diff --git a/configs/cls/ch_PP-OCRv3/ch_PP-OCRv3_rotnet.yml b/configs/cls/ch_PP-OCRv3/ch_PP-OCRv3_rotnet.yml new file mode 100644 index 0000000000000000000000000000000000000000..0e0b15e46c51546f63e400a1858ad9be15dfd982 --- /dev/null +++ b/configs/cls/ch_PP-OCRv3/ch_PP-OCRv3_rotnet.yml @@ -0,0 +1,98 @@ +Global: + debug: false + use_gpu: true + epoch_num: 100 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_ppocr_v3_rotnet + save_epoch_step: 3 + eval_batch_step: [0, 2000] + cal_metric_during_train: true + pretrained_model: null + checkpoints: null + save_inference_dir: null + use_visualdl: false + infer_img: doc/imgs_words/ch/word_1.jpg + character_dict_path: ppocr/utils/ppocr_keys_v1.txt + max_text_length: 25 + infer_mode: false + use_space_char: true + save_res_path: ./output/rec/predicts_chinese_lite_v2.0.txt +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: L2 + factor: 1.0e-05 +Architecture: + model_type: cls + algorithm: CLS + Transform: null + Backbone: + name: MobileNetV1Enhance + scale: 0.5 + last_conv_stride: [1, 2] + last_pool_type: avg + Neck: + Head: + name: ClsHead + class_dim: 4 + +Loss: + name: ClsLoss + main_indicator: acc + +PostProcess: + name: ClsPostProcess + +Metric: + name: ClsMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: + - ./train_data/train_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - BaseDataAugmentation: + - RandAugment: + - SSLRotateResize: + image_shape: [3, 48, 320] + - KeepKeys: + keep_keys: ["image", "label"] + loader: + collate_fn: "SSLRotateCollate" + shuffle: true + batch_size_per_card: 32 + drop_last: true + num_workers: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: + - ./train_data/val_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - SSLRotateResize: + image_shape: [3, 48, 320] + - KeepKeys: + keep_keys: ["image", "label"] + loader: + collate_fn: "SSLRotateCollate" + shuffle: false + drop_last: false + batch_size_per_card: 64 + num_workers: 8 +profiler_options: null diff --git a/configs/cls/cls_mv3.yml b/configs/cls/cls_mv3.yml new file mode 100644 index 0000000000000000000000000000000000000000..2a3c250396da7dd76864b71c96d13061bd2cade9 --- /dev/null +++ b/configs/cls/cls_mv3.yml @@ -0,0 +1,95 @@ +Global: + device: gpu + epoch_num: 100 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/cls/mv3/ + eval_epoch_step: [0, 1] + cal_metric_during_train: True + pretrained_model: + checkpoints: + use_tensorboard: False + infer_img: img2 + label_list: &label_list ['0','180'] + +Export: + export_dir: + export_shape: [ 1, 3, 48, 192 ] + dynamic_axes: [0] + +Architecture: + model_type: cls + algorithm: CLS + Transform: + Backbone: + name: MobileNetV3 + scale: 0.35 + model_name: small + Neck: + Head: + name: ClsHead + class_dim: 2 + +Loss: + name: ClsLoss + +Optimizer: + name: Adam + lr: 0.001 + weight_decay: 0 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 2 + +PostProcess: + name: ClsPostProcess + label_list: *label_list + +Metric: + name: ClsMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/cls + label_file_list: + - ./train_data/cls/train.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - ClsLabelEncode: # Class handling label + - BaseDataAugmentation: + - RandAugment: + - ClsResizeImg: + image_shape: [3, 48, 192] + - KeepKeys: + keep_keys: ['image', 'label'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 512 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/cls + label_file_list: + - ./train_data/cls/test.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - ClsLabelEncode: # Class handling label + - ClsResizeImg: + image_shape: [3, 48, 192] + - KeepKeys: + keep_keys: ['image', 'label'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 512 + num_workers: 4 diff --git a/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_cml.yml b/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_cml.yml new file mode 100644 index 0000000000000000000000000000000000000000..51b6b3efe39b787fd0d9bc0822be2e98b7a60db0 --- /dev/null +++ b/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_cml.yml @@ -0,0 +1,206 @@ +Global: + use_gpu: true + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/ch_db_mv3/ + save_epoch_step: 1200 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [3000, 2000] + cal_metric_during_train: False + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_db/predicts_db.txt + use_amp: False + amp_level: O2 + amp_dtype: bfloat16 + +Architecture: + name: DistillationModel + algorithm: Distillation + model_type: det + Models: + Teacher: + pretrained: ./pretrain_models/ch_ppocr_server_v2.0_det_train/best_accuracy + freeze_params: true + return_all_feats: false + model_type: det + algorithm: DB + Transform: + Backbone: + name: ResNet_vd + layers: 18 + Neck: + name: DBFPN + out_channels: 256 + Head: + name: DBHead + k: 50 + Student: + pretrained: + freeze_params: false + return_all_feats: false + model_type: det + algorithm: DB + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: True + Neck: + name: DBFPN + out_channels: 96 + Head: + name: DBHead + k: 50 + Student2: + pretrained: + freeze_params: false + return_all_feats: false + model_type: det + algorithm: DB + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: True + Neck: + name: DBFPN + out_channels: 96 + Head: + name: DBHead + k: 50 + +Loss: + name: CombinedLoss + loss_config_list: + - DistillationDilaDBLoss: + weight: 1.0 + model_name_pairs: + - ["Student", "Teacher"] + - ["Student2", "Teacher"] + key: maps + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + - DistillationDMLLoss: + model_name_pairs: + - ["Student", "Student2"] + maps_name: "thrink_maps" + weight: 1.0 + # act: None + model_name_pairs: ["Student", "Student2"] + key: maps + - DistillationDBLoss: + weight: 1.0 + model_name_list: ["Student", "Student2"] + # key: maps + # name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 2 + regularizer: + name: 'L2' + factor: 0 + +PostProcess: + name: DistillationDBPostProcess + model_name: ["Student", "Student2", "Teacher"] + # key: maps + thresh: 0.3 + box_thresh: 0.6 + max_candidates: 1000 + unclip_ratio: 1.5 + +Metric: + name: DistillationMetric + base_metric_name: DetMetric + main_indicator: hmean + key: "Student" + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - CopyPaste: + - 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_distill.yml b/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_distill.yml new file mode 100644 index 0000000000000000000000000000000000000000..2b865d73cdbfa9b4aeb28d78a8881e5c8f45b37b --- /dev/null +++ b/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_distill.yml @@ -0,0 +1,175 @@ +Global: + use_gpu: true + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/ch_db_mv3/ + save_epoch_step: 1200 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [3000, 2000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_db/predicts_db.txt + +Architecture: + name: DistillationModel + algorithm: Distillation + model_type: det + Models: + Student: + pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained + freeze_params: false + return_all_feats: false + model_type: det + algorithm: DB + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: True + Neck: + name: DBFPN + out_channels: 96 + Head: + name: DBHead + k: 50 + Teacher: + pretrained: ./pretrain_models/ch_ppocr_server_v2.0_det_train/best_accuracy + freeze_params: true + return_all_feats: false + model_type: det + algorithm: DB + Transform: + Backbone: + name: ResNet_vd + layers: 18 + Neck: + name: DBFPN + out_channels: 256 + Head: + name: DBHead + k: 50 + +Loss: + name: CombinedLoss + loss_config_list: + - DistillationDilaDBLoss: + weight: 1.0 + model_name_pairs: + - ["Student", "Teacher"] + key: maps + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + - DistillationDBLoss: + weight: 1.0 + model_name_list: ["Student"] + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 2 + regularizer: + name: 'L2' + factor: 0 + +PostProcess: + name: DistillationDBPostProcess + model_name: ["Student"] + key: head_out + thresh: 0.3 + box_thresh: 0.6 + max_candidates: 1000 + unclip_ratio: 1.5 + +Metric: + name: DistillationMetric + base_metric_name: DetMetric + main_indicator: hmean + key: "Student" + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - CopyPaste: + - 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: +# image_shape: [736, 1280] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_dml.yml b/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_dml.yml new file mode 100644 index 0000000000000000000000000000000000000000..9260151f7e7a15f5eda646ba6f82abc184d18d6f --- /dev/null +++ b/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_dml.yml @@ -0,0 +1,178 @@ +Global: + use_gpu: true + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/ch_db_mv3/ + save_epoch_step: 1200 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [3000, 2000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_db/predicts_db.txt + +Architecture: + name: DistillationModel + algorithm: Distillation + model_type: det + Models: + Student: + pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained + freeze_params: false + return_all_feats: false + model_type: det + algorithm: DB + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: True + Neck: + name: DBFPN + out_channels: 96 + Head: + name: DBHead + k: 50 + Teacher: + pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained + freeze_params: false + return_all_feats: false + model_type: det + algorithm: DB + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: True + Neck: + name: DBFPN + out_channels: 96 + Head: + name: DBHead + k: 50 + + +Loss: + name: CombinedLoss + loss_config_list: + - DistillationDMLLoss: + model_name_pairs: + - ["Student", "Teacher"] + maps_name: "thrink_maps" + weight: 1.0 + # act: None + model_name_pairs: ["Student", "Teacher"] + key: maps + - DistillationDBLoss: + weight: 1.0 + model_name_list: ["Student", "Teacher"] + # key: maps + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 2 + regularizer: + name: 'L2' + factor: 0 + +PostProcess: + name: DistillationDBPostProcess + model_name: ["Student", "Teacher"] + key: head_out + thresh: 0.3 + box_thresh: 0.6 + max_candidates: 1000 + unclip_ratio: 1.5 + +Metric: + name: DistillationMetric + base_metric_name: DetMetric + main_indicator: hmean + key: "Student" + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - CopyPaste: + - 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: +# image_shape: [736, 1280] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_student.yml b/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_student.yml new file mode 100644 index 0000000000000000000000000000000000000000..ba2f053010a13a9eb4d2d8368354aa3450f5dac4 --- /dev/null +++ b/configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_student.yml @@ -0,0 +1,132 @@ +Global: + use_gpu: true + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/ch_db_mv3/ + save_epoch_step: 1200 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 400] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/student.pdparams + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_db/predicts_db.txt + +Architecture: + model_type: det + algorithm: DB + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: True + Neck: + name: DBFPN + out_channels: 96 + Head: + name: DBHead + k: 50 + +Loss: + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 2 + regularizer: + name: 'L2' + factor: 0 + +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: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: +# image_shape: [736, 1280] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml b/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml new file mode 100644 index 0000000000000000000000000000000000000000..0f4da24cfbc1817dfd2b3d31e3f49878dc0bb9ad --- /dev/null +++ b/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml @@ -0,0 +1,217 @@ +Global: + device: gpu + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det/ch_PP-OCR_V3_det_cml/ + eval_epoch_step: [0, 1] + cal_metric_during_train: false + pretrained_model: + checkpoints: + use_tensorboard: false + infer_img: doc/imgs/1.jpg + +Export: + export_dir: + export_shape: [ 1, 3, 640, 640 ] + dynamic_axes: [ 0, 2, 3 ] + +Architecture: + name: DistillationModel + algorithm: Distillation + model_type: det + Models: + Student: + pretrained: + model_type: det + algorithm: DB + Transform: null + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: true + Neck: + name: RSEFPN + out_channels: 96 + shortcut: True + Head: + name: DBHead + k: 50 + Student2: + pretrained: + model_type: det + algorithm: DB + Transform: null + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: true + Neck: + name: RSEFPN + out_channels: 96 + shortcut: True + Head: + name: DBHead + k: 50 + Teacher: + freeze_params: true + return_all_feats: false + model_type: det + algorithm: DB + Backbone: + name: ResNet_vd + in_channels: 3 + layers: 50 + Neck: + name: LKPAN + out_channels: 256 + Head: + name: DBHead + kernel_list: [7,2,2] + k: 50 + +Loss: + name: CombinedLoss + loss_config_list: + - DistillationDilaDBLoss: + weight: 1.0 + model_name_pairs: + - ["Student", "Teacher"] + - ["Student2", "Teacher"] + key: res + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + - DistillationDMLLoss: + maps_name: "thrink_maps" + weight: 1.0 + model_name_pairs: ["Student", "Student2"] + key: res + - DistillationDBLoss: + weight: 1.0 + model_name_list: ["Student", "Student2"] + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + lr: 0.001 + weight_decay: 5.0e-05 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 2 + +PostProcess: + name: DistillationDBPostProcess + model_name: ["Student"] + key: head_out + thresh: 0.3 + box_thresh: 0.6 + max_candidates: 1000 + unclip_ratio: 1.5 + +Metric: + name: DistillationMetric + base_metric_name: DetMetric + main_indicator: hmean + key: "Student" + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - DetLabelEncode: null + - CopyPaste: + - 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: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_dml.yml b/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_dml.yml new file mode 100644 index 0000000000000000000000000000000000000000..6110f7c8d8093539bd512bbfd902b88c5f7ff2be --- /dev/null +++ b/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_dml.yml @@ -0,0 +1,166 @@ +Global: + device: gpu + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det/ch_PP-OCR_V3_det_dml/ + eval_epoch_step: [0, 1] + cal_metric_during_train: false + pretrained_model: + checkpoints: + use_tensorboard: false + infer_img: doc/imgs/1.jpg + +Export: + export_dir: + export_shape: [ 1, 3, 640, 640 ] + dynamic_axes: [ 0, 2, 3 ] + +Architecture: + name: DistillationModel + algorithm: Distillation + model_type: det + Models: + Student: + return_all_feats: false + model_type: det + algorithm: DB + Backbone: + name: ResNet_vd + in_channels: 3 + layers: 50 + Neck: + name: LKPAN + out_channels: 256 + Head: + name: DBHead + kernel_list: [7,2,2] + k: 50 + Student2: + return_all_feats: false + model_type: det + algorithm: DB + Backbone: + name: ResNet_vd + in_channels: 3 + layers: 50 + Neck: + name: LKPAN + out_channels: 256 + Head: + name: DBHead + kernel_list: [7,2,2] + k: 50 + + +Loss: + name: CombinedLoss + loss_config_list: + - DistillationDMLLoss: + maps_name: "thrink_maps" + weight: 1.0 + model_name_pairs: ["Student", "Student2"] + key: res + - DistillationDBLoss: + weight: 1.0 + model_name_list: ["Student", "Student2"] + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + lr: 0.001 + weight_decay: 0 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 2 + +PostProcess: + name: DistillationDBPostProcess + model_name: ["Student", "Student2"] + key: head_out + thresh: 0.3 + box_thresh: 0.6 + max_candidates: 1000 + unclip_ratio: 1.5 + +Metric: + name: DistillationMetric + base_metric_name: DetMetric + main_indicator: hmean + key: "Student" + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - CopyPaste: + - 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: +# image_shape: [736, 1280] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_student.yml b/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_student.yml new file mode 100644 index 0000000000000000000000000000000000000000..f365de2c7b7957a60d71e519a0402d6eb8b342ec --- /dev/null +++ b/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_student.yml @@ -0,0 +1,161 @@ +Global: + device: gpu + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det/ch_PP-OCR_V3_det/ + eval_epoch_step: [0, 1] + cal_metric_during_train: false + pretrained_model: + checkpoints: + use_tensorboard: false + infer_img: doc/imgs/1.jpg + +Export: + export_dir: + export_shape: [ 1, 3, 640, 640 ] + dynamic_axes: [ 0, 2, 3 ] + +Architecture: + model_type: det + algorithm: DB + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: True + Neck: + name: RSEFPN + out_channels: 96 + shortcut: True + Head: + name: DBHead + k: 50 + +Loss: + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + lr: 0.001 + weight_decay: 5.0e-05 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 2 + +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: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + 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: 8 + num_workers: 4 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_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 diff --git a/configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_cml.yml b/configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_cml.yml new file mode 100644 index 0000000000000000000000000000000000000000..71c089faaaf06e0832e026a6da6447da122dbfaf --- /dev/null +++ b/configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_cml.yml @@ -0,0 +1,234 @@ +Global: + device: gpu + epoch_num: &epoch_num 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det/ch_PP-OCRv4_cml + eval_epoch_step: [0, 1] + cal_metric_during_train: false + pretrained_model: + checkpoints: + use_tensorboard: false + infer_img: img + +Export: + export_dir: + export_shape: [ 1, 3, 640, 640 ] + dynamic_axes: [ 0, 2, 3 ] + +Architecture: + name: DistillationModel + algorithm: Distillation + model_type: det + Models: + Student: + pretrained: + model_type: det + algorithm: DB + Transform: null + Backbone: + name: PPLCNetV3 + scale: 0.75 + det: True + Neck: + name: RSEFPN + out_channels: 96 + shortcut: True + Head: + name: DBHead + k: 50 + Student2: + pretrained: + model_type: det + algorithm: DB + Transform: null + Backbone: + name: PPLCNetV3 + scale: 0.75 + det: True + Neck: + name: RSEFPN + out_channels: 96 + shortcut: True + Head: + name: DBHead + k: 50 + Teacher: + freeze_params: true + return_all_feats: false + pretrained: + model_type: det + algorithm: DB + Transform: null + Backbone: + name: PPHGNet_small + det: True + Neck: + name: LKPAN + out_channels: 256 + intracl: true + Head: + name: PFHeadLocal + k: 50 + mode: "large" + +Loss: + name: CombinedLoss + loss_config_list: + - DistillationDilaDBLoss: + weight: 1.0 + model_name_pairs: + - ["Student", "Teacher"] + - ["Student2", "Teacher"] + key: res + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + kd_loss: + name: KnowledgeDistillationKLDivLoss + loss_weight: 1.0e8 + T: 10 + - DistillationDMLLoss: + model_name_pairs: + - ["Student", "Student2"] + maps_name: "thrink_maps" + weight: 1.0 + key: res + - DistillationDBLoss: + weight: 1.0 + model_name_list: ["Student", "Student2"] + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + lr: 0.001 #(8*8c) + weight_decay: 1.0e-6 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 2 + +PostProcess: + name: DistillationDBPostProcess + model_name: ["Student"] + key: head_out + thresh: 0.3 + box_thresh: 0.6 + max_candidates: 1000 + unclip_ratio: 1.5 + +Metric: + name: DistillationMetric + base_metric_name: DetMetric + main_indicator: hmean + key: "Student" + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - DetLabelEncode: null + - CopyPaste: 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: + - 640 + - 640 + max_tries: 50 + keep_ratio: true + - MakeBorderMap: + shrink_ratio: 0.4 + thresh_min: 0.3 + thresh_max: 0.7 + total_epoch: *epoch_num + - MakeShrinkMap: + shrink_ratio: 0.4 + min_text_size: 8 + total_epoch: *epoch_num + - 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: 8 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - DetLabelEncode: null + - DetResizeForTest: + - 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 diff --git a/configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_student.yml b/configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_student.yml new file mode 100644 index 0000000000000000000000000000000000000000..ee833009fe3a57b8d0ef5b592652992993e2100c --- /dev/null +++ b/configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_student.yml @@ -0,0 +1,167 @@ +Global: + device: gpu + epoch_num: &epoch_num 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det/ch_PP-OCRv4/ + eval_epoch_step: [0, 1] + cal_metric_during_train: false + pretrained_model: + checkpoints: + use_tensorboard: false + infer_img: img + +Export: + export_dir: + export_shape: [ 1, 3, 640, 640 ] + dynamic_axes: [ 0, 2, 3 ] + +Architecture: + model_type: det + algorithm: DB + Transform: null + Backbone: + name: PPLCNetV3 + scale: 0.75 + det: True + Neck: + name: RSEFPN + out_channels: 96 + shortcut: True + Head: + name: DBHead + k: 50 + +Loss: + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + lr: 0.001 #(8*8c) + weight_decay: 5.0e-05 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 2 + + +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: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - DetLabelEncode: null + - CopyPaste: 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: + - 640 + - 640 + max_tries: 50 + keep_ratio: true + - MakeBorderMap: + shrink_ratio: 0.4 + thresh_min: 0.3 + thresh_max: 0.7 + total_epoch: *epoch_num + - MakeShrinkMap: + shrink_ratio: 0.4 + min_text_size: 8 + total_epoch: *epoch_num + - 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: 8 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - DetLabelEncode: null + - DetResizeForTest: + - 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 +profiler_options: null diff --git a/configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_teacher.yml b/configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_teacher.yml new file mode 100644 index 0000000000000000000000000000000000000000..43dde6b29243e370814a28482e8511069b22302d --- /dev/null +++ b/configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_teacher.yml @@ -0,0 +1,167 @@ +Global: + device: gpu + epoch_num: &epoch_num 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det/h_PP-OCRv4_hgnet/ + eval_epoch_step: [0, 1] + cal_metric_during_train: false + pretrained_model: + checkpoints: + use_tensorboard: false + infer_img: img + +Export: + export_dir: + export_shape: [ 1, 3, 640, 640 ] + dynamic_axes: [ 0, 2, 3 ] + +Architecture: + model_type: det + algorithm: DB + Transform: null + Backbone: + name: PPHGNet_small + det: True + Neck: + name: LKPAN + out_channels: 256 + intracl: true + Head: + name: PFHeadLocal + k: 50 + mode: "large" + + +Loss: + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + lr: 0.001 #(8*8c) + weight_decay: 1.0e-6 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 2 + +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: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - DetLabelEncode: null + - CopyPaste: 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: + - 640 + - 640 + max_tries: 50 + keep_ratio: true + - MakeBorderMap: + shrink_ratio: 0.4 + thresh_min: 0.3 + thresh_max: 0.7 + total_epoch: *epoch_num + - MakeShrinkMap: + shrink_ratio: 0.4 + min_text_size: 8 + total_epoch: *epoch_num + - 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: 8 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - DetLabelEncode: null + - DetResizeForTest: + - 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 +profiler_options: null diff --git a/configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml b/configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml new file mode 100644 index 0000000000000000000000000000000000000000..7c00612c65b27fab24a214bde2e8522e76b04cd9 --- /dev/null +++ b/configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml @@ -0,0 +1,132 @@ +Global: + use_gpu: true + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/ch_db_mv3/ + save_epoch_step: 1200 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [3000, 2000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_db/predicts_db.txt + +Architecture: + model_type: det + algorithm: DB + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + disable_se: True + Neck: + name: DBFPN + out_channels: 96 + Head: + name: DBHead + k: 50 + +Loss: + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 2 + regularizer: + name: 'L2' + factor: 0 + +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: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: +# image_shape: [736, 1280] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml b/configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml new file mode 100644 index 0000000000000000000000000000000000000000..00a1d97b6232a156ff083101d6f50d9f35b3b8f0 --- /dev/null +++ b/configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml @@ -0,0 +1,131 @@ +Global: + use_gpu: true + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/ch_db_res18/ + save_epoch_step: 1200 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [3000, 2000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet18_vd_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_db/predicts_db.txt + +Architecture: + model_type: det + algorithm: DB + Transform: + Backbone: + name: ResNet_vd + layers: 18 + disable_se: True + Neck: + name: DBFPN + out_channels: 256 + Head: + name: DBHead + k: 50 + +Loss: + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 2 + regularizer: + name: 'L2' + factor: 0 + +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: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: +# image_shape: [736, 1280] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/det/det_mv3_db.yml b/configs/det/det_mv3_db.yml new file mode 100644 index 0000000000000000000000000000000000000000..d84305423766953f1a21bc776974a4154ebc0d39 --- /dev/null +++ b/configs/det/det_mv3_db.yml @@ -0,0 +1,131 @@ +Global: + use_gpu: true + use_xpu: false + use_mlu: false + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/db_mv3/ + save_epoch_step: 1200 + # evaluation is run every 2000 iterations + eval_batch_step: [0, 2000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_db/predicts_db.txt + +Architecture: + model_type: det + algorithm: DB + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + Neck: + name: DBFPN + out_channels: 256 + Head: + name: DBHead + k: 50 + +Loss: + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0 + +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: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - IaaAugment: + augmenter_args: + - { 'type': Fliplr, 'args': { 'p': 0.5 } } + - { 'type': Affine, 'args': { 'rotate': [-10, 10] } } + - { 'type': Resize, 'args': { 'size': [0.5, 3] } } + - EastRandomCropData: + size: [640, 640] + 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 16 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + image_shape: [736, 1280] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 8 diff --git a/configs/det/det_mv3_east.yml b/configs/det/det_mv3_east.yml new file mode 100644 index 0000000000000000000000000000000000000000..e6fa397e73c5a2c3b53ee02a4e884eb1238713a1 --- /dev/null +++ b/configs/det/det_mv3_east.yml @@ -0,0 +1,109 @@ +Global: + use_gpu: true + epoch_num: 10000 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/east_mv3/ + save_epoch_step: 1000 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [4000, 5000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + save_res_path: ./output/det_east/predicts_east.txt + +Architecture: + model_type: det + algorithm: EAST + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + Neck: + name: EASTFPN + model_name: small + Head: + name: EASTHead + model_name: small + +Loss: + name: EASTLoss + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + # name: Cosine + learning_rate: 0.001 + # warmup_epoch: 0 + regularizer: + name: 'L2' + factor: 0 + +PostProcess: + name: EASTPostProcess + score_thresh: 0.8 + cover_thresh: 0.1 + nms_thresh: 0.2 + +Metric: + name: DetMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - EASTProcessTrain: + image_shape: [512, 512] + background_ratio: 0.125 + min_crop_side_ratio: 0.1 + min_text_size: 10 + - KeepKeys: + keep_keys: ['image', 'score_map', 'geo_map', 'training_mask'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 16 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + limit_side_len: 2400 + limit_type: max + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 \ No newline at end of file diff --git a/configs/det/det_mv3_pse.yml b/configs/det/det_mv3_pse.yml new file mode 100644 index 0000000000000000000000000000000000000000..91712f69c2d6eae4e927944c8f55509316cf4f3a --- /dev/null +++ b/configs/det/det_mv3_pse.yml @@ -0,0 +1,135 @@ +Global: + use_gpu: true + epoch_num: 600 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det_mv3_pse/ + save_epoch_step: 600 + # evaluation is run every 63 iterations + eval_batch_step: [ 0,63 ] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained + checkpoints: #./output/det_r50_vd_pse_batch8_ColorJitter/best_accuracy + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_pse/predicts_pse.txt + +Architecture: + model_type: det + algorithm: PSE + Transform: null + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + Neck: + name: FPN + out_channels: 96 + Head: + name: PSEHead + hidden_dim: 96 + out_channels: 7 + +Loss: + name: PSELoss + alpha: 0.7 + ohem_ratio: 3 + kernel_sample_mask: pred + reduction: none + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Step + learning_rate: 0.001 + step_size: 200 + gamma: 0.1 + regularizer: + name: 'L2' + factor: 0.0005 + +PostProcess: + name: PSEPostProcess + thresh: 0 + box_thresh: 0.85 + min_area: 16 + box_type: quad # 'quad' or 'poly' + scale: 1 + +Metric: + name: DetMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - ColorJitter: + brightness: 0.12549019607843137 + saturation: 0.5 + - IaaAugment: + augmenter_args: + - { 'type': Resize, 'args': { 'size': [ 0.5, 3 ] } } + - { 'type': Fliplr, 'args': { 'p': 0.5 } } + - { 'type': Affine, 'args': { 'rotate': [ -10, 10 ] } } + - MakePseGt: + kernel_num: 7 + min_shrink_ratio: 0.4 + size: 640 + - RandomCropImgMask: + size: [ 640,640 ] + main_key: gt_text + crop_keys: [ 'image', 'gt_text', 'gt_kernels', 'mask' ] + - NormalizeImage: + scale: 1./255. + mean: [ 0.485, 0.456, 0.406 ] + std: [ 0.229, 0.224, 0.225 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'gt_text', 'gt_kernels', 'mask' ] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 16 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + limit_side_len: 736 + limit_type: min + - NormalizeImage: + scale: 1./255. + mean: [ 0.485, 0.456, 0.406 ] + std: [ 0.229, 0.224, 0.225 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'shape', 'polys', 'ignore_tags' ] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 8 \ No newline at end of file diff --git a/configs/det/det_r18_vd_ct.yml b/configs/det/det_r18_vd_ct.yml new file mode 100644 index 0000000000000000000000000000000000000000..d7aaa16e173904ea8466b24704d539af182f5a92 --- /dev/null +++ b/configs/det/det_r18_vd_ct.yml @@ -0,0 +1,107 @@ +Global: + use_gpu: true + epoch_num: 600 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det_ct/ + save_epoch_step: 10 + # evaluation is run every 2000 iterations + eval_batch_step: [0,1000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet18_vd_pretrained.pdparams + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img623.jpg + save_res_path: ./output/det_ct/predicts_ct.txt + +Architecture: + model_type: det + algorithm: CT + Transform: + Backbone: + name: ResNet_vd + layers: 18 + Neck: + name: CTFPN + Head: + name: CT_Head + in_channels: 512 + hidden_dim: 128 + num_classes: 3 + +Loss: + name: CTLoss + +Optimizer: + name: Adam + lr: #PolynomialDecay + name: Linear + learning_rate: 0.001 + end_lr: 0. + epochs: 600 + step_each_epoch: 1254 + power: 0.9 + +PostProcess: + name: CTPostProcess + box_type: poly + +Metric: + name: CTMetric + main_indicator: f_score + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/total_text/train + label_file_list: + - ./train_data/total_text/train/train.txt + ratio_list: [1.0] + transforms: + - DecodeImage: + img_mode: RGB + channel_first: False + - CTLabelEncode: # Class handling label + - RandomScale: + - MakeShrink: + - GroupRandomHorizontalFlip: + - GroupRandomRotate: + - GroupRandomCropPadding: + - MakeCentripetalShift: + - ColorJitter: + brightness: 0.125 + saturation: 0.5 + - ToCHWImage: + - NormalizeImage: + - KeepKeys: + keep_keys: ['image', 'gt_kernel', 'training_mask', 'gt_instance', 'gt_kernel_instance', 'training_mask_distance', 'gt_distance'] # the order of the dataloader list + loader: + shuffle: True + drop_last: True + batch_size_per_card: 4 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/total_text/test + label_file_list: + - ./train_data/total_text/test/test.txt + ratio_list: [1.0] + transforms: + - DecodeImage: + img_mode: RGB + channel_first: False + - CTLabelEncode: # Class handling label + - ScaleAlignedShort: + - NormalizeImage: + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'texts'] # the order of the dataloader list + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 + num_workers: 2 diff --git a/configs/det/det_r50_db++_icdar15.yml b/configs/det/det_r50_db++_icdar15.yml new file mode 100644 index 0000000000000000000000000000000000000000..271c61679b8d07743bbaaed33a3632e011f10e81 --- /dev/null +++ b/configs/det/det_r50_db++_icdar15.yml @@ -0,0 +1,164 @@ +Global: + debug: false + use_gpu: true + epoch_num: 1000 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det_r50_icdar15/ + save_epoch_step: 200 + eval_batch_step: + - 0 + - 2000 + cal_metric_during_train: false + pretrained_model: ./pretrain_models/ResNet50_dcn_asf_synthtext_pretrained + checkpoints: null + save_inference_dir: null + use_visualdl: false + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./checkpoints/det_db/predicts_db.txt +Architecture: + model_type: det + algorithm: DB++ + Transform: null + Backbone: + name: ResNet + layers: 50 + dcn_stage: [False, True, True, True] + Neck: + name: DBFPN + out_channels: 256 + use_asf: True + Head: + name: DBHead + k: 50 +Loss: + name: DBLoss + balance_loss: true + main_loss_type: BCELoss + alpha: 5 + beta: 10 + ohem_ratio: 3 +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: DecayLearningRate + learning_rate: 0.007 + epochs: 1000 + factor: 0.9 + end_lr: 0 + weight_decay: 0.0001 +PostProcess: + name: DBPostProcess + thresh: 0.3 + box_thresh: 0.6 + max_candidates: 1000 + unclip_ratio: 1.5 + det_box_type: 'quad' # 'quad' or 'poly' +Metric: + name: DetMetric + main_indicator: hmean +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: + - 1.0 + 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: + - 640 + - 640 + max_tries: 10 + keep_ratio: true + - MakeShrinkMap: + shrink_ratio: 0.4 + min_text_size: 8 + - MakeBorderMap: + shrink_ratio: 0.4 + thresh_min: 0.3 + thresh_max: 0.7 + - NormalizeImage: + scale: 1./255. + mean: + - 0.48109378172549 + - 0.45752457890196 + - 0.40787054090196 + std: + - 1.0 + - 1.0 + - 1.0 + 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: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - DetLabelEncode: null + - DetResizeForTest: + image_shape: + - 1152 + - 2048 + - NormalizeImage: + scale: 1./255. + mean: + - 0.48109378172549 + - 0.45752457890196 + - 0.40787054090196 + std: + - 1.0 + - 1.0 + - 1.0 + 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 +profiler_options: null diff --git a/configs/det/det_r50_db++_td_tr.yml b/configs/det/det_r50_db++_td_tr.yml new file mode 100644 index 0000000000000000000000000000000000000000..1892f005b90c3f3458a412e5d5cc1d33793f949f --- /dev/null +++ b/configs/det/det_r50_db++_td_tr.yml @@ -0,0 +1,167 @@ +Global: + debug: false + use_gpu: true + epoch_num: 1000 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det_r50_td_tr/ + save_epoch_step: 200 + eval_batch_step: + - 0 + - 2000 + cal_metric_during_train: false + pretrained_model: ./pretrain_models/ResNet50_dcn_asf_synthtext_pretrained + checkpoints: null + save_inference_dir: null + use_visualdl: false + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./checkpoints/det_db/predicts_db.txt +Architecture: + model_type: det + algorithm: DB++ + Transform: null + Backbone: + name: ResNet + layers: 50 + dcn_stage: [False, True, True, True] + Neck: + name: DBFPN + out_channels: 256 + use_asf: True + Head: + name: DBHead + k: 50 +Loss: + name: DBLoss + balance_loss: true + main_loss_type: BCELoss + alpha: 5 + beta: 10 + ohem_ratio: 3 +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: DecayLearningRate + learning_rate: 0.007 + epochs: 1000 + factor: 0.9 + end_lr: 0 + weight_decay: 0.0001 +PostProcess: + name: DBPostProcess + thresh: 0.3 + box_thresh: 0.5 + max_candidates: 1000 + unclip_ratio: 1.5 + det_box_type: 'quad' # 'quad' or 'poly' +Metric: + name: DetMetric + main_indicator: hmean +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: + - ./train_data/TD_TR/TD500/train_gt_labels.txt + - ./train_data/TD_TR/TR400/gt_labels.txt + ratio_list: + - 1.0 + - 1.0 + 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: + - 640 + - 640 + max_tries: 10 + keep_ratio: true + - MakeShrinkMap: + shrink_ratio: 0.4 + min_text_size: 8 + - MakeBorderMap: + shrink_ratio: 0.4 + thresh_min: 0.3 + thresh_max: 0.7 + - NormalizeImage: + scale: 1./255. + mean: + - 0.48109378172549 + - 0.45752457890196 + - 0.40787054090196 + std: + - 1.0 + - 1.0 + - 1.0 + 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: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: + - ./train_data/TD_TR/TD500/test_gt_labels.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - DetLabelEncode: null + - DetResizeForTest: + image_shape: + - 736 + - 736 + keep_ratio: True + - NormalizeImage: + scale: 1./255. + mean: + - 0.48109378172549 + - 0.45752457890196 + - 0.40787054090196 + std: + - 1.0 + - 1.0 + - 1.0 + 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 +profiler_options: null diff --git a/configs/det/det_r50_drrg_ctw.yml b/configs/det/det_r50_drrg_ctw.yml new file mode 100644 index 0000000000000000000000000000000000000000..605a789d86a77e57886b9650a83fcbeb404b6949 --- /dev/null +++ b/configs/det/det_r50_drrg_ctw.yml @@ -0,0 +1,133 @@ +Global: + use_gpu: true + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 5 + output_dir: ./output/det_r50_drrg_ctw/ + save_epoch_step: 100 + # evaluation is run every 1260 iterations + eval_batch_step: [37800, 1260] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained.pdparams + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_drrg/predicts_drrg.txt + + +Architecture: + model_type: det + algorithm: DRRG + Transform: + Backbone: + name: ResNet_vd + layers: 50 + Neck: + name: FPN_UNet + in_channels: [256, 512, 1024, 2048] + out_channels: 32 + Head: + name: DRRGHead + in_channels: 32 + text_region_thr: 0.3 + center_region_thr: 0.4 +Loss: + name: DRRGLoss + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: DecayLearningRate + learning_rate: 0.028 + epochs: 1200 + factor: 0.9 + end_lr: 0.0000001 + weight_decay: 0.0001 + +PostProcess: + name: DRRGPostprocess + link_thr: 0.8 + +Metric: + name: DetFCEMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ctw1500/imgs/ + label_file_list: + - ./train_data/ctw1500/imgs/training.