Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+checkpoint-54/54.pdema filter=lfs diff=lfs merge=lfs -text
+checkpoint-54/54.pdopt filter=lfs diff=lfs merge=lfs -text
+checkpoint-54/54.pdparams filter=lfs diff=lfs merge=lfs -text
+inference/inference.pdiparams filter=lfs diff=lfs merge=lfs -text

.ipynb_checkpoints/config-checkpoint.yaml

@@ -0,0 +1,285 @@
+DETR:
+  backbone: PPHGNetV2
+  detr_head: DocLayoutV3Head
+  neck: MaskHybridEncoder
+  post_process: DocLayoutV3PostProcess
+  transformer: DocLayoutV3Transformer
+DocLayoutV3Head:
+  loss:
+    aux_loss: true
+    loss_coeff:
+      bbox: 5
+      class: 4
+      dice: 5
+      giou: 2
+      mask: 5
+      order: 50
+    matcher:
+      matcher_coeff:
+        bbox: 5
+        class: 4
+        dice: 5
+        giou: 2
+        mask: 5
+      name: HungarianMatcher
+    name: DocLayoutV3Loss
+    use_focal_loss: true
+    use_vfl: true
+    vfl_iou_type: mask
+DocLayoutV3Metric:
+  eval_mask: false
+DocLayoutV3PostProcess:
+  mask_stride: 4
+  num_top_queries: 300
+  resize_mask: false
+DocLayoutV3Transformer:
+  activation: relu
+  box_noise_scale: 1.0
+  dim_feedforward: 1024
+  dropout: 0.0
+  feat_strides:
+  - 8
+  - 16
+  - 32
+  label_noise_ratio: 0.5
+  learnt_init_query: false
+  mask_enhanced: true
+  nhead: 8
+  num_denoising: 0
+  num_levels: 3
+  num_queries: 300
+  position_embed_type: sine
+EvalDataset:
+  anno_path: annotations/instance_val.json
+  dataset_dir: /root/work/paddleocr-det/PaddleX/paddlex/dataset/handwriting
+  image_dir: images
+  name: COCOInstSegDataset
+EvalReader:
+  batch_size: 1
+  drop_last: false
+  sample_transforms:
+  - Decode: {}
+  - Resize:
+      interp: 2
+      keep_ratio: false
+      target_size:
+      - 800
+      - 800
+  - NormalizeImage:
+      mean:
+      - 0.0
+      - 0.0
+      - 0.0
+      norm_type: none
+      std:
+      - 1.0
+      - 1.0
+      - 1.0
+  - Permute: {}
+  shuffle: false
+LearningRate:
+  base_lr: 0.0001
+  schedulers:
+  - !PiecewiseDecay
+    gamma:
+    - 1.0
+    milestones:
+    - 100
+    use_warmup: true
+    values: null
+  - !LinearWarmup
+    epochs: null
+    epochs_first: false
+    start_factor: 0.001
+    steps: 200
+MaskHybridEncoder:
+  encoder_layer:
+    activation: gelu
+    d_model: 256
+    dim_feedforward: 1024
+    dropout: 0.0
+    name: TransformerLayer
+    nhead: 8
+  expansion: 1.0
+  hidden_dim: 256
+  mask_feat_channels:
+  - 64
+  - 64
+  num_encoder_layers: 1
+  use_encoder_idx:
+  - 3
+OptimizerBuilder:
+  clip_grad_by_norm: 0.1
+  optimizer:
+    type: AdamW
+    weight_decay: 0.0001
+  regularizer: false
+PPHGNetV2:
+  arch: L
+  freeze_at: 0
+  freeze_norm: true
+  freeze_stem_only: true
+  lr_mult_list:
+  - 0.0
+  - 0.05
+  - 0.05
+  - 0.05
+  - 0.05
+  return_idx:
+  - 0
+  - 1
+  - 2
+  - 3
+TestDataset: !ImageFolder
+  anno_path: annotations/instance_val.json
+  dataset_dir: /root/work/paddleocr-det/PaddleX/paddlex/dataset/handwriting
