_ratio=[0.25, 0.25, 0.25],\n        direction=['horizontal', 'vertical', 'diagonal'],\n        version='le90'),\n    dict(type='ExtraAttrs', tag='unsup_weak')\n]\nunsup_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(type='LoadEmptyAnnotations', with_bbox=True),\n    dict(\n        type='STMultiBranch',\n        unsup_strong=[\n            dict(type='DTToPILImage'),\n            dict(\n                type='DTRandomApply',\n                operations=[\n                    ColorJitter(\n                        brightness=[0.6, 1.4],\n                        contrast=[0.6, 1.4],\n                        saturation=[0.6, 1.4],\n                        hue=[-0.1, 0.1])\n                ],\n                p=0.8),\n            dict(type='DTRandomGrayscale', p=0.2),\n            dict(\n                type='DTRandomApply',\n                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,\n                                                                   2.0])]),\n            dict(type='DTToNumpy'),\n            dict(type='ExtraAttrs', tag='unsup_strong')\n        ],\n        unsup_weak=[\n            dict(type='RResize', img_scale=(1024, 1024)),\n            dict(\n                type='RRandomFlip',\n                flip_ratio=[0.25, 0.25, 0.25],\n                direction=['horizontal', 'vertical', 'diagonal'],\n                version='le90'),\n            dict(type='ExtraAttrs', tag='unsup_weak')\n        ],\n        common_pipeline=[\n            dict(\n                type='Normalize',\n                mean=[123.675, 116.28, 103.53],\n                std=[58.395, 57.12, 57.375],\n                to_rgb=True),\n            dict(type='Pad', size_divisor=32),\n            dict(type='DefaultFormatBundle'),\n            dict(\n                type='Collect',\n                keys=['img', 'gt_bboxes', 'gt_labels'],\n                meta_keys=('filename', 'ori_filename', 'ori_shape',\n                           'img_shape', 'pad_shape', 'scale_factor', 'flip',\n                           'flip_direction', 'img_norm_cfg', 'tag'))\n        ],\n        is_seq=True)\n]\nsup_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(type='LoadAnnotations', with_bbox=True),\n    dict(type='AddNoise', p=0.3),\n    dict(type='RResize', img_scale=(1024, 1024)),\n    dict(\n        type='RRandomFlip',\n        flip_ratio=[0.25, 0.25, 0.25],\n        direction=['horizontal', 'vertical', 'diagonal'],\n        version='le90'),\n    dict(type='ExtraAttrs', tag='sup_weak'),\n    dict(\n        type='Normalize',\n        mean=[123.675, 116.28, 103.53],\n        std=[58.395, 57.12, 57.375],\n        to_rgb=True),\n    dict(type='Pad', size_divisor=32),\n    dict(type='DefaultFormatBundle'),\n    dict(\n        type='Collect',\n        keys=['img', 'gt_bboxes', 'gt_labels'],\n        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',\n                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',\n                   'img_norm_cfg', 'tag'))\n]\ntest_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(\n        type='MultiScaleFlipAug',\n        img_scale=(1024, 1024),\n        flip=False,\n        transforms=[\n            dict(type='RResize'),\n            dict(\n                type='Normalize',\n                mean=[123.675, 116.28, 103.53],\n                std=[58.395, 57.12, 57.375],\n                to_rgb=True),\n            dict(type='Pad', size_divisor=32),\n            dict(type='DefaultFormatBundle'),\n            dict(type='Collect', keys=['img'])\n        ])\n]\ndataset_type = 'DOTAv15WSOODDataset'\nclasses = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n           'basketball-court', 'storage-tank', 'soccer-ball-field',\n           'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n           'container-crane')\ndata = dict(\n    samples_per_gpu=3,\n    workers_per_gpu=5,\n    train=dict(\n        type='SemiDataset',\n        sup=dict(\n            type='DOTAv15WSOODDataset',\n            pipeline=[\n                dict(type='LoadImageFromFile'),\n                dict(type='LoadAnnotations', with_bbox=True),\n                dict(type='AddNoise', p=0.3),\n                dict(type='RResize', img_scale=(1024, 1024)),\n                dict(\n                    type='RRandomFlip',\n                    flip_ratio=[0.25, 0.25, 0.25],\n                    direction=['horizontal', 'vertical', 'diagonal'],\n                    version='le90'),\n                dict(type='ExtraAttrs', tag='sup_weak'),\n                dict(\n                    type='Normalize',\n                    mean=[123.675, 116.28, 103.53],\n                    std=[58.395, 57.12, 57.375],\n                    to_rgb=True),\n                dict(type='Pad', size_divisor=32),\n                dict(type='DefaultFormatBundle'),\n                dict(\n                    type='Collect',\n                    keys=['img', 'gt_bboxes', 'gt_labels'],\n                    meta_keys=('filename', 'ori_filename', 'ori_shape',\n                               'img_shape', 'pad_shape', 'scale_factor',\n                               'flip', 'flip_direction', 'img_norm_cfg',\n                               'tag'))\n            ],\n            ann_file='data/train_20p_labeled/annfiles/',\n            img_prefix='data/train_20p_labeled/images/',\n            version='le90',\n            classes=('plane', 'baseball-diamond', 'bridge',\n                     'ground-track-field', 'small-vehicle', 'large-vehicle',\n                     'ship', 'tennis-court', 'basketball-court',\n                     'storage-tank', 'soccer-ball-field', 'roundabout',\n                     'harbor', 'swimming-pool', 'helicopter',\n                     'container-crane')),\n        unsup=dict(\n            type='DOTAv15WSOODDataset',\n            pipeline=[\n                dict(type='LoadImageFromFile'),\n                dict(type='LoadEmptyAnnotations', with_bbox=True),\n                dict(\n                    type='STMultiBranch',\n                    unsup_strong=[\n                        dict(type='DTToPILImage'),\n                        dict(\n                            type='DTRandomApply',\n                            operations=[\n                                ColorJitter(\n                                    brightness=[0.6, 1.4],\n                                    contrast=[0.6, 1.4],\n                                    saturation=[0.6, 1.4],\n                                    hue=[-0.1, 0.1])\n                            ],\n                            p=0.8),\n                        dict(type='DTRandomGrayscale', p=0.2),\n                        dict(\n                            type='DTRandomApply',\n                            operations=[\n                                dict(\n                                    type='DTGaussianBlur',\n                                    rad_range=[0.1, 2.0])\n                            ]),\n                        dict(type='DTToNumpy'),\n                        dict(type='ExtraAttrs', tag='unsup_strong')\n                    ],\n                    unsup_weak=[\n                        dict(type='RResize', img_scale=(1024, 1024)),\n                        dict(\n                            type='RRandomFlip',\n                            flip_ratio=[0.25, 0.25, 0.25],\n                            direction=['horizontal', 'vertical', 'diagonal'],\n                            version='le90'),\n                        dict(type='ExtraAttrs', tag='unsup_weak')\n                    ],\n                    common_pipeline=[\n                        dict(\n                            type='Normalize',\n                            mean=[123.675, 116.28, 103.53],\n                            std=[58.395, 57.12, 57.375],\n                            to_rgb=True),\n                        dict(type='Pad', size_divisor=32),\n                        dict(type='DefaultFormatBundle'),\n                        dict(\n                            type='Collect',\n                            keys=['img', 'gt_bboxes', 'gt_labels'],\n                            meta_keys=('filename', 'ori_filename', 'ori_shape',\n                                       'img_shape', 'pad_shape',\n                                       'scale_factor', 'flip',\n                                       'flip_direction', 'img_norm_cfg',\n                                       'tag'))\n                    ],\n                    is_seq=True)\n            ],\n            ann_file='data/train_20p_unlabeled/empty_annfiles/',\n            img_prefix='data/train_20p_unlabeled/images/',\n            version='le90',\n            classes=('plane', 'baseball-diamond', 'bridge',\n                     'ground-track-field', 'small-vehicle', 'large-vehicle',\n                     'ship', 'tennis-court', 'basketball-court',\n                     'storage-tank', 'soccer-ball-field', 'roundabout',\n                     'harbor', 'swimming-pool', 'helicopter',\n                     'container-crane'),\n            filter_empty_gt=False)),\n    val=dict(\n        type='DOTAv15WSOODDataset',\n        pipeline=[\n            dict(type='LoadImageFromFile'),\n            dict(\n                type='MultiScaleFlipAug',\n                img_scale=(1024, 1024),\n                flip=False,\n                transforms=[\n                    dict(type='RResize'),\n                    dict(\n                        type='Normalize',\n                        mean=[123.675, 116.28, 103.53],\n                        std=[58.395, 57.12, 57.375],\n                        to_rgb=True),\n                    dict(type='Pad', size_divisor=32),\n                    dict(type='DefaultFormatBundle'),\n                    dict(type='Collect', keys=['img'])\n                ])\n        ],\n        img_prefix='data/val/images/',\n        ann_file='data/val/annfiles/',\n        version='le90',\n        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n                 'basketball-court', 'storage-tank', 'soccer-ball-field',\n                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n                 'container-crane')),\n    test=dict(\n        type='DOTAv15WSOODDataset',\n        pipeline=[\n            dict(type='LoadImageFromFile'),\n            dict(\n                type='MultiScaleFlipAug',\n                img_scale=(1024, 1024),\n                flip=False,\n                transforms=[\n                    dict(type='RResize'),\n                    dict(\n                        type='Normalize',\n                        mean=[123.675, 116.28, 103.53],\n                        std=[58.395, 57.12, 57.375],\n                        to_rgb=True),\n                    dict(type='Pad', size_divisor=32),\n                    dict(type='DefaultFormatBundle'),\n                    dict(type='Collect', keys=['img'])\n                ])\n        ],\n        img_prefix='data/vis_val/images/',\n        ann_file='data/vis_val/annfiles/',\n        version='le90',\n        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n                 'basketball-court', 'storage-tank', 'soccer-ball-field',\n                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n                 'container-crane')),\n    sampler=dict(\n        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))\ncustom_hooks = [\n    dict(type='NumClassCheckHook'),\n    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)\n]\nevaluation = dict(\n    type='SubModulesDistEvalHook',\n    interval=3200,\n    metric='mAP',\n    save_best='mAP')\noptimizer = dict(\n    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)\noptimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))\nlr_config = dict(\n    policy='step',\n    warmup='linear',\n    warmup_iters=500,\n    warmup_ratio=0.3333333333333333,\n    step=120000)\nrunner = dict(type='IterBasedRunner', max_iters=120000)\ncheckpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)\nlog_config = dict(\n    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nopencv_num_threads = 0\nmp_start_method = 'fork'\ncustom_imports = dict(imports=['semi_mmrotate'], allow_failed_imports=False)\nwork_dir = 'work_dir/h2rv2_mcl/a_noise/0.3/'\nauto_resume = False\ngpu_ids = range(0, 2)\n", "seed": 42, "exp_name": "a_gmm_noise_3.py"}

