2025-07-07 11:27:50,676 - PropVG - INFO - dataset = 'MixedSeg' data_root = './data/seqtr_type/' img_norm_cfg = dict( mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]) train_pipeline = [ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file='data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ] val_pipeline = [ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file='data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ] test_pipeline = [ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file='data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ] word_emb_cfg = dict(type='GloVe') data = dict( samples_per_gpu=8, workers_per_gpu=4, train=dict( type='MixedSeg', which_set='train', img_source=['coco'], annsfile= './data/seqtr_type/annotations/mixed-seg/instances_nogoogle_withid.json', imgsfile='./data/seqtr_type/images/mscoco/train2014', pipeline=[ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file= 'data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ], word_emb_cfg=dict(type='GloVe')), val_refcoco_unc=dict( type='MixedSeg', which_set='val_refcoco_unc', img_source=['coco'], annsfile= './data/seqtr_type/annotations/mixed-seg/instances_nogoogle_withid.json', imgsfile='./data/seqtr_type/images/mscoco/train2014', pipeline=[ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file= 'data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ], word_emb_cfg=dict(type='GloVe')), testA_refcoco_unc=dict( type='MixedSeg', which_set='testA_refcoco_unc', img_source=['coco'], annsfile= './data/seqtr_type/annotations/mixed-seg/instances_nogoogle_withid.json', imgsfile='./data/seqtr_type/images/mscoco/train2014', pipeline=[ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file= 'data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ], word_emb_cfg=dict(type='GloVe')), testB_refcoco_unc=dict( type='MixedSeg', which_set='testB_refcoco_unc', img_source=['coco'], annsfile= './data/seqtr_type/annotations/mixed-seg/instances_nogoogle_withid.json', imgsfile='./data/seqtr_type/images/mscoco/train2014', pipeline=[ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file= 'data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ], word_emb_cfg=dict(type='GloVe')), val_refcocoplus_unc=dict( type='MixedSeg', which_set='val_refcocoplus_unc', img_source=['coco'], annsfile= './data/seqtr_type/annotations/mixed-seg/instances_nogoogle_withid.json', imgsfile='./data/seqtr_type/images/mscoco/train2014', pipeline=[ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file= 'data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ], word_emb_cfg=dict(type='GloVe')), testA_refcocoplus_unc=dict( type='MixedSeg', which_set='testA_refcocoplus_unc', img_source=['coco'], annsfile= './data/seqtr_type/annotations/mixed-seg/instances_nogoogle_withid.json', imgsfile='./data/seqtr_type/images/mscoco/train2014', pipeline=[ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file= 'data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ], word_emb_cfg=dict(type='GloVe')), testB_refcocoplus_unc=dict( type='MixedSeg', which_set='testB_refcocoplus_unc', img_source=['coco'], annsfile= './data/seqtr_type/annotations/mixed-seg/instances_nogoogle_withid.json', imgsfile='./data/seqtr_type/images/mscoco/train2014', pipeline=[ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file= 