+  image_dir: ''
+  sample_num: -1
+  use_default_label: null
+TestReader:
+  batch_size: 1
+  drop_last: false
+  inputs_def:
+    image_shape:
+    - 3
+    - 800
+    - 800
+  sample_transforms:
+  - Decode: {}
+  - Resize:
+      interp: 2
+      keep_ratio: false
+      target_size:
+      - 800
+      - 800
+  - NormalizeImage:
+      mean:
+      - 0.0
+      - 0.0
+      - 0.0
+      norm_type: none
+      std:
+      - 1.0
+      - 1.0
+      - 1.0
+  - Permute: {}
+  shuffle: false
+TrainDataset:
+  anno_path: annotations/instance_train.json
+  data_fields:
+  - image
+  - gt_bbox
+  - gt_class
+  - gt_poly
+  - is_crowd
+  - gt_read_order
+  dataset_dir: /root/work/paddleocr-det/PaddleX/paddlex/dataset/handwriting
+  image_dir: images
+  name: COCOInstSegDataset
+TrainReader:
+  batch_size: 8
+  batch_transforms:
+  - BatchRandomResize:
+      keep_ratio: false
+      random_interp: true
+      random_size: true
+      target_size:
+      - 672
+      - 704
+      - 736
+      - 768
+      - 800
+      - 800
+      - 800
+      - 800
+      - 832
+      - 864
+      - 896
+      - 928
+  - UnpackMask: {}
+  - NormalizeImage:
+      mean:
+      - 0.0
+      - 0.0
+      - 0.0
+      norm_type: none
+      std:
+      - 1.0
+      - 1.0
+      - 1.0
+  - NormalizeBox: {}
+  - BboxXYXY2XYWH: {}
+  - Permute: {}
+  collate_batch: false
+  drop_last: true
+  sample_transforms:
+  - Decode: {}
+  - Poly2MaskPack:
+      del_poly: true
+  - RandomDistort:
+      prob: 0.8
+  - UpdateBBoxFromMask: {}
+  - RandomExpand:
+      fill_value:
+      - 123.675
+      - 116.28
+      - 103.53
+      prob: 0.5
+      ratio: 1.5
+  - RandomCrop:
+      prob: 0.8
+      use_box_candidates: true
+  shuffle: true
+  use_shared_memory: true
+amp: false
+architecture: DETR
+ema_decay: 0.9999
+ema_decay_type: exponential
+ema_filter_no_grad: true
+enable_ce: false
+epoch: 200
+eval: true
+eval_size:
+- 800
+- 800
+export:
+  benchmark: false
+  fuse_conv_bn: false
+  nms: true
+  post_process: true
+export_with_pir: true
+filename: instancesegmodel_PP-DocLayoutV3
+find_unused_parameters: true
+fleet: false
+hidden_dim: 256
+log_iter: 10
+metric: DocLayoutV3Metric
+norm_type: sync_bn
+num_classes: 6
+num_prototypes: 32
+pdx_model_name: PP-DocLayoutV3
+pretrain_weights: /root/work/paddleocr-det/PaddleX/PP-DocLayoutV3_pretrained.pdparams
+print_flops: false
+print_params: false
+profiler_options: null
+proposals_path: sniper/proposals.json
+resume: null
+save_dir: /root/work/paddleocr-det/PaddleX/output
+save_prediction_only: false
+save_proposals: false
+snapshot_epoch: 5
+target_metrics: mask
+to_static: false
+uniform_output_enabled: true
+use_ema: true
+use_focal_loss: true
+use_gpu: true
+use_iluvatar_gpu: false
+use_mlu: false
+use_npu: false
+use_vdl: true
+use_wandb: false
+use_xpu: false
+vdl_log_dir: /root/work/paddleocr-det/PaddleX/output
+with_mask: true
+worker_num: 24