work_dir_h/PWOOD/a_noise/0.3/a_gmm_noise_3.py

@@ -0,0 +1,468 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='SemiGMMLoss', cls_channels=16, policy='high'),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='AddNoise', p=0.3),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='AddNoise', p=0.3),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+custom_imports = dict(imports=['semi_mmrotate'], allow_failed_imports=False)
+work_dir = 'work_dir/h2rv2_mcl/a_noise/0.3/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/ablation/0.015/lr_0.03.py

@@ -0,0 +1,465 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLossV2', cls_channels=16, thres=0.015),
+    train_cfg=dict(
+        iter_count=35200,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = 'work_dir/h2rv2_mcl/ablation/0.015/iter_35200.pth'
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/ablation/0.015/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/ablation/0.01_topk/lr_h2rv2_mcl_adamw_dota15_20p.py

@@ -0,0 +1,465 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLossV2', cls_channels=16, thres=0.01),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/ablation/0.01_topk/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/ablation/0.025/20250302_073702.log

@@ -0,0 +1,557 @@
+2025-03-02 07:37:02,579 - mmrotate - INFO - Environment info:
+------------------------------------------------------------
+sys.platform: linux
+Python: 3.8.0 (default, Nov  6 2019, 21:49:08) [GCC 7.3.0]
+CUDA available: True
+GPU 0,1: NVIDIA GeForce RTX 3090
+CUDA_HOME: /usr/local/cuda
+NVCC: Cuda compilation tools, release 11.6, V11.6.55
+GCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0
+PyTorch: 1.13.1+cu116
+PyTorch compiling details: PyTorch built with:
+  - GCC 9.3
+  - C++ Version: 201402
+  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications
+  - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)
+  - OpenMP 201511 (a.k.a. OpenMP 4.5)
+  - LAPACK is enabled (usually provided by MKL)
+  - NNPACK is enabled
+  - CPU capability usage: AVX2
+  - CUDA Runtime 11.6
+  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86
+  - CuDNN 8.3.2  (built against CUDA 11.5)
+  - Magma 2.6.1
+  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.6, CUDNN_VERSION=8.3.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, 
+
+TorchVision: 0.14.1+cu116
+OpenCV: 4.11.0
+MMCV: 1.7.2
+MMCV Compiler: GCC 9.3
+MMCV CUDA Compiler: 11.6
+MMRotate: 0.3.4+7833b87
+------------------------------------------------------------
+
+2025-03-02 07:37:03,726 - mmrotate - INFO - Distributed training: True
+2025-03-02 07:37:05,068 - mmrotate - INFO - Config:
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLossV2', cls_channels=16, thres=0.025),
+    train_cfg=dict(
+        iter_count=28800,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = 'work_dir/h2rv2_mcl/ablation/0.025/iter_28800.pth'
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/ablation/0.025/'
+auto_resume = False
+gpu_ids = range(0, 2)
+
+2025-03-02 07:37:05,069 - mmrotate - INFO - Set random seed to 42, deterministic: True
+2025-03-02 07:38:51,196 - mmrotate - INFO - load checkpoint from local path: work_dir/h2rv2_mcl/ablation/0.025/iter_28800.pth
+2025-03-02 07:38:57,284 - mmrotate - INFO - resumed from epoch: 1, iter 28799
+2025-03-02 07:38:57,287 - mmrotate - INFO - Start running, host: yangxue@thinklab-105-227, work_dir: /mnt/nas2/home/yangxue/lmx/sood-mcl/work_dir/h2rv2_mcl/ablation/0.025
+2025-03-02 07:38:57,288 - mmrotate - INFO - Hooks will be executed in the following order:
+before_run:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(NORMAL      ) CheckpointHook                     
+(NORMAL      ) MeanTeacher                        
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_train_epoch:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(NORMAL      ) NumClassCheckHook                  
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_train_iter:
+(VERY_HIGH   ) StepLrUpdaterHook                  
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+ -------------------- 
+after_train_iter:
+(ABOVE_NORMAL) OptimizerHook                      
+(NORMAL      ) CheckpointHook                     
+(NORMAL      ) MeanTeacher                        
+(LOW         ) IterTimerHook                      
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+after_train_epoch:
+(NORMAL      ) CheckpointHook                     
+(LOW         ) SubModulesDistEvalHook             
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_val_epoch:
+(NORMAL      ) NumClassCheckHook                  
+(LOW         ) IterTimerHook                      
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+before_val_iter:
+(LOW         ) IterTimerHook                      
+ -------------------- 
+after_val_iter:
+(LOW         ) IterTimerHook                      
+ -------------------- 
+after_val_epoch:
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+after_run:
+(VERY_LOW    ) TextLoggerHook                     
+ -------------------- 
+2025-03-02 07:38:57,288 - mmrotate - INFO - workflow: [('train', 1)], max: 120000 iters
+2025-03-02 07:38:57,288 - mmrotate - INFO - Checkpoints will be saved to /mnt/nas2/home/yangxue/lmx/sood-mcl/work_dir/h2rv2_mcl/ablation/0.025 by HardDiskBackend.

