first commit

2025-07-02 08:57:16 +03:00
commit 56532cc9a9
1901 changed files with 457695 additions and 0 deletions
--- a/deploy/slim/auto_compression/configs/ppocrv4/ppocrv4_det_qat_dist.yaml
+++ b/deploy/slim/auto_compression/configs/ppocrv4/ppocrv4_det_qat_dist.yaml
@@ -0,0 +1,163 @@
+Global:
+  model_type: det
+  model_dir: ./models/ch_PP-OCRv4_det_infer
+  model_filename: inference.pdmodel
+  params_filename: inference.pdiparams
+  algorithm: DB
+  
+Distillation:
+  alpha: 1.0
+  loss: l2
+
+QuantAware:
+  use_pact: false
+  activation_bits: 8
+  is_full_quantize: false
+  onnx_format: false
+  activation_quantize_type: moving_average_abs_max
+  weight_quantize_type: channel_wise_abs_max
+  not_quant_pattern:
+  - skip_quant
+  quantize_op_types:
+  - conv2d
+  weight_bits: 8
+
+TrainConfig:
+  epochs: 2
+  eval_iter: 200
+  learning_rate: 
+    type: CosineAnnealingDecay 
+    learning_rate: 0.000005
+  optimizer_builder:
+    optimizer:
+      type: Adam
+    weight_decay: 5.0e-05
+
+PostProcess:
+  name: DBPostProcess
+  thresh: 0.3
+  box_thresh: 0.6
+  max_candidates: 1000
+  unclip_ratio: 1.5
+  
+Metric:
+  name: DetMetric
+  main_indicator: hmean
+  
+Train:
+  dataset:
+    name: SimpleDataSet
+    data_dir: datasets/chinese
+    label_file_list:
+        - datasets/chinese/zhongce_training_fix_1.6k.txt
+        - datasets/chinese/label_train_all_f4_part2.txt
+        - datasets/chinese/label_train_all_f4_part3.txt
+        - datasets/chinese/label_train_all_f4_part4.txt
+        - datasets/chinese/label_train_all_f4_part5.txt
+        - datasets/chinese/synth_en_my_clip.txt
+        - datasets/chinese/synth_ch_my_clip.txt
+        - datasets/chinese/synth_en_my_largeword_clip.txt
+    ratio_list:
+        - 0.3
+        - 0.2
+        - 0.1
+        - 0.2
+        - 0.2
+        - 0.1
+        - 0.2
+        - 0.2
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - DetLabelEncode: null
+    - IaaAugment:
+        augmenter_args:
+        - type: Fliplr
+          args:
+            p: 0.5
+        - type: Affine
+          args:
+            rotate:
+            - -10
+            - 10
+        - type: Resize
+          args:
+            size:
+            - 0.5
+            - 3
+    - EastRandomCropData:
+        size:
+        - 960
+        - 960
+        max_tries: 50
+        keep_ratio: true
+    - MakeBorderMap:
+        shrink_ratio: 0.4
+        thresh_min: 0.3
+        thresh_max: 0.7
+    - MakeShrinkMap:
+        shrink_ratio: 0.4
+        min_text_size: 8
+    - NormalizeImage:
+        scale: 1./255.
+        mean:
+        - 0.485
+        - 0.456
+        - 0.406
+        std:
+        - 0.229
+        - 0.224
+        - 0.225
+        order: hwc
+    - ToCHWImage: null
+    - KeepKeys:
+        keep_keys:
+        - image
+        - threshold_map
+        - threshold_mask
+        - shrink_map
+        - shrink_mask
+  loader:
+    shuffle: true
+    drop_last: false
+    batch_size_per_card: 4
+    num_workers: 4
+    
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: datasets/v4_4_test_dataset
+    label_file_list:
+      - datasets/v4_4_test_dataset/label.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - DetLabelEncode: null
+    - DetResizeForTest:
+        limit_side_len: 960
+        limit_type: max
+    - NormalizeImage:
+        scale: 1./255.