check_dataset_result.json

@@ -0,0 +1,53 @@
+{
+  "done_flag": true,
+  "check_pass": true,
+  "attributes": {
+    "num_classes": 6,
+    "train_samples": 12612,
+    "train_sample_paths": [
+      "check_dataset\/demo_img\/IMG_3087.jpg",
+      "check_dataset\/demo_img\/IMG_3203.jpg",
+      "check_dataset\/demo_img\/IMG_2603.jpg",
+      "check_dataset\/demo_img\/IMG_1558.jpg",
+      "check_dataset\/demo_img\/IMG_2656.jpg",
+      "check_dataset\/demo_img\/IMG_1405.jpg",
+      "check_dataset\/demo_img\/IMG_1291.jpg",
+      "check_dataset\/demo_img\/IMG_1471.jpg",
+      "check_dataset\/demo_img\/IMG_2073.jpg",
+      "check_dataset\/demo_img\/IMG_3326.jpg"
+    ],
+    "val_samples": 1454,
+    "val_sample_paths": [
+      "check_dataset\/demo_img\/IMG_3197.jpg",
+      "check_dataset\/demo_img\/4cfa0dec-02d9-44df-be65-67cf14562540.jpg",
+      "check_dataset\/demo_img\/IMG_3147.jpg",
+      "check_dataset\/demo_img\/IMG_2023.jpg",
+      "check_dataset\/demo_img\/IMG_1335.jpg",
+      "check_dataset\/demo_img\/IMG_3240.jpg",
+      "check_dataset\/demo_img\/IMG_2250.jpg",
+      "check_dataset\/demo_img\/IMG_3591.jpg",
+      "check_dataset\/demo_img\/IMG_3625.jpg",
+      "check_dataset\/demo_img\/IMG_3262.jpg"
+    ],
+    "read_order_validation": {
+      "instance_train": {
+        "total_images": 2119,
+        "valid_images": 2119,
+        "invalid_images": [],
+        "pass_rate": 1.0
+      },
+      "instance_val": {
+        "total_images": 236,
+        "valid_images": 236,
+        "invalid_images": [],
+        "pass_rate": 1.0
+      }
+    }
+  },
+  "analysis": {
+    "histogram": "check_dataset\/histogram.png"
+  },
+  "dataset_path": "handwriting",
+  "show_type": "image",
+  "dataset_type": "COCOInstSegDataset"
+}

checkpoint-54/.ipynb_checkpoints/54.info-checkpoint.json

@@ -0,0 +1 @@
+{"metric": 0.0, "epoch": 55}

checkpoint-54/54.info.json

@@ -0,0 +1 @@
+{"metric": 0.0, "epoch": 55}

checkpoint-54/54.pdema

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0fd8de42ff8635510bea7863908ae017e531e95bd507d2d4112b495a646b6f97
+size 133222291

checkpoint-54/54.pdopt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8ebe428c3dd4dab2ce2a586d905698f7de0641aba63e7a545692b0a594fb995
+size 265302158

checkpoint-54/54.pdparams

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e91fe515c85693f3db3e1e0244781cdc3fd316bb5327ec425c9a047b85df16eb
+size 133211611