work_dir_h/PWOOD/ablation/0.025/20250302_073702.log.json

@@ -0,0 +1 @@
+{"env_info": "sys.platform: linux\nPython: 3.8.0 (default, Nov  6 2019, 21:49:08) [GCC 7.3.0]\nCUDA available: True\nGPU 0,1: NVIDIA GeForce RTX 3090\nCUDA_HOME: /usr/local/cuda\nNVCC: Cuda compilation tools, release 11.6, V11.6.55\nGCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0\nPyTorch: 1.13.1+cu116\nPyTorch compiling details: PyTorch built with:\n  - GCC 9.3\n  - C++ Version: 201402\n  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications\n  - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)\n  - OpenMP 201511 (a.k.a. OpenMP 4.5)\n  - LAPACK is enabled (usually provided by MKL)\n  - NNPACK is enabled\n  - CPU capability usage: AVX2\n  - CUDA Runtime 11.6\n  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86\n  - CuDNN 8.3.2  (built against CUDA 11.5)\n  - Magma 2.6.1\n  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.6, CUDNN_VERSION=8.3.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, \n\nTorchVision: 0.14.1+cu116\nOpenCV: 4.11.0\nMMCV: 1.7.2\nMMCV Compiler: GCC 9.3\nMMCV CUDA Compiler: 11.6\nMMRotate: 0.3.4+7833b87", "config": "angle_version = 'le90'\ndetector = dict(\n    type='SemiRotatedFCOS',\n    backbone=dict(\n        type='ResNet',\n        depth=50,\n        num_stages=4,\n        out_indices=(0, 1, 2, 3),\n        frozen_stages=1,\n        zero_init_residual=False,\n        norm_cfg=dict(type='BN', requires_grad=True),\n        norm_eval=True,\n        style='pytorch',\n        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),\n    neck=dict(\n        type='FPN',\n        in_channels=[256, 512, 1024, 2048],\n        out_channels=256,\n        start_level=1,\n        add_extra_convs='on_output',\n        num_outs=5,\n        relu_before_extra_convs=True),\n    bbox_head=dict(\n        type='SemiRotatedFCOSHeadH2RV2MCL',\n        num_classes=16,\n        in_channels=256,\n        stacked_convs=4,\n        feat_channels=256,\n        strides=[8, 16, 32, 64, 128],\n        center_sampling=True,\n        center_sample_radius=1.5,\n        norm_on_bbox=True,\n        centerness_on_reg=True,\n        square_cls=[1, 9, 11],\n        resize_cls=[1],\n        scale_angle=False,\n        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),\n        loss_cls=dict(\n            type='FocalLoss',\n            use_sigmoid=True,\n            gamma=2.0,\n            alpha=0.25,\n            loss_weight=1.0),\n        loss_bbox=dict(type='IoULoss', loss_weight=1.0),\n        loss_centerness=dict(\n            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),\n        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),\n    train_cfg=None,\n    test_cfg=dict(\n        nms_pre=2000,\n        min_bbox_size=0,\n        score_thr=0.05,\n        nms=dict(iou_thr=0.1),\n        max_per_img=2000))\nmodel = dict(\n    type='H2RV2MCLTeacher',\n    model=dict(\n        type='SemiRotatedFCOS',\n        backbone=dict(\n            type='ResNet',\n            depth=50,\n            num_stages=4,\n            out_indices=(0, 1, 2, 3),\n            frozen_stages=1,\n            zero_init_residual=False,\n            norm_cfg=dict(type='BN', requires_grad=True),\n            norm_eval=True,\n            style='pytorch',\n            init_cfg=dict(\n                type='Pretrained', checkpoint='torchvision://resnet50')),\n        neck=dict(\n            type='FPN',\n            in_channels=[256, 512, 1024, 2048],\n            out_channels=256,\n            start_level=1,\n            add_extra_convs='on_output',\n            num_outs=5,\n            relu_before_extra_convs=True),\n        bbox_head=dict(\n            type='SemiRotatedFCOSHeadH2RV2MCL',\n            num_classes=16,\n            in_channels=256,\n            stacked_convs=4,\n            feat_channels=256,\n            strides=[8, 16, 32, 64, 128],\n            center_sampling=True,\n            center_sample_radius=1.5,\n            norm_on_bbox=True,\n            centerness_on_reg=True,\n            square_cls=[1, 9, 11],\n            resize_cls=[1],\n            scale_angle=False,\n            bbox_coder=dict(\n                type='DistanceAnglePointCoder', angle_version='le90'),\n            loss_cls=dict(\n                type='FocalLoss',\n                use_sigmoid=True,\n                gamma=2.0,\n                alpha=0.25,\n                loss_weight=1.0),\n            loss_bbox=dict(type='IoULoss', loss_weight=1.0),\n            loss_centerness=dict(\n                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),\n            loss_ss_symmetry=dict(\n                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),\n        train_cfg=None,\n        test_cfg=dict(\n            nms_pre=2000,\n            min_bbox_size=0,\n            score_thr=0.05,\n            nms=dict(iou_thr=0.1),\n            max_per_img=2000)),\n    semi_loss=dict(type='RotatedMCLLossV2', cls_channels=16, thres=0.025),\n    train_cfg=dict(\n        iter_count=28800,\n        burn_in_steps=12800,\n        sup_weight=1.0,\n        unsup_weight=1.0,\n        weight_suppress='exp',\n        logit_specific_weights=dict(),\n        cls_channels=16),\n    test_cfg=dict(inference_on='teacher'))\nimg_norm_cfg = dict(\n    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ncommon_pipeline = [\n    dict(\n        type='Normalize',\n        mean=[123.675, 116.28, 103.53],\n        std=[58.395, 57.12, 57.375],\n        to_rgb=True),\n    dict(type='Pad', size_divisor=32),\n    dict(type='DefaultFormatBundle'),\n    dict(\n        type='Collect',\n        keys=['img', 'gt_bboxes', 'gt_labels'],\n        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',\n                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',\n                   'img_norm_cfg', 'tag'))\n]\nstrong_pipeline = [\n    dict(type='DTToPILImage'),\n    dict(\n        type='DTRandomApply',\n        operations=[\n            ColorJitter(\n                brightness=[0.6, 1.4],\n                contrast=[0.6, 1.4],\n                saturation=[0.6, 1.4],\n                hue=[-0.1, 0.1])\n        ],\n        p=0.8),\n    dict(type='DTRandomGrayscale', p=0.2),\n    dict(\n        type='DTRandomApply',\n        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),\n    dict(type='DTToNumpy'),\n    dict(type='ExtraAttrs', tag='unsup_strong')\n]\nweak_pipeline = [\n    dict(type='RResize', img_scale=(1024, 1024)),\n    dict(\n        type='RRandomFlip',\n        flip_ratio=[0.25, 0.25, 0.25],\n        direction=['horizontal', 'vertical', 'diagonal'],\n        version='le90'),\n    dict(type='ExtraAttrs', tag='unsup_weak')\n]\nunsup_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(type='LoadEmptyAnnotations', with_bbox=True),\n    dict(\n        type='STMultiBranch',\n        unsup_strong=[\n            dict(type='DTToPILImage'),\n            dict(\n                type='DTRandomApply',\n                operations=[\n                    ColorJitter(\n                        brightness=[0.6, 1.4],\n                        contrast=[0.6, 1.4],\n                        saturation=[0.6, 1.4],\n                        hue=[-0.1, 0.1])\n                ],\n                p=0.8),\n            dict(type='DTRandomGrayscale', p=0.2),\n            dict(\n                type='DTRandomApply',\n                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,\n                                                                   2.0])]),\n            dict(type='DTToNumpy'),\n            dict(type='ExtraAttrs', tag='unsup_strong')\n        ],\n        unsup_weak=[\n            dict(type='RResize', img_scale=(1024, 1024)),\n            dict(\n                type='RRandomFlip',\n                flip_ratio=[0.25, 0.25, 0.25],\n                direction=['horizontal', 'vertical', 'diagonal'],\n                version='le90'),\n            dict(type='ExtraAttrs', tag='unsup_weak')\n        ],\n        common_pipeline=[\n            dict(\n                type='Normalize',\n                mean=[123.675, 116.28, 103.53],\n                std=[58.395, 57.12, 57.375],\n                to_rgb=True),\n            dict(type='Pad', size_divisor=32),\n            dict(type='DefaultFormatBundle'),\n            dict(\n                type='Collect',\n                keys=['img', 'gt_bboxes', 'gt_labels'],\n                meta_keys=('filename', 'ori_filename', 'ori_shape',\n                           'img_shape', 'pad_shape', 'scale_factor', 'flip',\n                           'flip_direction', 'img_norm_cfg', 'tag'))\n        ],\n        is_seq=True)\n]\nsup_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(type='LoadAnnotations', with_bbox=True),\n    dict(type='RResize', img_scale=(1024, 1024)),\n    dict(\n        type='RRandomFlip',\n        flip_ratio=[0.25, 0.25, 0.25],\n        direction=['horizontal', 'vertical', 'diagonal'],\n        version='le90'),\n    dict(type='ExtraAttrs', tag='sup_weak'),\n    dict(\n        type='Normalize',\n        mean=[123.675, 116.28, 103.53],\n        std=[58.395, 57.12, 57.375],\n        to_rgb=True),\n    dict(type='Pad', size_divisor=32),\n    dict(type='DefaultFormatBundle'),\n    dict(\n        type='Collect',\n        keys=['img', 'gt_bboxes', 'gt_labels'],\n        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',\n                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',\n                   