+        mean:
+        - 0.485
+        - 0.456
+        - 0.406
+        std:
+        - 0.229
+        - 0.224
+        - 0.225
+        order: hwc
+    - ToCHWImage: null
+    - KeepKeys:
+        keep_keys:
+        - image
+        - shape
+        - polys
+        - ignore_tags
+  loader:
+    shuffle: false
+    drop_last: false
+    batch_size_per_card: 1
+    num_workers: 10
--- a/deploy/slim/auto_compression/configs/ppocrv4/ppocrv4_det_server_qat_dist.yaml
+++ b/deploy/slim/auto_compression/configs/ppocrv4/ppocrv4_det_server_qat_dist.yaml
@@ -0,0 +1,161 @@
+Global:
+  model_type: det
+  model_dir: ./models/ch_PP-OCRv4_det_server_infer
+  model_filename: inference.pdmodel
+  params_filename: inference.pdiparams
+  algorithm: DB
+  
+Distillation:
+  alpha: 1.0
+  loss: l2
+
+QuantAware:
+  use_pact: false
+  activation_bits: 8
+  is_full_quantize: false
+  onnx_format: false
+  activation_quantize_type: moving_average_abs_max
+  weight_quantize_type: channel_wise_abs_max
+  not_quant_pattern:
+  - skip_quant
+  quantize_op_types:
+  - conv2d
+  weight_bits: 8
+
+TrainConfig:
+  epochs: 1
+  eval_iter: 200
+  learning_rate: 
+    type: CosineAnnealingDecay 
+    learning_rate: 0.000005
+  optimizer_builder:
+    optimizer:
+      type: Adam
+    weight_decay: 5.0e-05
+
+PostProcess:
+  name: DBPostProcess
+  thresh: 0.3
+  box_thresh: 0.6
+  max_candidates: 1000
+  unclip_ratio: 1.5
+  
+Metric:
+  name: DetMetric
+  main_indicator: hmean
+  
+Train:
+  dataset:
+    name: SimpleDataSet
+    data_dir: datasets/chinese
+    label_file_list:
+        - datasets/chinese/zhongce_training_fix_1.6k.txt
+        - datasets/chinese/label_train_all_f4_part2.txt
+        - datasets/chinese/label_train_all_f4_part3.txt
+        - datasets/chinese/label_train_all_f4_part4.txt
+        - datasets/chinese/label_train_all_f4_part5.txt
+        - datasets/chinese/synth_en_my_clip.txt
+        - datasets/chinese/synth_ch_my_clip.txt
+        - datasets/chinese/synth_en_my_largeword_clip.txt
+    ratio_list:
+        - 0.3
+        - 0.2
+        - 0.1
+        - 0.2
+        - 0.2
+        - 0.1
+        - 0.2
+        - 0.2
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - DetLabelEncode: null
+    - IaaAugment:
+        augmenter_args:
+        - type: Fliplr
+          args:
+            p: 0.5
+        - type: Affine
+          args:
+            rotate:
+            - -10
+            - 10
+        - type: Resize
+          args:
+            size:
+            - 0.5
+            - 3
+    - EastRandomCropData:
+        size:
+        - 960
+        - 960
+        max_tries: 50
+        keep_ratio: true
+    - MakeBorderMap:
+        shrink_ratio: 0.4
+        thresh_min: 0.3
+        thresh_max: 0.7
+    - MakeShrinkMap:
+        shrink_ratio: 0.4
+        min_text_size: 8
+    - NormalizeImage:
+        scale: 1./255.
+        mean:
+        - 0.485
+        - 0.456
+        - 0.406
+        std:
+        - 0.229
+        - 0.224
+        - 0.225
+        order: hwc
+    - ToCHWImage: null
+    - KeepKeys:
+        keep_keys:
+        - image
+        - threshold_map
+        - threshold_mask
+        - shrink_map
+        - shrink_mask
+  loader:
+    shuffle: true
+    drop_last: false
+    batch_size_per_card: 2
+    num_workers: 8
+    
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: datasets/v4_4_test_dataset_small
+    label_file_list:
+      - datasets/v4_4_test_dataset_small/label.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - DetLabelEncode: null
+    - DetResizeForTest: null
+    - NormalizeImage:
+        scale: 1./255.