config.yaml

@@ -0,0 +1,285 @@
+DETR:
+  backbone: PPHGNetV2
+  detr_head: DocLayoutV3Head
+  neck: MaskHybridEncoder
+  post_process: DocLayoutV3PostProcess
+  transformer: DocLayoutV3Transformer
+DocLayoutV3Head:
+  loss:
+    aux_loss: true
+    loss_coeff:
+      bbox: 5
+      class: 4
+      dice: 5
+      giou: 2
+      mask: 5
+      order: 50
+    matcher:
+      matcher_coeff:
+        bbox: 5
+        class: 4
+        dice: 5
+        giou: 2
+        mask: 5
+      name: HungarianMatcher
+    name: DocLayoutV3Loss
+    use_focal_loss: true
+    use_vfl: true
+    vfl_iou_type: mask
+DocLayoutV3Metric:
+  eval_mask: false
+DocLayoutV3PostProcess:
+  mask_stride: 4
+  num_top_queries: 300
+  resize_mask: false
+DocLayoutV3Transformer:
+  activation: relu
+  box_noise_scale: 1.0
+  dim_feedforward: 1024
+  dropout: 0.0
+  feat_strides:
+  - 8
+  - 16
+  - 32
+  label_noise_ratio: 0.5
+  learnt_init_query: false
+  mask_enhanced: true
+  nhead: 8
+  num_denoising: 0
+  num_levels: 3
+  num_queries: 300
+  position_embed_type: sine
+EvalDataset:
+  anno_path: annotations/instance_val.json
+  dataset_dir: /root/work/paddleocr-det/PaddleX/paddlex/dataset/handwriting
+  image_dir: images
+  name: COCOInstSegDataset
+EvalReader:
+  batch_size: 1
+  drop_last: false
+  sample_transforms:
+  - Decode: {}
+  - Resize:
+      interp: 2
+      keep_ratio: false
+      target_size:
+      - 800
+      - 800
+  - NormalizeImage:
+      mean:
+      - 0.0
+      - 0.0
+      - 0.0
+      norm_type: none
+      std:
+      - 1.0
+      - 1.0
+      - 1.0
+  - Permute: {}
+  shuffle: false
+LearningRate:
+  base_lr: 0.0001
+  schedulers:
+  - !PiecewiseDecay
+    gamma:
+    - 1.0
+    milestones:
+    - 100
+    use_warmup: true
+    values: null
+  - !LinearWarmup
+    epochs: null
+    epochs_first: false
+    start_factor: 0.001
+    steps: 200
+MaskHybridEncoder:
+  encoder_layer:
+    activation: gelu
+    d_model: 256
+    dim_feedforward: 1024
+    dropout: 0.0
+    name: TransformerLayer
+    nhead: 8
+  expansion: 1.0
+  hidden_dim: 256
+  mask_feat_channels:
+  - 64
+  - 64
+  num_encoder_layers: 1
+  use_encoder_idx:
+  - 3
+OptimizerBuilder:
+  clip_grad_by_norm: 0.1
+  optimizer:
+    type: AdamW
+    weight_decay: 0.0001
+  regularizer: false
+PPHGNetV2:
+  arch: L
+  freeze_at: 0
+  freeze_norm: true
+  freeze_stem_only: true
+  lr_mult_list:
+  - 0.0
+  - 0.05
+  - 0.05
+  - 0.05
+  - 0.05
+  return_idx:
+  - 0
+  - 1
+  - 2
+  - 3
+TestDataset: !ImageFolder
+  anno_path: annotations/instance_val.json
+  dataset_dir: /root/work/paddleocr-det/PaddleX/paddlex/dataset/handwriting
+  image_dir: ''
+  sample_num: -1
+  use_default_label: null
+TestReader:
+  batch_size: 1
+  drop_last: false
+  inputs_def:
+    image_shape:
+    - 3
+    - 800
+    - 800
+  sample_transforms:
+  - Decode: {}
+  - Resize:
+      interp: 2
+      keep_ratio: false
+      target_size:
+      - 800
+      - 800
+  - NormalizeImage:
+      mean:
+      - 0.0
+      - 0.0
+      - 0.0
+      norm_type: none
+      std:
+      - 1.0
+      - 1.0
+      - 1.0
+  - Permute: {}
+  shuffle: false
+TrainDataset:
+  anno_path: annotations/instance_train.json
+  data_fields:
+  - image
+  - gt_bbox
+  - gt_class
+  - gt_poly
+  - is_crowd
+  - gt_read_order
+  dataset_dir: /root/work/paddleocr-det/PaddleX/paddlex/dataset/handwriting
+  image_dir: images
+  name: COCOInstSegDataset
+TrainReader:
+  batch_size: 8
+  batch_transforms:
+  - BatchRandomResize:
+      keep_ratio: false
+      random_interp: true
+      random_size: true
+      target_size:
+      - 672
+      - 704
+      - 736
+      - 768
+      - 800
+      - 800
+      - 800
+      - 800
+      - 832
+      - 864
+      - 896
+      - 928
+  - UnpackMask: {}
+  - NormalizeImage:
+      mean:
+      - 0.0
+      - 0.0
+      - 0.0
+      norm_type: none
+      std:
+      - 1.0
+      - 1.0
+      - 1.0
+  - NormalizeBox: {}
+  - BboxXYXY2XYWH: {}
+  - Permute: {}
+  collate_batch: false
+  drop_last: true
+  sample_transforms:
+  - Decode: {}
+  - Poly2MaskPack:
+      del_poly: true
+  - RandomDistort:
+      prob: 0.8
+  - UpdateBBoxFromMask: {}
+  - RandomExpand:
+      fill_value:
+      - 123.675
+      - 116.28
+      - 103.53
+      prob: 0.5
+      ratio: 1.5
+  - RandomCrop:
+      prob: 0.8
+      use_box_candidates: true
+  shuffle: true
+  use_shared_memory: true
+amp: false
+architecture: DETR
+ema_decay: 0.9999
+ema_decay_type: exponential
+ema_filter_no_grad: true
+enable_ce: false
+epoch: 200
+eval: true
+eval_size:
+- 800
+- 800
+export:
+  benchmark: false
+  fuse_conv_bn: false
+  nms: true
+  post_process: true
+export_with_pir: true
+filename: instancesegmodel_PP-DocLayoutV3
+find_unused_parameters: true
+fleet: false
+hidden_dim: 256
+log_iter: 10
+metric: DocLayoutV3Metric
+norm_type: sync_bn
+num_classes: 6
+num_prototypes: 32
+pdx_model_name: PP-DocLayoutV3
+pretrain_weights: /root/work/paddleocr-det/PaddleX/PP-DocLayoutV3_pretrained.pdparams
+print_flops: false
+print_params: false
+profiler_options: null
+proposals_path: sniper/proposals.json
+resume: null
+save_dir: /root/work/paddleocr-det/PaddleX/output
+save_prediction_only: false
+save_proposals: false
+snapshot_epoch: 5
+target_metrics: mask
+to_static: false
+uniform_output_enabled: true
+use_ema: true
+use_focal_loss: true
+use_gpu: true
+use_iluvatar_gpu: false
+use_mlu: false
+use_npu: false
+use_vdl: true
+use_wandb: false
+use_xpu: false
+vdl_log_dir: /root/work/paddleocr-det/PaddleX/output
+with_mask: true
+worker_num: 24