'data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ], word_emb_cfg=dict(type='GloVe')), val_refcocog_umd=dict( type='MixedSeg', which_set='val_refcocog_umd', img_source=['coco'], annsfile= './data/seqtr_type/annotations/mixed-seg/instances_nogoogle_withid.json', imgsfile='./data/seqtr_type/images/mscoco/train2014', pipeline=[ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file= 'data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ], word_emb_cfg=dict(type='GloVe')), test_refcocog_umd=dict( type='MixedSeg', which_set='test_refcocog_umd', img_source=['coco'], annsfile= './data/seqtr_type/annotations/mixed-seg/instances_nogoogle_withid.json', imgsfile='./data/seqtr_type/images/mscoco/train2014', pipeline=[ dict( type='LoadImageAnnotationsFromFile_TO', max_token=20, with_mask=True, with_bbox=True, dataset='MixedSeg', use_token_type='beit3', refer_file= 'data/seqtr_type/annotations/mixed-seg/coco_all.json', object_area_filter=100, object_area_rate_filter=[0.05, 0.8]), dict(type='Resize', img_scale=(384, 384), keep_ratio=False), dict( type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]), dict(type='DefaultFormatBundle'), dict( type='CollectData', keys=[ 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', 'gt_bbox' ], meta_keys=[ 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', 'empty', 'refer_target_index' ]) ], word_emb_cfg=dict(type='GloVe'))) ema = False ema_factor = 0.999 use_fp16 = False seed = 6666 deterministic = True log_level = 'INFO' log_interval = 50 save_interval = -1 resume_from = None load_from = 'work_dir/refcoco-mix/PropVG-refcoco-mix.pth' finetune_from = None evaluate_interval = 1 start_evaluate_epoch = 0 start_save_checkpoint = 20 max_token = 20 img_size = 384 patch_size = 16 model = dict( type='MIXRefUniModel_OMG', vis_enc=dict( type='BEIT3', img_size=384, patch_size=16, vit_type='base', drop_path_rate=0.1, vocab_size=64010, freeze_layer=-1, vision_embed_proj_interpolate=False, pretrain='pretrain_weights/beit3_base_patch16_224.zip'), lan_enc=None, fusion=None, head=dict( type='REFHead', input_channels=768, hidden_channels=256, num_queries=20, detr_loss=dict( criterion=dict(loss_class=1.0, loss_bbox=5.0, loss_giou=2.0), matcher=dict(cost_class=1.0, cost_bbox=5.0, cost_giou=2.0)), loss_weight=dict( mask=dict(dice=1.0, bce=1.0, nt=0.2, neg=0), bbox=0.1, allbbox=0.1, refer=1.0), MTD=dict(K=100)), post_params=dict( score_weighted=False, mask_threshold=0.5, score_threshold=0.7, with_nms=False, with_mask=True), process_visual=False, visualize_params=dict(row_columns=(4, 5)), visual_mode='test') grad_norm_clip = 0.15 lr = 0.0005 optimizer_config = dict( type='Adam', lr=0.0005, lr_vis_enc=5e-05, lr_lan_enc=0.0005, betas=(0.9, 0.98), eps=1e-09, weight_decay=0, amsgrad=True) scheduler_config = dict( type='MultiStepLRWarmUp', warmup_epochs=1, decay_steps=[21, 27], decay_ratio=0.1, max_epoch=30) launcher = 'pytorch' distributed = True rank = 0 world_size = 1 2025-07-07 11:27:58,403 - PropVG - INFO - Mixed-val_refcoco_unc size: 10834 2025-07-07 11:28:06,594 - PropVG - INFO - Mixed-testA_refcoco_unc size: 5657 2025-07-07 11:28:15,164 - PropVG - INFO - Mixed-testB_refcoco_unc size: 5095 2025-07-07 11:28:23,677 - PropVG - INFO - Mixed-val_refcocoplus_unc size: 10758 2025-07-07 11:28:30,907 - PropVG - INFO - Mixed-testA_refcocoplus_unc size: 5726 2025-07-07 11:28:38,494 - PropVG - INFO - Mixed-testB_refcocoplus_unc size: 4889 2025-07-07 11:28:49,090 - PropVG - INFO - Mixed-val_refcocog_umd size: 4896 2025-07-07 11:28:54,576 - PropVG - INFO - Mixed-test_refcocog_umd size: 9602 2025-07-07 11:29:02,664 - PropVG - INFO - loaded checkpoint from work_dir/refcoco-mix/PropVG-refcoco-mix.pth 2025-07-07 11:29:02,665 - PropVG - INFO - PropVG - evaluating set val_refcoco_unc 2025-07-07 11:32:39,213 - PropVG - INFO - ------------ validate ------------ time: 216.54, DetACC: 92.70, mIoU: 81.96, oIoU: 81.80, MaskACC@0.5-0.9: [92.24, 90.71, 87.59, 79.79, 46.59]DetACC@0.5-0.9: [92.70, 91.43, 88.90, 83.85, 66.30] 2025-07-07 11:32:43,474 - PropVG - INFO - PropVG - evaluating set testA_refcoco_unc 2025-07-07 11:34:47,838 - PropVG - INFO - ------------ validate ------------ time: 124.36, DetACC: 95.07, mIoU: 83.58, oIoU: 83.74, MaskACC@0.5-0.9: [94.56, 93.48, 90.93, 82.91, 46.61]DetACC@0.5-0.9: [95.07, 93.99, 92.17, 88.17, 69.29] 2025-07-07 11:34:53,297 - PropVG - INFO - PropVG - evaluating set testB_refcoco_unc 2025-07-07 11:36:51,290 - PropVG - INFO - ------------ validate ------------ time: 117.99, DetACC: 89.58, mIoU: 80.02, oIoU: 79.33, MaskACC@0.5-0.9: [89.19, 86.99, 83.45, 76.76, 51.07]DetACC@0.5-0.9: [89.58, 87.56, 84.61, 79.14, 61.83] 2025-07-07 11:36:56,652 - PropVG - INFO - PropVG - evaluating set val_refcocoplus_unc 2025-07-07 11:40:28,540 - PropVG - INFO - ------------ validate ------------ time: 211.88, DetACC: 87.27, mIoU: 77.14, oIoU: 74.81, MaskACC@0.5-0.9: [86.67, 85.36, 82.52, 75.28, 44.34]DetACC@0.5-0.9: [87.27, 86.30, 84.09, 79.64, 63.62] 2025-07-07 11:40:36,392 - PropVG - INFO - PropVG - evaluating set testA_refcocoplus_unc 2025-07-07 11:42:43,800 - PropVG - INFO - ------------ validate ------------ time: 127.40, DetACC: 90.87, mIoU: 79.83, oIoU: 78.72, MaskACC@0.5-0.9: [90.13, 88.79, 86.57, 79.46, 45.04]DetACC@0.5-0.9: [90.87, 89.82, 87.81, 83.92, 66.33] 2025-07-07 11:42:48,169 - PropVG - INFO - PropVG - evaluating set testB_refcocoplus_unc 2025-07-07 11:44:41,261 - PropVG - INFO - ------------ validate ------------ time: 113.09, DetACC: 81.26, mIoU: 72.18, oIoU: 69.15, MaskACC@0.5-0.9: [80.18, 78.20, 74.78, 68.68, 45.88]DetACC@0.5-0.9: [81.26, 79.40, 76.95, 72.20, 56.78] 2025-07-07 11:44:45,751 - PropVG - INFO - PropVG - evaluating set val_refcocog_umd 2025-07-07 11:46:42,173 - PropVG - INFO - ------------ validate ------------ time: 116.42, DetACC: 88.15, mIoU: 76.97, oIoU: 75.54, MaskACC@0.5-0.9: [86.17, 83.58, 79.43, 72.16, 44.87]DetACC@0.5-0.9: [88.15, 85.97, 82.90, 78.00, 63.09] 2025-07-07 11:46:46,257 - PropVG - INFO - PropVG - evaluating set test_refcocog_umd 2025-07-07 11:50:06,821 - PropVG - INFO - ------------ validate ------------ time: 200.56, DetACC: 88.30, mIoU: 77.72, oIoU: 77.40, MaskACC@0.5-0.9: [87.14, 85.01, 80.84, 72.78, 45.79]DetACC@0.5-0.9: [88.30, 86.71, 83.98, 79.07, 65.00] 2025-07-07 11:50:11,168 - PropVG - INFO - sucessfully save the results to work_dir/refcoco-mix/refer_output_thr0.7_no-nms_no-sw_0.5_100.xlsx !!!