'img_norm_cfg', 'tag'))\n]\ntest_pipeline = [\n    dict(type='LoadImageFromFile'),\n    dict(\n        type='MultiScaleFlipAug',\n        img_scale=(1024, 1024),\n        flip=False,\n        transforms=[\n            dict(type='RResize'),\n            dict(\n                type='Normalize',\n                mean=[123.675, 116.28, 103.53],\n                std=[58.395, 57.12, 57.375],\n                to_rgb=True),\n            dict(type='Pad', size_divisor=32),\n            dict(type='DefaultFormatBundle'),\n            dict(type='Collect', keys=['img'])\n        ])\n]\ndataset_type = 'DOTAv15WSOODDataset'\nclasses = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n           'basketball-court', 'storage-tank', 'soccer-ball-field',\n           'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n           'container-crane')\ndata = dict(\n    samples_per_gpu=3,\n    workers_per_gpu=5,\n    train=dict(\n        type='SemiDataset',\n        sup=dict(\n            type='DOTAv15WSOODDataset',\n            pipeline=[\n                dict(type='LoadImageFromFile'),\n                dict(type='LoadAnnotations', with_bbox=True),\n                dict(type='RResize', img_scale=(1024, 1024)),\n                dict(\n                    type='RRandomFlip',\n                    flip_ratio=[0.25, 0.25, 0.25],\n                    direction=['horizontal', 'vertical', 'diagonal'],\n                    version='le90'),\n                dict(type='ExtraAttrs', tag='sup_weak'),\n                dict(\n                    type='Normalize',\n                    mean=[123.675, 116.28, 103.53],\n                    std=[58.395, 57.12, 57.375],\n                    to_rgb=True),\n                dict(type='Pad', size_divisor=32),\n                dict(type='DefaultFormatBundle'),\n                dict(\n                    type='Collect',\n                    keys=['img', 'gt_bboxes', 'gt_labels'],\n                    meta_keys=('filename', 'ori_filename', 'ori_shape',\n                               'img_shape', 'pad_shape', 'scale_factor',\n                               'flip', 'flip_direction', 'img_norm_cfg',\n                               'tag'))\n            ],\n            ann_file='data/train_20p_labeled/annfiles/',\n            img_prefix='data/train_20p_labeled/images/',\n            version='le90',\n            classes=('plane', 'baseball-diamond', 'bridge',\n                     'ground-track-field', 'small-vehicle', 'large-vehicle',\n                     'ship', 'tennis-court', 'basketball-court',\n                     'storage-tank', 'soccer-ball-field', 'roundabout',\n                     'harbor', 'swimming-pool', 'helicopter',\n                     'container-crane')),\n        unsup=dict(\n            type='DOTAv15WSOODDataset',\n            pipeline=[\n                dict(type='LoadImageFromFile'),\n                dict(type='LoadEmptyAnnotations', with_bbox=True),\n                dict(\n                    type='STMultiBranch',\n                    unsup_strong=[\n                        dict(type='DTToPILImage'),\n                        dict(\n                            type='DTRandomApply',\n                            operations=[\n                                ColorJitter(\n                                    brightness=[0.6, 1.4],\n                                    contrast=[0.6, 1.4],\n                                    saturation=[0.6, 1.4],\n                                    hue=[-0.1, 0.1])\n                            ],\n                            p=0.8),\n                        dict(type='DTRandomGrayscale', p=0.2),\n                        dict(\n                            type='DTRandomApply',\n                            operations=[\n                                dict(\n                                    type='DTGaussianBlur',\n                                    rad_range=[0.1, 2.0])\n                            ]),\n                        dict(type='DTToNumpy'),\n                        dict(type='ExtraAttrs', tag='unsup_strong')\n                    ],\n                    unsup_weak=[\n                        dict(type='RResize', img_scale=(1024, 1024)),\n                        dict(\n                            type='RRandomFlip',\n                            flip_ratio=[0.25, 0.25, 0.25],\n                            direction=['horizontal', 'vertical', 'diagonal'],\n                            version='le90'),\n                        dict(type='ExtraAttrs', tag='unsup_weak')\n                    ],\n                    common_pipeline=[\n                        dict(\n                            type='Normalize',\n                            mean=[123.675, 116.28, 103.53],\n                            std=[58.395, 57.12, 57.375],\n                            to_rgb=True),\n                        dict(type='Pad', size_divisor=32),\n                        dict(type='DefaultFormatBundle'),\n                        dict(\n                            type='Collect',\n                            keys=['img', 'gt_bboxes', 'gt_labels'],\n                            meta_keys=('filename', 'ori_filename', 'ori_shape',\n                                       'img_shape', 'pad_shape',\n                                       'scale_factor', 'flip',\n                                       'flip_direction', 'img_norm_cfg',\n                                       'tag'))\n                    ],\n                    is_seq=True)\n            ],\n            ann_file='data/train_20p_unlabeled/empty_annfiles/',\n            img_prefix='data/train_20p_unlabeled/images/',\n            version='le90',\n            classes=('plane', 'baseball-diamond', 'bridge',\n                     'ground-track-field', 'small-vehicle', 'large-vehicle',\n                     'ship', 'tennis-court', 'basketball-court',\n                     'storage-tank', 'soccer-ball-field', 'roundabout',\n                     'harbor', 'swimming-pool', 'helicopter',\n                     'container-crane'),\n            filter_empty_gt=False)),\n    val=dict(\n        type='DOTAv15WSOODDataset',\n        pipeline=[\n            dict(type='LoadImageFromFile'),\n            dict(\n                type='MultiScaleFlipAug',\n                img_scale=(1024, 1024),\n                flip=False,\n                transforms=[\n                    dict(type='RResize'),\n                    dict(\n                        type='Normalize',\n                        mean=[123.675, 116.28, 103.53],\n                        std=[58.395, 57.12, 57.375],\n                        to_rgb=True),\n                    dict(type='Pad', size_divisor=32),\n                    dict(type='DefaultFormatBundle'),\n                    dict(type='Collect', keys=['img'])\n                ])\n        ],\n        img_prefix='data/val/images/',\n        ann_file='data/val/annfiles/',\n        version='le90',\n        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n                 'basketball-court', 'storage-tank', 'soccer-ball-field',\n                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n                 'container-crane')),\n    test=dict(\n        type='DOTAv15WSOODDataset',\n        pipeline=[\n            dict(type='LoadImageFromFile'),\n            dict(\n                type='MultiScaleFlipAug',\n                img_scale=(1024, 1024),\n                flip=False,\n                transforms=[\n                    dict(type='RResize'),\n                    dict(\n                        type='Normalize',\n                        mean=[123.675, 116.28, 103.53],\n                        std=[58.395, 57.12, 57.375],\n                        to_rgb=True),\n                    dict(type='Pad', size_divisor=32),\n                    dict(type='DefaultFormatBundle'),\n                    dict(type='Collect', keys=['img'])\n                ])\n        ],\n        img_prefix='data/vis_val/images/',\n        ann_file='data/vis_val/annfiles/',\n        version='le90',\n        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',\n                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',\n                 'basketball-court', 'storage-tank', 'soccer-ball-field',\n                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',\n                 'container-crane')),\n    sampler=dict(\n        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))\ncustom_hooks = [\n    dict(type='NumClassCheckHook'),\n    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)\n]\nevaluation = dict(\n    type='SubModulesDistEvalHook',\n    interval=3200,\n    metric='mAP',\n    save_best='mAP')\noptimizer = dict(\n    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)\noptimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))\nlr_config = dict(\n    policy='step',\n    warmup='linear',\n    warmup_iters=500,\n    warmup_ratio=0.3333333333333333,\n    step=120000)\nrunner = dict(type='IterBasedRunner', max_iters=120000)\ncheckpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)\nlog_config = dict(\n    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nload_from = None\nresume_from = 'work_dir/h2rv2_mcl/ablation/0.025/iter_28800.pth'\nworkflow = [('train', 1)]\nopencv_num_threads = 0\nmp_start_method = 'fork'\nwork_dir = 'work_dir/h2rv2_mcl/ablation/0.025/'\nauto_resume = False\ngpu_ids = range(0, 2)\n", "seed": 42, "exp_name": "lr_0.025.py"}