inference/.ipynb_checkpoints/inference-checkpoint.yml

@@ -0,0 +1,82 @@
+mode: paddle
+draw_threshold: 0.5
+metric: DocLayoutV3Metric
+use_dynamic_shape: false
+Global:
+  model_name: PP-DocLayoutV3
+arch: DETR
+min_subgraph_size: 3
+mask: true
+Preprocess:
+- interp: 2
+  keep_ratio: false
+  target_size:
+  - 800
+  - 800
+  type: Resize
+- mean:
+  - 0.0
+  - 0.0
+  - 0.0
+  norm_type: none
+  std:
+  - 1.0
+  - 1.0
+  - 1.0
+  type: NormalizeImage
+- type: Permute
+label_list:
+- exercise
+- figure
+- question
+- table
+- text
+- unknown
+Hpi:
+  backend_configs:
+    paddle_infer:
+      trt_dynamic_shapes: &id001
+        image:
+        - - 1
+          - 3
+          - 800
+          - 800
+        - - 1
+          - 3
+          - 800
+          - 800
+        - - 8
+          - 3
+          - 800
+          - 800
+        scale_factor:
+        - - 1
+          - 2
+        - - 1
+          - 2
+        - - 8
+          - 2
+      trt_dynamic_shape_input_data:
+        scale_factor:
+        - - 2
+          - 2
+        - - 1
+          - 1
+        - - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+    tensorrt:
+      dynamic_shapes: *id001