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + ignore_orientation: True + - DetLabelEncode: # Class handling label + - ColorJitter: + brightness: 0.12549019607843137 + saturation: 0.5 + - RandomScaling: + - RandomCropFlip: + crop_ratio: 0.5 + - RandomCropPolyInstances: + crop_ratio: 0.8 + min_side_ratio: 0.3 + - RandomRotatePolyInstances: + rotate_ratio: 0.5 + max_angle: 60 + pad_with_fixed_color: False + - SquareResizePad: + target_size: 800 + pad_ratio: 0.6 + - IaaAugment: + augmenter_args: + - { 'type': Fliplr, 'args': { 'p': 0.5 } } + - DRRGTargets: + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'gt_text_mask', 'gt_center_region_mask', 'gt_mask', + 'gt_top_height_map', 'gt_bot_height_map', 'gt_sin_map', + 'gt_cos_map', 'gt_comp_attribs'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 4 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ctw1500/imgs/ + label_file_list: + - ./train_data/ctw1500/imgs/test.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + ignore_orientation: True + - DetLabelEncode: # Class handling label + - DetResizeForTest: + limit_type: 'min' + limit_side_len: 640 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - Pad: + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 \ No newline at end of file diff --git a/configs/det/det_r50_vd_db.yml b/configs/det/det_r50_vd_db.yml new file mode 100644 index 0000000000000000000000000000000000000000..7133d10e8769bcafa45be37fbdb712f1b4353bd5 --- /dev/null +++ b/configs/det/det_r50_vd_db.yml @@ -0,0 +1,128 @@ +Global: + use_gpu: true + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det_r50_vd/ + save_epoch_step: 1200 + # evaluation is run every 2000 iterations + eval_batch_step: [0,2000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_db/predicts_db.txt + +Architecture: + model_type: det + algorithm: DB + Transform: + Backbone: + name: ResNet_vd + layers: 50 + Neck: + name: DBFPN + out_channels: 256 + Head: + name: DBHead + k: 50 + +Loss: + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0 + +PostProcess: + name: DBPostProcess + thresh: 0.3 + box_thresh: 0.7 + max_candidates: 1000 + unclip_ratio: 1.5 + +Metric: + name: DetMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - IaaAugment: + augmenter_args: + - { 'type': Fliplr, 'args': { 'p': 0.5 } } + - { 'type': Affine, 'args': { 'rotate': [-10, 10] } } + - { 'type': Resize, 'args': { 'size': [0.5, 3] } } + - EastRandomCropData: + size: [640, 640] + 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 16 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + image_shape: [736, 1280] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 8 diff --git a/configs/det/det_r50_vd_dcn_fce_ctw.yml b/configs/det/det_r50_vd_dcn_fce_ctw.yml new file mode 100644 index 0000000000000000000000000000000000000000..31db5935ec3cdae69636208f073de80bc70b59b6 --- /dev/null +++ b/configs/det/det_r50_vd_dcn_fce_ctw.yml @@ -0,0 +1,139 @@ +Global: + use_gpu: true + epoch_num: 1500 + log_smooth_window: 20 + print_batch_step: 20 + output_dir: ./output/det_r50_dcn_fce_ctw/ + save_epoch_step: 100 + # evaluation is run every 835 iterations + eval_batch_step: [0, 835] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_fce/predicts_fce.txt + + +Architecture: + model_type: det + algorithm: FCE + Transform: + Backbone: + name: ResNet_vd + layers: 50 + dcn_stage: [False, True, True, True] + out_indices: [1,2,3] + Neck: + name: FCEFPN + out_channels: 256 + has_extra_convs: False + extra_stage: 0 + Head: + name: FCEHead + fourier_degree: 5 +Loss: + name: FCELoss + fourier_degree: 5 + num_sample: 50 + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + learning_rate: 0.0001 + regularizer: + name: 'L2' + factor: 0 + +PostProcess: + name: FCEPostProcess + scales: [8, 16, 32] + alpha: 1.0 + beta: 1.0 + fourier_degree: 5 + box_type: 'poly' + +Metric: + name: DetFCEMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ctw1500/imgs/ + label_file_list: + - ./train_data/ctw1500/imgs/training.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + ignore_orientation: True + - DetLabelEncode: # Class handling label + - ColorJitter: + brightness: 0.142 + saturation: 0.5 + contrast: 0.5 + - RandomScaling: + - RandomCropFlip: + crop_ratio: 0.5 + - RandomCropPolyInstances: + crop_ratio: 0.8 + min_side_ratio: 0.3 + - RandomRotatePolyInstances: + rotate_ratio: 0.5 + max_angle: 30 + pad_with_fixed_color: False + - SquareResizePad: + target_size: 800 + pad_ratio: 0.6 + - IaaAugment: + augmenter_args: + - { 'type': Fliplr, 'args': { 'p': 0.5 } } + - FCENetTargets: + fourier_degree: 5 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'p3_maps', 'p4_maps', 'p5_maps'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 6 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ctw1500/imgs/ + label_file_list: + - ./train_data/ctw1500/imgs/test.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + ignore_orientation: True + - DetLabelEncode: # Class handling label + - DetResizeForTest: + limit_type: 'min' + limit_side_len: 736 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - Pad: + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 \ No newline at end of file diff --git a/configs/det/det_r50_vd_east.yml b/configs/det/det_r50_vd_east.yml new file mode 100644 index 0000000000000000000000000000000000000000..a03576598ba8a92a0763a43d09917757e449a5e1 --- /dev/null +++ b/configs/det/det_r50_vd_east.yml @@ -0,0 +1,108 @@ +Global: + use_gpu: true + epoch_num: 10000 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/east_r50_vd/ + save_epoch_step: 1000 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [4000, 5000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet50_vd_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + save_res_path: ./output/det_east/predicts_east.txt + +Architecture: + model_type: det + algorithm: EAST + Transform: + Backbone: + name: ResNet_vd + layers: 50 + Neck: + name: EASTFPN + model_name: large + Head: + name: EASTHead + model_name: large + +Loss: + name: EASTLoss + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + # name: Cosine + learning_rate: 0.001 + # warmup_epoch: 0 + regularizer: + name: 'L2' + factor: 0 + +PostProcess: + name: EASTPostProcess + score_thresh: 0.8 + cover_thresh: 0.1 + nms_thresh: 0.2 + +Metric: + name: DetMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - EASTProcessTrain: + image_shape: [512, 512] + background_ratio: 0.125 + min_crop_side_ratio: 0.1 + min_text_size: 10 + - KeepKeys: + keep_keys: ['image', 'score_map', 'geo_map', 'training_mask'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + limit_side_len: 2400 + limit_type: max + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 \ No newline at end of file diff --git a/configs/det/det_r50_vd_pse.yml b/configs/det/det_r50_vd_pse.yml new file mode 100644 index 0000000000000000000000000000000000000000..af6c1e6ed4b7e881bf2e8695435d434cd745944e --- /dev/null +++ b/configs/det/det_r50_vd_pse.yml @@ -0,0 +1,134 @@ +Global: + use_gpu: true + epoch_num: 600 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/det_r50_vd_pse/ + save_epoch_step: 600 + # evaluation is run every 125 iterations + eval_batch_step: [ 0,125 ] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained + checkpoints: #./output/det_r50_vd_pse_batch8_ColorJitter/best_accuracy + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_pse/predicts_pse.txt + +Architecture: + model_type: det + algorithm: PSE + Transform: + Backbone: + name: ResNet_vd + layers: 50 + Neck: + name: FPN + out_channels: 256 + Head: + name: PSEHead + hidden_dim: 256 + out_channels: 7 + +Loss: + name: PSELoss + alpha: 0.7 + ohem_ratio: 3 + kernel_sample_mask: pred + reduction: none + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Step + learning_rate: 0.0001 + step_size: 200 + gamma: 0.1 + regularizer: + name: 'L2' + factor: 0.0005 + +PostProcess: + name: PSEPostProcess + thresh: 0 + box_thresh: 0.85 + min_area: 16 + box_type: quad # 'quad' or 'poly' + scale: 1 + +Metric: + name: DetMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - ColorJitter: + brightness: 0.12549019607843137 + saturation: 0.5 + - IaaAugment: + augmenter_args: + - { 'type': Resize, 'args': { 'size': [ 0.5, 3 ] } } + - { 'type': Fliplr, 'args': { 'p': 0.5 } } + - { 'type': Affine, 'args': { 'rotate': [ -10, 10 ] } } + - MakePseGt: + kernel_num: 7 + min_shrink_ratio: 0.4 + size: 640 + - RandomCropImgMask: + size: [ 640,640 ] + main_key: gt_text + crop_keys: [ 'image', 'gt_text', 'gt_kernels', 'mask' ] + - NormalizeImage: + scale: 1./255. + mean: [ 0.485, 0.456, 0.406 ] + std: [ 0.229, 0.224, 0.225 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'gt_text', 'gt_kernels', 'mask' ] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + limit_side_len: 736 + limit_type: min + - NormalizeImage: + scale: 1./255. + mean: [ 0.485, 0.456, 0.406 ] + std: [ 0.229, 0.224, 0.225 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'shape', 'polys', 'ignore_tags' ] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 8 \ No newline at end of file diff --git a/configs/det/det_r50_vd_sast_icdar15.yml b/configs/det/det_r50_vd_sast_icdar15.yml new file mode 100644 index 0000000000000000000000000000000000000000..ce9582dbd206f22d11f3e80216696fd7a9cff617 --- /dev/null +++ b/configs/det/det_r50_vd_sast_icdar15.yml @@ -0,0 +1,109 @@ +Global: + use_gpu: true + epoch_num: 5000 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/sast_r50_vd_ic15/ + save_epoch_step: 1000 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [4000, 5000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + save_res_path: ./output/sast_r50_vd_ic15/predicts_sast.txt + + +Architecture: + model_type: det + algorithm: SAST + Transform: + Backbone: + name: ResNet_SAST + layers: 50 + Neck: + name: SASTFPN + with_cab: True + Head: + name: SASTHead + +Loss: + name: SASTLoss + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + # name: Cosine + learning_rate: 0.001 + # warmup_epoch: 0 + regularizer: + name: 'L2' + factor: 0 + +PostProcess: + name: SASTPostProcess + score_thresh: 0.5 + sample_pts_num: 2 + nms_thresh: 0.2 + expand_scale: 1.0 + shrink_ratio_of_width: 0.3 + +Metric: + name: DetMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: [./train_data/icdar2013/train_label_json.txt, ./train_data/icdar2015/train_label_json.txt, ./train_data/icdar17_mlt_latin/train_label_json.txt, ./train_data/coco_text_icdar_4pts/train_label_json.txt] + ratio_list: [0.1, 0.45, 0.3, 0.15] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - SASTProcessTrain: + image_shape: [512, 512] + min_crop_side_ratio: 0.3 + min_crop_size: 24 + min_text_size: 4 + max_text_size: 512 + - KeepKeys: + keep_keys: ['image', 'score_map', 'border_map', 'training_mask', 'tvo_map', 'tco_map'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 4 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + resize_long: 1536 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/det/det_r50_vd_sast_totaltext.yml b/configs/det/det_r50_vd_sast_totaltext.yml new file mode 100644 index 0000000000000000000000000000000000000000..ce28e203b4e64533438fe0cbbc392d9ea7332b56 --- /dev/null +++ b/configs/det/det_r50_vd_sast_totaltext.yml @@ -0,0 +1,108 @@ +Global: + use_gpu: true + epoch_num: 5000 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/sast_r50_vd_tt/ + save_epoch_step: 1000 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [4000, 5000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + save_res_path: ./output/sast_r50_vd_tt/predicts_sast.txt + +Architecture: + model_type: det + algorithm: SAST + Transform: + Backbone: + name: ResNet_SAST + layers: 50 + Neck: + name: SASTFPN + with_cab: True + Head: + name: SASTHead + +Loss: + name: SASTLoss + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + # name: Cosine + learning_rate: 0.001 + # warmup_epoch: 0 + regularizer: + name: 'L2' + factor: 0 + +PostProcess: + name: SASTPostProcess + score_thresh: 0.5 + sample_pts_num: 6 + nms_thresh: 0.2 + expand_scale: 1.2 + shrink_ratio_of_width: 0.2 + +Metric: + name: DetMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: [./train_data/art_latin_icdar_14pt/train_no_tt_test/train_label_json.txt, ./train_data/total_text_icdar_14pt/train_label_json.txt] + ratio_list: [0.5, 0.5] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - SASTProcessTrain: + image_shape: [512, 512] + min_crop_side_ratio: 0.3 + min_crop_size: 24 + min_text_size: 4 + max_text_size: 512 + - KeepKeys: + keep_keys: ['image', 'score_map', 'border_map', 'training_mask', 'tvo_map', 'tco_map'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 4 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: + - ./train_data/total_text_icdar_14pt/test_label_json.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: + resize_long: 768 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 \ No newline at end of file diff --git a/configs/det/det_res18_db_v2.0.yml b/configs/det/det_res18_db_v2.0.yml new file mode 100644 index 0000000000000000000000000000000000000000..00a1d97b6232a156ff083101d6f50d9f35b3b8f0 --- /dev/null +++ b/configs/det/det_res18_db_v2.0.yml @@ -0,0 +1,131 @@ +Global: + use_gpu: true + epoch_num: 1200 + log_smooth_window: 20 + print_batch_step: 2 + output_dir: ./output/ch_db_res18/ + save_epoch_step: 1200 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [3000, 2000] + cal_metric_during_train: False + pretrained_model: ./pretrain_models/ResNet18_vd_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_en/img_10.jpg + save_res_path: ./output/det_db/predicts_db.txt + +Architecture: + model_type: det + algorithm: DB + Transform: + Backbone: + name: ResNet_vd + layers: 18 + disable_se: True + Neck: + name: DBFPN + out_channels: 256 + Head: + name: DBHead + k: 50 + +Loss: + name: DBLoss + balance_loss: true + main_loss_type: DiceLoss + alpha: 5 + beta: 10 + ohem_ratio: 3 + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 2 + regularizer: + name: 'L2' + factor: 0 + +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: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/train_icdar2015_label.txt + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - 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: + - KeepKeys: + keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/icdar2015/text_localization/ + label_file_list: + - ./train_data/icdar2015/text_localization/test_icdar2015_label.txt + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - DetLabelEncode: # Class handling label + - DetResizeForTest: +# image_shape: [736, 1280] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: ['image', 'shape', 'polys', 'ignore_tags'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/e2e/e2e_r50_vd_pg.yml b/configs/e2e/e2e_r50_vd_pg.yml new file mode 100644 index 0000000000000000000000000000000000000000..0e144bbab2ad4c9166612a8e5dd8e70d46e0f580 --- /dev/null +++ b/configs/e2e/e2e_r50_vd_pg.yml @@ -0,0 +1,121 @@ +Global: + use_gpu: True + epoch_num: 600 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/pgnet_r50_vd_totaltext/ + save_epoch_step: 10 + # evaluation is run every 0 iterationss after the 1000th iteration + eval_batch_step: [ 0, 1000 ] + cal_metric_during_train: False + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + infer_visual_type: EN # two mode: EN is for english datasets, CN is for chinese datasets + valid_set: totaltext # two mode: totaltext valid curved words, partvgg valid non-curved words + save_res_path: ./output/pgnet_r50_vd_totaltext/predicts_pgnet.txt + character_dict_path: ppocr/utils/ic15_dict.txt + character_type: EN + max_text_length: 50 # the max length in seq + max_text_nums: 30 # the max seq nums in a pic + tcl_len: 64 + +Architecture: + model_type: e2e + algorithm: PGNet + Transform: + Backbone: + name: ResNet + layers: 50 + Neck: + name: PGFPN + Head: + name: PGHead + character_dict_path: ppocr/utils/ic15_dict.txt # the same as Global:character_dict_path + +Loss: + name: PGLoss + tcl_bs: 64 + max_text_length: 50 # the same as Global: max_text_length + max_text_nums: 30 # the same as Global：max_text_nums + pad_num: 36 # the length of dict for pad + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 50 + regularizer: + name: 'L2' + factor: 0.0001 + +PostProcess: + name: PGPostProcess + score_thresh: 0.5 + mode: fast # fast or slow two ways + point_gather_mode: align # same as PGProcessTrain: point_gather_mode + +Metric: + name: E2EMetric + mode: A # two ways for eval, A: label from txt, B: label from gt_mat + gt_mat_dir: ./train_data/total_text/gt # the dir of gt_mat + character_dict_path: ppocr/utils/ic15_dict.txt + main_indicator: f_score_e2e + +Train: + dataset: + name: PGDataSet + data_dir: ./train_data/total_text/train + label_file_list: [./train_data/total_text/train/train.txt] + ratio_list: [1.0] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - E2ELabelEncodeTrain: + - PGProcessTrain: + batch_size: 14 # same as loader: batch_size_per_card + use_resize: True + use_random_crop: False + min_crop_size: 24 + min_text_size: 4 + max_text_size: 512 + point_gather_mode: align # two mode: align and none, align mode is better than none mode + - KeepKeys: + keep_keys: [ 'images', 'tcl_maps', 'tcl_label_maps', 'border_maps','direction_maps', 'training_masks', 'label_list', 'pos_list', 'pos_mask' ] # dataloader will return list in this order + loader: + shuffle: True + drop_last: True + batch_size_per_card: 14 + num_workers: 16 + +Eval: + dataset: + name: PGDataSet + data_dir: ./train_data/total_text/test + label_file_list: [./train_data/total_text/test/test.txt] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - E2ELabelEncodeTest: + - E2EResizeForTest: + max_side_len: 768 + - NormalizeImage: + scale: 1./255. + mean: [ 0.485, 0.456, 0.406 ] + std: [ 0.229, 0.224, 0.225 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'shape', 'polys', 'texts', 'ignore_tags', 'img_id'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 2 diff --git a/configs/kie/layoutlm_series/re_layoutlmv2_xfund_zh.yml b/configs/kie/layoutlm_series/re_layoutlmv2_xfund_zh.yml new file mode 100644 index 0000000000000000000000000000000000000000..9b45460835d529803df28ba8e142ad2278a0b9f1 --- /dev/null +++ b/configs/kie/layoutlm_series/re_layoutlmv2_xfund_zh.yml @@ -0,0 +1,123 @@ +Global: + use_gpu: True + epoch_num: &epoch_num 200 + log_smooth_window: 10 + print_batch_step: 10 + output_dir: ./output/re_layoutlmv2_xfund_zh + save_epoch_step: 2000 + # evaluation is run every 10 iterations after the 0th iteration + eval_batch_step: [ 0, 19 ] + cal_metric_during_train: False + save_inference_dir: + use_visualdl: False + seed: 2022 + infer_img: ppstructure/docs/kie/input/zh_val_21.jpg + save_res_path: ./output/re_layoutlmv2_xfund_zh/res/ + +Architecture: + model_type: kie + algorithm: &algorithm "LayoutLMv2" + Transform: + Backbone: + name: LayoutLMv2ForRe + pretrained: True + checkpoints: + +Loss: + name: LossFromOutput + key: loss + reduction: mean + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + clip_norm: 10 + lr: + learning_rate: 0.00005 + warmup_epoch: 10 + regularizer: + name: L2 + factor: 0.00000 + +PostProcess: + name: VQAReTokenLayoutLMPostProcess + +Metric: + name: VQAReTokenMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_train/image + label_file_list: + - train_data/XFUND/zh_train/train.json + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: True + algorithm: *algorithm + class_path: &class_path train_data/XFUND/class_list_xfun.txt + - VQATokenPad: + max_seq_len: &max_seq_len 512 + return_attention_mask: True + - VQAReTokenRelation: + - VQAReTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids','image', 'entities', 'relations'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 8 + collate_fn: ListCollator + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_val/image + label_file_list: + - train_data/XFUND/zh_val/val.