work_dir_h/PWOOD/ablation/0.025/lr_0.025.py

@@ -0,0 +1,465 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLossV2', cls_channels=16, thres=0.025),
+    train_cfg=dict(
+        iter_count=28800,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = 'work_dir/h2rv2_mcl/ablation/0.025/iter_28800.pth'
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/ablation/0.025/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/ablation/dior/0.01/lr_0.01.py

@@ -0,0 +1,465 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=20,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[2, 5, 9, 14, 15, 19],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=20,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[2, 5, 9, 14, 15, 19],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLossV2', cls_channels=20, thres=0.01),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=20),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(800, 800)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(800, 800)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(800, 800)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(800, 800),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DIOR_DOTAWSOODDataset'
+classes = ('airplane', 'airport', 'baseballfield', 'basketballcourt', 'bridge',
+           'chimney', 'expressway-service-area', 'expressway-toll-station',
+           'dam', 'golffield', 'groundtrackfield', 'harbor', 'overpass',
+           'ship', 'stadium', 'storagetank', 'tenniscourt', 'trainstation',
+           'vehicle', 'windmill')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DIOR_DOTAWSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(800, 800)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data_dior/train_20p_labeled_txt/',
+            img_prefix='data_dior/train_20p_labeled_jpg/',
+            version='le90',
+            classes=('airplane', 'airport', 'baseballfield', 'basketballcourt',
+                     'bridge', 'chimney', 'expressway-service-area',
+                     'expressway-toll-station', 'dam', 'golffield',
+                     'groundtrackfield', 'harbor', 'overpass', 'ship',
+                     'stadium', 'storagetank', 'tenniscourt', 'trainstation',
+                     'vehicle', 'windmill')),
+        unsup=dict(
+            type='DIOR_DOTAWSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(800, 800)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data_dior/train_20p_unlabeled_txt/',
+            img_prefix='data_dior/train_20p_unlabeled_jpg/',
+            version='le90',
+            classes=('airplane', 'airport', 'baseballfield', 'basketballcourt',
+                     'bridge', 'chimney', 'expressway-service-area',
+                     'expressway-toll-station', 'dam', 'golffield',
+                     'groundtrackfield', 'harbor', 'overpass', 'ship',
+                     'stadium', 'storagetank', 'tenniscourt', 'trainstation',
+                     'vehicle', 'windmill'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DIOR_DOTAWSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(800, 800),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data_dior/test_jpg/',
+        ann_file='data_dior/test_txt/',
+        version='le90',
+        classes=('airplane', 'airport', 'baseballfield', 'basketballcourt',
+                 'bridge', 'chimney', 'expressway-service-area',
+                 'expressway-toll-station', 'dam', 'golffield',
+                 'groundtrackfield', 'harbor', 'overpass', 'ship', 'stadium',
+                 'storagetank', 'tenniscourt', 'trainstation', 'vehicle',
+                 'windmill')),
+    test=dict(
+        type='DIOR_DOTAWSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(800, 800),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data_dior/test_jpg/',
+        ann_file='data_dior/test_txt/',
+        version='le90',
+        classes=('airplane', 'airport', 'baseballfield', 'basketballcourt',
+                 'bridge', 'chimney', 'expressway-service-area',
+                 'expressway-toll-station', 'dam', 'golffield',
+                 'groundtrackfield', 'harbor', 'overpass', 'ship', 'stadium',
+                 'storagetank', 'tenniscourt', 'trainstation', 'vehicle',
+                 'windmill')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/ablation/dior/0.01/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/ablation/dior/0.015/lr_0.015.py

@@ -0,0 +1,465 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=20,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[2, 5, 9, 14, 15, 19],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=20,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[2, 5, 9, 14, 15, 19],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLossV2', cls_channels=20, thres=0.015),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=20),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(800, 800)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(800, 800)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(800, 800)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(800, 800),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DIOR_DOTAWSOODDataset'
+classes = ('airplane', 'airport', 'baseballfield', 'basketballcourt', 'bridge',
+           'chimney', 'expressway-service-area', 'expressway-toll-station',
+           'dam', 'golffield', 'groundtrackfield', 'harbor', 'overpass',
+           'ship', 'stadium', 'storagetank', 'tenniscourt', 'trainstation',
+           'vehicle', 'windmill')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DIOR_DOTAWSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(800, 800)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data_dior/train_20p_labeled_txt/',
+            img_prefix='data_dior/train_20p_labeled_jpg/',
+            version='le90',
+            classes=('airplane', 'airport', 'baseballfield', 'basketballcourt',
+                     'bridge', 'chimney', 'expressway-service-area',
+                     'expressway-toll-station', 'dam', 'golffield',
+                     'groundtrackfield', 'harbor', 'overpass', 'ship',
+                     'stadium', 'storagetank', 'tenniscourt', 'trainstation',
+                     'vehicle', 'windmill')),
+        unsup=dict(
+            type='DIOR_DOTAWSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(800, 800)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data_dior/train_20p_unlabeled_txt/',
+            img_prefix='data_dior/train_20p_unlabeled_jpg/',
+            version='le90',
+            classes=('airplane', 'airport', 'baseballfield', 'basketballcourt',
+                     'bridge', 'chimney', 'expressway-service-area',
+                     'expressway-toll-station', 'dam', 'golffield',
+                     'groundtrackfield', 'harbor', 'overpass', 'ship',
+                     'stadium', 'storagetank', 'tenniscourt', 'trainstation',
+                     'vehicle', 'windmill'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DIOR_DOTAWSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(800, 800),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data_dior/test_jpg/',
+        ann_file='data_dior/test_txt/',
+        version='le90',
+        classes=('airplane', 'airport', 'baseballfield', 'basketballcourt',
+                 'bridge', 'chimney', 'expressway-service-area',
+                 'expressway-toll-station', 'dam', 'golffield',
+                 'groundtrackfield', 'harbor', 'overpass', 'ship', 'stadium',
+                 'storagetank', 'tenniscourt', 'trainstation', 'vehicle',
+                 'windmill')),
+    test=dict(
+        type='DIOR_DOTAWSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(800, 800),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data_dior/test_jpg/',
+        ann_file='data_dior/test_txt/',
+        version='le90',
+        classes=('airplane', 'airport', 'baseballfield', 'basketballcourt',
+                 'bridge', 'chimney', 'expressway-service-area',
+                 'expressway-toll-station', 'dam', 'golffield',
+                 'groundtrackfield', 'harbor', 'overpass', 'ship', 'stadium',
+                 'storagetank', 'tenniscourt', 'trainstation', 'vehicle',
+                 'windmill')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/ablation/dior/0.015/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/ablation/middle/gmm.py

@@ -0,0 +1,465 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='SemiGMMLoss', cls_channels=16, policy='middle'),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=2)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/ablation/middle/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/ablation/ntopk/gmm_ntopk.py

@@ -0,0 +1,465 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacherLog',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='SemiGMMLossNtopk', cls_channels=16, policy='high'),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'DEBUG'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/ablation/ntopk/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/ablation/topk_0.03/lr_0.03.py

@@ -0,0 +1,465 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='SemiRotatedFCOSHeadH2RV2MCL',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        center_sampling=True,
+        center_sample_radius=1.5,
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        square_cls=[1, 9, 11],
+        resize_cls=[1],
+        scale_angle=False,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+        loss_ss_symmetry=dict(type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='H2RV2MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='SemiRotatedFCOSHeadH2RV2MCL',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            center_sampling=True,
+            center_sample_radius=1.5,
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            square_cls=[1, 9, 11],
+            resize_cls=[1],
+            scale_angle=False,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
+            loss_ss_symmetry=dict(
+                type='SmoothL1Loss', loss_weight=0.2, beta=0.1)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLossV2', cls_channels=16, thres=0.03),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='exp',
+        logit_specific_weights=dict(),
+        cls_channels=16),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTAv15WSOODDataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ],
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane')),
+        unsup=dict(
+            type='DOTAv15WSOODDataset',
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            version='le90',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    test=dict(
+        type='DOTAv15WSOODDataset',
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ],
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        version='le90',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane')),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/ablation/topk_0.03/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/base/10p/mcl_fcos_adamw_dota15_10p.py