+        mean:
+        - 0.485
+        - 0.456
+        - 0.406
+        std:
+        - 0.229
+        - 0.224
+        - 0.225
+        order: hwc
+    - ToCHWImage: null
+    - KeepKeys:
+        keep_keys:
+        - image
+        - shape
+        - polys
+        - ignore_tags
+  loader:
+    shuffle: false
+    drop_last: false
+    batch_size_per_card: 1
+    num_workers: 2
--- a/deploy/slim/auto_compression/configs/ppocrv4/ppocrv4_rec_qat_dist.yaml
+++ b/deploy/slim/auto_compression/configs/ppocrv4/ppocrv4_rec_qat_dist.yaml
@@ -0,0 +1,115 @@
+Global:
+  model_dir: ./models/ch_PP-OCRv4_rec_infer
+  model_filename: inference.pdmodel
+  params_filename: inference.pdiparams
+  model_type: rec
+  algorithm: SVTR
+  character_dict_path: ./ppocr_keys_v1.txt
+  max_text_length: &max_text_length 25
+  use_space_char: true
+
+Distillation:
+  alpha: [1.0, 1.0]
+  loss: ['skd', 'l2']
+  node:
+   - ['softmax_11.tmp_0']
+   - ['linear_170.tmp_1']
+
+QuantAware:
+  use_pact: false
+  activation_bits: 8
+  is_full_quantize: false
+  onnx_format: false
+  activation_quantize_type: moving_average_abs_max
+  weight_quantize_type: channel_wise_abs_max
+  not_quant_pattern:
+  - skip_quant
+  quantize_op_types:
+  - conv2d
+  weight_bits: 8
+
+TrainConfig:
+  epochs: 1
+  eval_iter: 1000
+  logging_iter: 100
+  learning_rate: 
+    type: CosineAnnealingDecay 
+    learning_rate: 0.00001
+  optimizer_builder:
+    optimizer:
+      type: Adam
+    weight_decay: 5.0e-05
+
+PostProcess:
+  name: CTCLabelDecode 
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  ignore_space: False 
+
+Train:
+  dataset:
+    name: MultiScaleDataSet
+    ds_width: false
+    data_dir: datasets/real_data/
+    label_file_list:
+      - datasets/real_data/train_list.txt
+    ext_op_transform_idx: 1
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - RecConAug:
+        prob: 0.5
+        ext_data_num: 2
+        image_shape: [48, 320, 3]
+        max_text_length: *max_text_length
+    - RecAug:
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  sampler:
+    name: MultiScaleSampler
+    scales: [[320, 32], [320, 48], [320, 64]]
+    first_bs: &bs 64
+    fix_bs: false
+    divided_factor: [8, 16] # w, h
+    is_training: True
+  loader:
+    shuffle: true
+    batch_size_per_card: *bs
+    drop_last: true
+    num_workers: 8
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: datasets/real_data/
+    label_file_list:
+      - datasets/real_data/val_list.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - RecResizeImg:
+        image_shape: [3, 48, 320]
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  loader:
+    shuffle: false
+    drop_last: false
+    batch_size_per_card: 1
+    num_workers: 4
--- a/deploy/slim/auto_compression/configs/ppocrv4/ppocrv4_rec_server_qat_dist.yaml
+++ b/deploy/slim/auto_compression/configs/ppocrv4/ppocrv4_rec_server_qat_dist.yaml
@@ -0,0 +1,113 @@
+Global:
+  model_dir: ./models/ch_PP-OCRv4_rec_server_infer
+  model_filename: inference.pdmodel
+  params_filename: inference.pdiparams
+  model_type: rec
+  algorithm: SVTR
+  character_dict_path: ./ppocr_keys_v1.txt
+  max_text_length: &max_text_length 25
+  use_space_char: true
+
+Distillation:
+  alpha: 1.0
+  loss: 'l2'
+
+QuantAware:
+  use_pact: false
+  activation_bits: 8
+  is_full_quantize: false
+  onnx_format: false
+  activation_quantize_type: moving_average_abs_max
+  weight_quantize_type: channel_wise_abs_max
+  not_quant_pattern:
+  - skip_quant
+  quantize_op_types:
+  - conv2d
+  weight_bits: 8
+
+TrainConfig:
+  epochs: 1
+  eval_iter: 1000
+  logging_iter: 100
+  learning_rate: 
+    type: CosineAnnealingDecay 
+    learning_rate: 0.00001
+  optimizer_builder:
+    optimizer:
+      type: Adam
+    weight_decay: 5.0e-05
+
+PostProcess:
+  name: CTCLabelDecode 
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  ignore_space: False 
+
+Train:
+  dataset:
+    name: MultiScaleDataSet
+    ds_width: false
+    data_dir: datasets/real_data/
+    ext_op_transform_idx: 1
+    label_file_list:
+    - datasets/real_data/train_list.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - RecConAug:
+        prob: 0.5
+        ext_data_num: 2
+        image_shape: [48, 320, 3]
+        max_text_length: *max_text_length
+    - RecAug:
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  sampler:
+    name: MultiScaleSampler
+    scales: [[320, 32], [320, 48], [320, 64]]
+    first_bs: &bs 64
+    fix_bs: false
+    divided_factor: [8, 16] # w, h
+    is_training: True
+  loader:
+    shuffle: true
+    batch_size_per_card: *bs
+    drop_last: true
+    num_workers: 8
+
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: datasets/real_data/
+    label_file_list:
+    - datasets/real_data/val_list.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - RecResizeImg:
+        image_shape: [3, 48, 320]
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  loader:
+    shuffle: false
+    drop_last: false
+    batch_size_per_card: 1
+    num_workers: 4