inference/inference.pdiparams

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d60164c404fb3a2136d2e8e92c262e59358ba50d7b11e34291bfe8973a55110
+size 129235195

inference/inference.yml

@@ -0,0 +1,82 @@
+mode: paddle
+draw_threshold: 0.5
+metric: DocLayoutV3Metric
+use_dynamic_shape: false
+Global:
+  model_name: PP-DocLayoutV3
+arch: DETR
+min_subgraph_size: 3
+mask: true
+Preprocess:
+- interp: 2
+  keep_ratio: false
+  target_size:
+  - 800
+  - 800
+  type: Resize
+- mean:
+  - 0.0
+  - 0.0
+  - 0.0
+  norm_type: none
+  std:
+  - 1.0
+  - 1.0
+  - 1.0
+  type: NormalizeImage
+- type: Permute
+label_list:
+- exercise
+- figure
+- question
+- table
+- text
+- unknown
+Hpi:
+  backend_configs:
+    paddle_infer:
+      trt_dynamic_shapes: &id001
+        image:
+        - - 1
+          - 3
+          - 800
+          - 800
+        - - 1
+          - 3
+          - 800
+          - 800
+        - - 8
+          - 3
+          - 800
+          - 800
+        scale_factor:
+        - - 1
+          - 2
+        - - 1
+          - 2
+        - - 8
+          - 2
+      trt_dynamic_shape_input_data:
+        scale_factor:
+        - - 2
+          - 2
+        - - 1
+          - 1
+        - - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+          - 0.67
+    tensorrt:
+      dynamic_shapes: *id001

train_result.json

@@ -0,0 +1 @@
+{"model_name": "PP-DocLayoutV3", "label_dict": "", "visualdl_log": "", "train_log": "train.log", "config": "config.yaml", "models": {"last_1": {"score": 0.0, "pdparams": "164/164.pdparams", "pdopt": "164/164.pdopt", "pdstates": "164/164.pdstates", "pdema": "164/164.pdema", "inference_config": "164/inference/inference.yml", "pdmodel": "164/inference/inference.json", "pdiparams": "164/inference/inference.pdiparams"}, "last_2": {"score": 0.0, "pdparams": "159/159.pdparams", "pdopt": "159/159.pdopt", "pdstates": "159/159.pdstates", "pdema": "159/159.pdema", "inference_config": "159/inference/inference.yml", "pdmodel": "159/inference/inference.json", "pdiparams": "159/inference/inference.pdiparams"}, "last_3": {"score": 0.0, "pdparams": "154/154.pdparams", "pdopt": "154/154.pdopt", "pdstates": "154/154.pdstates", "pdema": "154/154.pdema", "inference_config": "154/inference/inference.yml", "pdmodel": "154/inference/inference.json", "pdiparams": "154/inference/inference.pdiparams"}, "last_4": {"score": 0.0, "pdparams": "149/149.pdparams", "pdopt": "149/149.pdopt", "pdstates": "149/149.pdstates", "pdema": "149/149.pdema", "inference_config": "149/inference/inference.yml", "pdmodel": "149/inference/inference.json", "pdiparams": "149/inference/inference.pdiparams"}, "last_5": {"score": 0.0, "pdparams": "144/144.pdparams", "pdopt": "144/144.pdopt", "pdstates": "144/144.pdstates", "pdema": "144/144.pdema", "inference_config": "144/inference/inference.yml", "pdmodel": "144/inference/inference.json", "pdiparams": "144/inference/inference.pdiparams"}, "best": {"score": 0.0, "pdparams": "best_model/best_model.pdparams", "pdopt": "best_model/best_model.pdopt", "pdstates": "best_model/best_model.pdstates", "pdema": "best_model/best_model.pdema", "inference_config": "best_model/inference/inference.yml", "pdmodel": "best_model/inference/inference.json", "pdiparams": "best_model/inference/inference.pdiparams"}}, "done_flag": false}