@@ -0,0 +1,453 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='BaseHead',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        separate_angle=False,
+        scale_angle=True,
+        beta=0.2,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='RotatedIoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='BaseHead',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            separate_angle=False,
+            scale_angle=True,
+            beta=0.2,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='RotatedIoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLoss', cls_channels=16),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=6400,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='linear',
+        logit_specific_weights=dict()),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTADataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTADataset',
+            ann_file='data/train_10p_labeled/annfiles/',
+            img_prefix='data/train_10p_labeled/images/',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ]),
+        unsup=dict(
+            type='DOTADataset',
+            ann_file='data/train_10p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_10p_unlabeled/images/',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTADataset',
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane'),
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ]),
+    test=dict(
+        type='DOTADataset',
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane'),
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ]),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/base/10p/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/base/20p/mcl_fcos_adamw_dota15_20p.py

@@ -0,0 +1,453 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='BaseHead',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        separate_angle=False,
+        scale_angle=True,
+        beta=0.2,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='RotatedIoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='BaseHead',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            separate_angle=False,
+            scale_angle=True,
+            beta=0.2,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='RotatedIoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLoss', cls_channels=16),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='linear',
+        logit_specific_weights=dict()),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTADataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTADataset',
+            ann_file='data/train_20p_labeled/annfiles/',
+            img_prefix='data/train_20p_labeled/images/',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ]),
+        unsup=dict(
+            type='DOTADataset',
+            ann_file='data/train_20p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_20p_unlabeled/images/',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTADataset',
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane'),
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ]),
+    test=dict(
+        type='DOTADataset',
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane'),
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ]),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=120000)
+runner = dict(type='IterBasedRunner', max_iters=120000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/base/20p/'
+auto_resume = False
+gpu_ids = range(0, 2)

work_dir_h/PWOOD/base/30p/mcl_fcos_adamw_dota15_30p.py

@@ -0,0 +1,453 @@
+angle_version = 'le90'
+detector = dict(
+    type='SemiRotatedFCOS',
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        zero_init_residual=False,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='BaseHead',
+        num_classes=16,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        norm_on_bbox=True,
+        centerness_on_reg=True,
+        separate_angle=False,
+        scale_angle=True,
+        beta=0.2,
+        bbox_coder=dict(type='DistanceAnglePointCoder', angle_version='le90'),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='RotatedIoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0)),
+    train_cfg=None,
+    test_cfg=dict(
+        nms_pre=2000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(iou_thr=0.1),
+        max_per_img=2000))
+model = dict(
+    type='MCLTeacher',
+    model=dict(
+        type='SemiRotatedFCOS',
+        backbone=dict(
+            type='ResNet',
+            depth=50,
+            num_stages=4,
+            out_indices=(0, 1, 2, 3),
+            frozen_stages=1,
+            zero_init_residual=False,
+            norm_cfg=dict(type='BN', requires_grad=True),
+            norm_eval=True,
+            style='pytorch',
+            init_cfg=dict(
+                type='Pretrained', checkpoint='torchvision://resnet50')),
+        neck=dict(
+            type='FPN',
+            in_channels=[256, 512, 1024, 2048],
+            out_channels=256,
+            start_level=1,
+            add_extra_convs='on_output',
+            num_outs=5,
+            relu_before_extra_convs=True),
+        bbox_head=dict(
+            type='BaseHead',
+            num_classes=16,
+            in_channels=256,
+            stacked_convs=4,
+            feat_channels=256,
+            strides=[8, 16, 32, 64, 128],
+            norm_on_bbox=True,
+            centerness_on_reg=True,
+            separate_angle=False,
+            scale_angle=True,
+            beta=0.2,
+            bbox_coder=dict(
+                type='DistanceAnglePointCoder', angle_version='le90'),
+            loss_cls=dict(
+                type='FocalLoss',
+                use_sigmoid=True,
+                gamma=2.0,
+                alpha=0.25,
+                loss_weight=1.0),
+            loss_bbox=dict(type='RotatedIoULoss', loss_weight=1.0),
+            loss_centerness=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0)),
+        train_cfg=None,
+        test_cfg=dict(
+            nms_pre=2000,
+            min_bbox_size=0,
+            score_thr=0.05,
+            nms=dict(iou_thr=0.1),
+            max_per_img=2000)),
+    semi_loss=dict(type='RotatedMCLLoss', cls_channels=16),
+    train_cfg=dict(
+        iter_count=0,
+        burn_in_steps=12800,
+        sup_weight=1.0,
+        unsup_weight=1.0,
+        weight_suppress='linear',
+        logit_specific_weights=dict()),
+    test_cfg=dict(inference_on='teacher'))
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+common_pipeline = [
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+strong_pipeline = [
+    dict(type='DTToPILImage'),
+    dict(
+        type='DTRandomApply',
+        operations=[
+            ColorJitter(
+                brightness=[0.6, 1.4],
+                contrast=[0.6, 1.4],
+                saturation=[0.6, 1.4],
+                hue=[-0.1, 0.1])
+        ],
+        p=0.8),
+    dict(type='DTRandomGrayscale', p=0.2),
+    dict(
+        type='DTRandomApply',
+        operations=[dict(type='DTGaussianBlur', rad_range=[0.1, 2.0])]),
+    dict(type='DTToNumpy'),
+    dict(type='ExtraAttrs', tag='unsup_strong')
+]
+weak_pipeline = [
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='unsup_weak')
+]
+unsup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadEmptyAnnotations', with_bbox=True),
+    dict(
+        type='STMultiBranch',
+        unsup_strong=[
+            dict(type='DTToPILImage'),
+            dict(
+                type='DTRandomApply',
+                operations=[
+                    ColorJitter(
+                        brightness=[0.6, 1.4],
+                        contrast=[0.6, 1.4],
+                        saturation=[0.6, 1.4],
+                        hue=[-0.1, 0.1])
+                ],
+                p=0.8),
+            dict(type='DTRandomGrayscale', p=0.2),
+            dict(
+                type='DTRandomApply',
+                operations=[dict(type='DTGaussianBlur', rad_range=[0.1,
+                                                                   2.0])]),
+            dict(type='DTToNumpy'),
+            dict(type='ExtraAttrs', tag='unsup_strong')
+        ],
+        unsup_weak=[
+            dict(type='RResize', img_scale=(1024, 1024)),
+            dict(
+                type='RRandomFlip',
+                flip_ratio=[0.25, 0.25, 0.25],
+                direction=['horizontal', 'vertical', 'diagonal'],
+                version='le90'),
+            dict(type='ExtraAttrs', tag='unsup_weak')
+        ],
+        common_pipeline=[
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(
+                type='Collect',
+                keys=['img', 'gt_bboxes', 'gt_labels'],
+                meta_keys=('filename', 'ori_filename', 'ori_shape',
+                           'img_shape', 'pad_shape', 'scale_factor', 'flip',
+                           'flip_direction', 'img_norm_cfg', 'tag'))
+        ],
+        is_seq=True)
+]
+sup_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RResize', img_scale=(1024, 1024)),
+    dict(
+        type='RRandomFlip',
+        flip_ratio=[0.25, 0.25, 0.25],
+        direction=['horizontal', 'vertical', 'diagonal'],
+        version='le90'),
+    dict(type='ExtraAttrs', tag='sup_weak'),
+    dict(
+        type='Normalize',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        to_rgb=True),
+    dict(type='Pad', size_divisor=32),
+    dict(type='DefaultFormatBundle'),
+    dict(
+        type='Collect',
+        keys=['img', 'gt_bboxes', 'gt_labels'],
+        meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
+                   'pad_shape', 'scale_factor', 'flip', 'flip_direction',
+                   'img_norm_cfg', 'tag'))
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(1024, 1024),
+        flip=False,
+        transforms=[
+            dict(type='RResize'),
+            dict(
+                type='Normalize',
+                mean=[123.675, 116.28, 103.53],
+                std=[58.395, 57.12, 57.375],
+                to_rgb=True),
+            dict(type='Pad', size_divisor=32),
+            dict(type='DefaultFormatBundle'),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+dataset_type = 'DOTADataset'
+classes = ('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+           'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+           'basketball-court', 'storage-tank', 'soccer-ball-field',
+           'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+           'container-crane')
+data = dict(
+    samples_per_gpu=3,
+    workers_per_gpu=5,
+    train=dict(
+        type='SemiDataset',
+        sup=dict(
+            type='DOTADataset',
+            ann_file='data/train_30p_labeled/annfiles/',
+            img_prefix='data/train_30p_labeled/images/',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadAnnotations', with_bbox=True),
+                dict(type='RResize', img_scale=(1024, 1024)),
+                dict(
+                    type='RRandomFlip',
+                    flip_ratio=[0.25, 0.25, 0.25],
+                    direction=['horizontal', 'vertical', 'diagonal'],
+                    version='le90'),
+                dict(type='ExtraAttrs', tag='sup_weak'),
+                dict(
+                    type='Normalize',
+                    mean=[123.675, 116.28, 103.53],
+                    std=[58.395, 57.12, 57.375],
+                    to_rgb=True),
+                dict(type='Pad', size_divisor=32),
+                dict(type='DefaultFormatBundle'),
+                dict(
+                    type='Collect',
+                    keys=['img', 'gt_bboxes', 'gt_labels'],
+                    meta_keys=('filename', 'ori_filename', 'ori_shape',
+                               'img_shape', 'pad_shape', 'scale_factor',
+                               'flip', 'flip_direction', 'img_norm_cfg',
+                               'tag'))
+            ]),
+        unsup=dict(
+            type='DOTADataset',
+            ann_file='data/train_30p_unlabeled/empty_annfiles/',
+            img_prefix='data/train_30p_unlabeled/images/',
+            classes=('plane', 'baseball-diamond', 'bridge',
+                     'ground-track-field', 'small-vehicle', 'large-vehicle',
+                     'ship', 'tennis-court', 'basketball-court',
+                     'storage-tank', 'soccer-ball-field', 'roundabout',
+                     'harbor', 'swimming-pool', 'helicopter',
+                     'container-crane'),
+            pipeline=[
+                dict(type='LoadImageFromFile'),
+                dict(type='LoadEmptyAnnotations', with_bbox=True),
+                dict(
+                    type='STMultiBranch',
+                    unsup_strong=[
+                        dict(type='DTToPILImage'),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                ColorJitter(
+                                    brightness=[0.6, 1.4],
+                                    contrast=[0.6, 1.4],
+                                    saturation=[0.6, 1.4],
+                                    hue=[-0.1, 0.1])
+                            ],
+                            p=0.8),
+                        dict(type='DTRandomGrayscale', p=0.2),
+                        dict(
+                            type='DTRandomApply',
+                            operations=[
+                                dict(
+                                    type='DTGaussianBlur',
+                                    rad_range=[0.1, 2.0])
+                            ]),
+                        dict(type='DTToNumpy'),
+                        dict(type='ExtraAttrs', tag='unsup_strong')
+                    ],
+                    unsup_weak=[
+                        dict(type='RResize', img_scale=(1024, 1024)),
+                        dict(
+                            type='RRandomFlip',
+                            flip_ratio=[0.25, 0.25, 0.25],
+                            direction=['horizontal', 'vertical', 'diagonal'],
+                            version='le90'),
+                        dict(type='ExtraAttrs', tag='unsup_weak')
+                    ],
+                    common_pipeline=[
+                        dict(
+                            type='Normalize',
+                            mean=[123.675, 116.28, 103.53],
+                            std=[58.395, 57.12, 57.375],
+                            to_rgb=True),
+                        dict(type='Pad', size_divisor=32),
+                        dict(type='DefaultFormatBundle'),
+                        dict(
+                            type='Collect',
+                            keys=['img', 'gt_bboxes', 'gt_labels'],
+                            meta_keys=('filename', 'ori_filename', 'ori_shape',
+                                       'img_shape', 'pad_shape',
+                                       'scale_factor', 'flip',
+                                       'flip_direction', 'img_norm_cfg',
+                                       'tag'))
+                    ],
+                    is_seq=True)
+            ],
+            filter_empty_gt=False)),
+    val=dict(
+        type='DOTADataset',
+        img_prefix='data/val/images/',
+        ann_file='data/val/annfiles/',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane'),
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ]),
+    test=dict(
+        type='DOTADataset',
+        img_prefix='data/vis_val/images/',
+        ann_file='data/vis_val/annfiles/',
+        classes=('plane', 'baseball-diamond', 'bridge', 'ground-track-field',
+                 'small-vehicle', 'large-vehicle', 'ship', 'tennis-court',
+                 'basketball-court', 'storage-tank', 'soccer-ball-field',
+                 'roundabout', 'harbor', 'swimming-pool', 'helicopter',
+                 'container-crane'),
+        pipeline=[
+            dict(type='LoadImageFromFile'),
+            dict(
+                type='MultiScaleFlipAug',
+                img_scale=(1024, 1024),
+                flip=False,
+                transforms=[
+                    dict(type='RResize'),
+                    dict(
+                        type='Normalize',
+                        mean=[123.675, 116.28, 103.53],
+                        std=[58.395, 57.12, 57.375],
+                        to_rgb=True),
+                    dict(type='Pad', size_divisor=32),
+                    dict(type='DefaultFormatBundle'),
+                    dict(type='Collect', keys=['img'])
+                ])
+        ]),
+    sampler=dict(
+        train=dict(type='MultiSourceSampler', sample_ratio=[2, 1], seed=42)))
+custom_hooks = [
+    dict(type='NumClassCheckHook'),
+    dict(type='MeanTeacher', momentum=0.9996, interval=1, start_steps=3200)
+]
+evaluation = dict(
+    type='SubModulesDistEvalHook',
+    interval=3200,
+    metric='mAP',
+    save_best='mAP')
+optimizer = dict(
+    type='AdamW', lr=0.0001, betas=(0.9, 0.999), weight_decay=0.05)
+optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.3333333333333333,
+    step=[120000, 160000])
+runner = dict(type='IterBasedRunner', max_iters=180000)
+checkpoint_config = dict(by_epoch=False, interval=3200, max_keep_ckpts=1)
+log_config = dict(
+    _delete_=True, interval=50, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+opencv_num_threads = 0
+mp_start_method = 'fork'
+work_dir = 'work_dir/h2rv2_mcl/base/30p/'
+auto_resume = False
+gpu_ids = range(0, 2)