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: True + algorithm: *algorithm + class_path: *class_path + - VQATokenPad: + max_seq_len: *max_seq_len + return_attention_mask: True + - VQAReTokenRelation: + - VQAReTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image','entities', 'relations'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 8 + collate_fn: ListCollator diff --git a/configs/kie/layoutlm_series/re_layoutxlm_xfund_zh.yml b/configs/kie/layoutlm_series/re_layoutxlm_xfund_zh.yml new file mode 100644 index 0000000000000000000000000000000000000000..49c48699b220b7035083128428c65f592e0a28a5 --- /dev/null +++ b/configs/kie/layoutlm_series/re_layoutxlm_xfund_zh.yml @@ -0,0 +1,123 @@ +Global: + use_gpu: True + epoch_num: &epoch_num 130 + log_smooth_window: 10 + print_batch_step: 10 + output_dir: ./output/re_layoutxlm_xfund_zh + save_epoch_step: 2000 + # evaluation is run every 10 iterations after the 0th iteration + eval_batch_step: [ 0, 19 ] + cal_metric_during_train: False + save_inference_dir: + use_visualdl: False + seed: 2022 + infer_img: ppstructure/docs/kie/input/zh_val_21.jpg + save_res_path: ./output/re_layoutxlm_xfund_zh/res/ + +Architecture: + model_type: kie + algorithm: &algorithm "LayoutXLM" + Transform: + Backbone: + name: LayoutXLMForRe + pretrained: True + checkpoints: + +Loss: + name: LossFromOutput + key: loss + reduction: mean + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + clip_norm: 10 + lr: + learning_rate: 0.00005 + warmup_epoch: 10 + regularizer: + name: L2 + factor: 0.00000 + +PostProcess: + name: VQAReTokenLayoutLMPostProcess + +Metric: + name: VQAReTokenMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_train/image + label_file_list: + - train_data/XFUND/zh_train/train.json + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: True + algorithm: *algorithm + class_path: &class_path train_data/XFUND/class_list_xfun.txt + - VQATokenPad: + max_seq_len: &max_seq_len 512 + return_attention_mask: True + - VQAReTokenRelation: + - VQAReTokenChunk: + max_seq_len: *max_seq_len + - TensorizeEntitiesRelations: + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox','attention_mask', 'token_type_ids', 'image', 'entities', 'relations'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 2 + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_val/image + label_file_list: + - train_data/XFUND/zh_val/val.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: True + algorithm: *algorithm + class_path: *class_path + - VQATokenPad: + max_seq_len: *max_seq_len + return_attention_mask: True + - VQAReTokenRelation: + - VQAReTokenChunk: + max_seq_len: *max_seq_len + - TensorizeEntitiesRelations: + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'entities', 'relations'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 8 diff --git a/configs/kie/layoutlm_series/ser_layoutlm_xfund_zh.yml b/configs/kie/layoutlm_series/ser_layoutlm_xfund_zh.yml new file mode 100644 index 0000000000000000000000000000000000000000..cd2d13227e1dd20bc02dc477f160dddfe44c23fa --- /dev/null +++ b/configs/kie/layoutlm_series/ser_layoutlm_xfund_zh.yml @@ -0,0 +1,121 @@ +Global: + use_gpu: True + epoch_num: &epoch_num 200 + log_smooth_window: 10 + print_batch_step: 10 + output_dir: ./output/ser_layoutlm_xfund_zh + save_epoch_step: 2000 + # evaluation is run every 10 iterations after the 0th iteration + eval_batch_step: [ 0, 19 ] + cal_metric_during_train: False + save_inference_dir: + use_visualdl: False + seed: 2022 + infer_img: ppstructure/docs/kie/input/zh_val_42.jpg + save_res_path: ./output/re_layoutlm_xfund_zh/res + +Architecture: + model_type: kie + algorithm: &algorithm "LayoutLM" + Transform: + Backbone: + name: LayoutLMForSer + pretrained: True + checkpoints: + num_classes: &num_classes 7 + +Loss: + name: VQASerTokenLayoutLMLoss + num_classes: *num_classes + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + lr: + name: Linear + learning_rate: 0.00005 + epochs: *epoch_num + warmup_epoch: 2 + regularizer: + name: L2 + factor: 0.00000 + +PostProcess: + name: VQASerTokenLayoutLMPostProcess + class_path: &class_path train_data/XFUND/class_list_xfun.txt + +Metric: + name: VQASerTokenMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_train/image + label_file_list: + - train_data/XFUND/zh_train/train.json + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + - VQATokenPad: + max_seq_len: &max_seq_len 512 + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 16 + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_val/image + label_file_list: + - train_data/XFUND/zh_val/val.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + - VQATokenPad: + max_seq_len: *max_seq_len + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 4 diff --git a/configs/kie/layoutlm_series/ser_layoutlmv2_xfund_zh.yml b/configs/kie/layoutlm_series/ser_layoutlmv2_xfund_zh.yml new file mode 100644 index 0000000000000000000000000000000000000000..d8b00e16614ac20128de2248e476b7bffcf93afb --- /dev/null +++ b/configs/kie/layoutlm_series/ser_layoutlmv2_xfund_zh.yml @@ -0,0 +1,122 @@ +Global: + use_gpu: True + epoch_num: &epoch_num 200 + log_smooth_window: 10 + print_batch_step: 10 + output_dir: ./output/ser_layoutlmv2_xfund_zh/ + save_epoch_step: 2000 + # evaluation is run every 10 iterations after the 0th iteration + eval_batch_step: [ 0, 19 ] + cal_metric_during_train: False + save_inference_dir: + use_visualdl: False + seed: 2022 + infer_img: ppstructure/docs/kie/input/zh_val_42.jpg + save_res_path: ./output/ser_layoutlmv2_xfund_zh/res/ + +Architecture: + model_type: kie + algorithm: &algorithm "LayoutLMv2" + Transform: + Backbone: + name: LayoutLMv2ForSer + pretrained: True + checkpoints: + num_classes: &num_classes 7 + +Loss: + name: VQASerTokenLayoutLMLoss + num_classes: *num_classes + key: "backbone_out" + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + lr: + name: Linear + learning_rate: 0.00005 + epochs: *epoch_num + warmup_epoch: 2 + regularizer: + + name: L2 + factor: 0.00000 + +PostProcess: + name: VQASerTokenLayoutLMPostProcess + class_path: &class_path train_data/XFUND/class_list_xfun.txt + +Metric: + name: VQASerTokenMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_train/image + label_file_list: + - train_data/XFUND/zh_train/train.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + - VQATokenPad: + max_seq_len: &max_seq_len 512 + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_val/image + label_file_list: + - train_data/XFUND/zh_val/val.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + - VQATokenPad: + max_seq_len: *max_seq_len + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 4 diff --git a/configs/kie/layoutlm_series/ser_layoutxlm_xfund_zh.yml b/configs/kie/layoutlm_series/ser_layoutxlm_xfund_zh.yml new file mode 100644 index 0000000000000000000000000000000000000000..e37e06432a8c14b9b79f75d44191048aed7fdf8f --- /dev/null +++ b/configs/kie/layoutlm_series/ser_layoutxlm_xfund_zh.yml @@ -0,0 +1,122 @@ +Global: + use_gpu: True + epoch_num: &epoch_num 200 + log_smooth_window: 10 + print_batch_step: 10 + output_dir: ./output/ser_layoutxlm_xfund_zh + save_epoch_step: 2000 + # evaluation is run every 10 iterations after the 0th iteration + eval_batch_step: [ 0, 19 ] + cal_metric_during_train: False + save_inference_dir: + use_visualdl: False + seed: 2022 + infer_img: ppstructure/docs/kie/input/zh_val_42.jpg + save_res_path: ./output/ser_layoutxlm_xfund_zh/res + +Architecture: + model_type: kie + algorithm: &algorithm "LayoutXLM" + Transform: + Backbone: + name: LayoutXLMForSer + pretrained: True + checkpoints: + num_classes: &num_classes 7 + +Loss: + name: VQASerTokenLayoutLMLoss + num_classes: *num_classes + key: "backbone_out" + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + lr: + name: Linear + learning_rate: 0.00005 + epochs: *epoch_num + warmup_epoch: 2 + regularizer: + name: L2 + factor: 0.00000 + +PostProcess: + name: VQASerTokenLayoutLMPostProcess + class_path: &class_path train_data/XFUND/class_list_xfun.txt + +Metric: + name: VQASerTokenMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_train/image + label_file_list: + - train_data/XFUND/zh_train/train.json + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + - VQATokenPad: + max_seq_len: &max_seq_len 512 + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_val/image + label_file_list: + - train_data/XFUND/zh_val/val.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + - VQATokenPad: + max_seq_len: *max_seq_len + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 4 diff --git a/configs/kie/sdmgr/kie_unet_sdmgr.yml b/configs/kie/sdmgr/kie_unet_sdmgr.yml new file mode 100644 index 0000000000000000000000000000000000000000..f6efb6b753ca5aba15ed125d44a8fdc248c9178c --- /dev/null +++ b/configs/kie/sdmgr/kie_unet_sdmgr.yml @@ -0,0 +1,111 @@ +Global: + use_gpu: True + epoch_num: 60 + log_smooth_window: 20 + print_batch_step: 50 + output_dir: ./output/kie_5/ + save_epoch_step: 50 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [ 0, 80 ] + # 1. If pretrained_model is saved in static mode, such as classification pretrained model + # from static branch, load_static_weights must be set as True. + # 2. If you want to finetune the pretrained models we provide in the docs, + # you should set load_static_weights as False. + load_static_weights: False + cal_metric_during_train: False + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + class_path: &class_path ./train_data/wildreceipt/class_list.txt + infer_img: ./train_data/wildreceipt/1.txt + save_res_path: ./output/sdmgr_kie/predicts_kie.txt + img_scale: [ 1024, 512 ] + +Architecture: + model_type: kie + algorithm: SDMGR + Transform: + Backbone: + name: Kie_backbone + Head: + name: SDMGRHead + +Loss: + name: SDMGRLoss + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Piecewise + learning_rate: 0.001 + decay_epochs: [ 60, 80, 100] + values: [ 0.001, 0.0001, 0.00001] + warmup_epoch: 2 + regularizer: + name: 'L2' + factor: 0.00005 + +PostProcess: + name: None + +Metric: + name: KIEMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/wildreceipt/ + label_file_list: [ './train_data/wildreceipt/wildreceipt_train.txt' ] + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - KieLabelEncode: # Class handling label + character_dict_path: ./train_data/wildreceipt/dict.txt + class_path: *class_path + - KieResize: + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'relations', 'texts', 'points', 'labels', 'tag', 'shape'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 4 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/wildreceipt + label_file_list: + - ./train_data/wildreceipt/wildreceipt_test.txt + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - KieLabelEncode: # Class handling label + character_dict_path: ./train_data/wildreceipt/dict.txt + - KieResize: + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'relations', 'texts', 'points', 'labels', 'tag', 'ori_image', 'ori_boxes', 'shape'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 # must be 1 + num_workers: 4 diff --git a/configs/kie/vi_layoutxlm/re_vi_layoutxlm_xfund_zh.yml b/configs/kie/vi_layoutxlm/re_vi_layoutxlm_xfund_zh.yml new file mode 100644 index 0000000000000000000000000000000000000000..0dabb4bb3a470e8cc5d7e22bb6710da4811f5a5e --- /dev/null +++ b/configs/kie/vi_layoutxlm/re_vi_layoutxlm_xfund_zh.yml @@ -0,0 +1,130 @@ +Global: + use_gpu: True + epoch_num: &epoch_num 130 + log_smooth_window: 10 + print_batch_step: 10 + output_dir: ./output/re_vi_layoutxlm_xfund_zh + save_epoch_step: 2000 + # evaluation is run every 10 iterations after the 0th iteration + eval_batch_step: [ 0, 19 ] + cal_metric_during_train: False + save_inference_dir: + use_visualdl: False + seed: 2022 + infer_img: ppstructure/docs/kie/input/zh_val_21.jpg + save_res_path: ./output/re/xfund_zh/with_gt + kie_rec_model_dir: + kie_det_model_dir: + +Architecture: + model_type: kie + algorithm: &algorithm "LayoutXLM" + Transform: + Backbone: + name: LayoutXLMForRe + pretrained: True + mode: vi + checkpoints: + +Loss: + name: LossFromOutput + key: loss + reduction: mean + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + clip_norm: 10 + lr: + learning_rate: 0.00005 + warmup_epoch: 10 + regularizer: + name: L2 + factor: 0.00000 + +PostProcess: + name: VQAReTokenLayoutLMPostProcess + +Metric: + name: VQAReTokenMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_train/image + label_file_list: + - train_data/XFUND/zh_train/train.json + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: True + algorithm: *algorithm + class_path: &class_path train_data/XFUND/class_list_xfun.txt + use_textline_bbox_info: &use_textline_bbox_info True + order_method: &order_method "tb-yx" + - VQATokenPad: + max_seq_len: &max_seq_len 512 + return_attention_mask: True + - VQAReTokenRelation: + - VQAReTokenChunk: + max_seq_len: *max_seq_len + - TensorizeEntitiesRelations: + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox','attention_mask', 'token_type_ids', 'entities', 'relations'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 2 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_val/image + label_file_list: + - train_data/XFUND/zh_val/val.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: True + algorithm: *algorithm + class_path: *class_path + use_textline_bbox_info: *use_textline_bbox_info + order_method: *order_method + - VQATokenPad: + max_seq_len: *max_seq_len + return_attention_mask: True + - VQAReTokenRelation: + - VQAReTokenChunk: + max_seq_len: *max_seq_len + - TensorizeEntitiesRelations: + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'entities', 'relations'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 8 diff --git a/configs/kie/vi_layoutxlm/re_vi_layoutxlm_xfund_zh_udml.yml b/configs/kie/vi_layoutxlm/re_vi_layoutxlm_xfund_zh_udml.yml new file mode 100644 index 0000000000000000000000000000000000000000..b763dc0e7c2b7ab532c7737bacc0447f14023d20 --- /dev/null +++ b/configs/kie/vi_layoutxlm/re_vi_layoutxlm_xfund_zh_udml.yml @@ -0,0 +1,177 @@ +Global: + use_gpu: True + epoch_num: &epoch_num 130 + log_smooth_window: 10 + print_batch_step: 10 + output_dir: ./output/re_vi_layoutxlm_xfund_zh_udml + save_epoch_step: 2000 + # evaluation is run every 10 iterations after the 0th iteration + eval_batch_step: [ 0, 19 ] + cal_metric_during_train: False + save_inference_dir: + use_visualdl: False + seed: 2022 + infer_img: ppstructure/docs/kie/input/zh_val_21.jpg + save_res_path: ./output/re/xfund_zh/with_gt + +Architecture: + model_type: &model_type "kie" + name: DistillationModel + algorithm: Distillation + Models: + Teacher: + pretrained: + freeze_params: false + return_all_feats: true + model_type: *model_type + algorithm: &algorithm "LayoutXLM" + Transform: + Backbone: + name: LayoutXLMForRe + pretrained: True + mode: vi + checkpoints: + Student: + pretrained: + freeze_params: false + return_all_feats: true + model_type: *model_type + algorithm: *algorithm + Transform: + Backbone: + name: LayoutXLMForRe + pretrained: True + mode: vi + checkpoints: + +Loss: + name: CombinedLoss + loss_config_list: + - DistillationLossFromOutput: + weight: 1.0 + model_name_list: ["Student", "Teacher"] + key: loss + reduction: mean + - DistillationVQADistanceLoss: + weight: 0.5 + mode: "l2" + model_name_pairs: + - ["Student", "Teacher"] + key: hidden_states + index: 5 + name: "loss_5" + - DistillationVQADistanceLoss: + weight: 0.5 + mode: "l2" + model_name_pairs: + - ["Student", "Teacher"] + key: hidden_states + index: 8 + name: "loss_8" + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + clip_norm: 10 + lr: + learning_rate: 0.00005 + warmup_epoch: 10 + regularizer: + name: L2 + factor: 0.00000 + +PostProcess: + name: DistillationRePostProcess + model_name: ["Student", "Teacher"] + key: null + + +Metric: + name: DistillationMetric + base_metric_name: VQAReTokenMetric + main_indicator: hmean + key: "Student" + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_train/image + label_file_list: + - train_data/XFUND/zh_train/train.json + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: True + algorithm: *algorithm + class_path: &class_path train_data/XFUND/class_list_xfun.txt + use_textline_bbox_info: &use_textline_bbox_info True + # [None, "tb-yx"] + order_method: &order_method "tb-yx" + - VQATokenPad: + max_seq_len: &max_seq_len 512 + return_attention_mask: True + - VQAReTokenRelation: + - VQAReTokenChunk: + max_seq_len: *max_seq_len + - TensorizeEntitiesRelations: + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox','attention_mask', 'token_type_ids', 'entities', 'relations'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 2 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_val/image + label_file_list: + - train_data/XFUND/zh_val/val.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: True + algorithm: *algorithm + class_path: *class_path + use_textline_bbox_info: *use_textline_bbox_info + order_method: *order_method + - VQATokenPad: + max_seq_len: *max_seq_len + return_attention_mask: True + - VQAReTokenRelation: + - VQAReTokenChunk: + max_seq_len: *max_seq_len + - TensorizeEntitiesRelations: + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'entities', 'relations'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 8 + + diff --git a/configs/kie/vi_layoutxlm/ser_vi_layoutxlm_xfund_zh.yml b/configs/kie/vi_layoutxlm/ser_vi_layoutxlm_xfund_zh.yml new file mode 100644 index 0000000000000000000000000000000000000000..57eaace89ca7b46dafd52f4c2a411a269206f868 --- /dev/null +++ b/configs/kie/vi_layoutxlm/ser_vi_layoutxlm_xfund_zh.yml @@ -0,0 +1,138 @@ +Global: + use_gpu: True + epoch_num: &epoch_num 200 + log_smooth_window: 10 + print_batch_step: 10 + output_dir: ./output/ser_vi_layoutxlm_xfund_zh + save_epoch_step: 2000 + # evaluation is run every 10 iterations after the 0th iteration + eval_batch_step: [ 0, 19 ] + cal_metric_during_train: False + save_inference_dir: + use_visualdl: False + seed: 2022 + infer_img: ppstructure/docs/kie/input/zh_val_42.jpg + d2s_train_image_shape: [3, 224, 224] + # if you want to predict using the groundtruth ocr info, + # you can use the following config + # infer_img: train_data/XFUND/zh_val/val.json + # infer_mode: False + + save_res_path: ./output/ser/xfund_zh/res + kie_rec_model_dir: + kie_det_model_dir: + amp_custom_white_list: ['scale', 'concat', 'elementwise_add'] + +Architecture: + model_type: kie + algorithm: &algorithm "LayoutXLM" + Transform: + Backbone: + name: LayoutXLMForSer + pretrained: True + checkpoints: + # one of base or vi + mode: vi + num_classes: &num_classes 7 + +Loss: + name: VQASerTokenLayoutLMLoss + num_classes: *num_classes + key: "backbone_out" + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + lr: + name: Linear + learning_rate: 0.00005 + epochs: *epoch_num + warmup_epoch: 2 + regularizer: + name: L2 + factor: 0.00000 + +PostProcess: + name: VQASerTokenLayoutLMPostProcess + class_path: &class_path train_data/XFUND/class_list_xfun.txt + +Metric: + name: VQASerTokenMetric + main_indicator: hmean + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_train/image + label_file_list: + - train_data/XFUND/zh_train/train.json + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + use_textline_bbox_info: &use_textline_bbox_info True + # one of [None, "tb-yx"] + order_method: &order_method "tb-yx" + - VQATokenPad: + max_seq_len: &max_seq_len 512 + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 8 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_val/image + label_file_list: + - train_data/XFUND/zh_val/val.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + use_textline_bbox_info: *use_textline_bbox_info + order_method: *order_method + - VQATokenPad: + max_seq_len: *max_seq_len + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 4 diff --git a/configs/kie/vi_layoutxlm/ser_vi_layoutxlm_xfund_zh_udml.yml b/configs/kie/vi_layoutxlm/ser_vi_layoutxlm_xfund_zh_udml.yml new file mode 100644 index 0000000000000000000000000000000000000000..79fa46c15d294d653edb52e7928a466ae7aba1ba --- /dev/null +++ b/configs/kie/vi_layoutxlm/ser_vi_layoutxlm_xfund_zh_udml.yml @@ -0,0 +1,182 @@ +Global: + use_gpu: True + epoch_num: &epoch_num 200 + log_smooth_window: 10 + print_batch_step: 10 + output_dir: ./output/ser_vi_layoutxlm_xfund_zh_udml + save_epoch_step: 2000 + # evaluation is run every 10 iterations after the 0th iteration + eval_batch_step: [ 0, 19 ] + cal_metric_during_train: False + save_inference_dir: + use_visualdl: False + seed: 2022 + infer_img: ppstructure/docs/kie/input/zh_val_42.jpg + save_res_path: ./output/ser_layoutxlm_xfund_zh/res + + +Architecture: + model_type: &model_type "kie" + name: DistillationModel + algorithm: Distillation + Models: + Teacher: + pretrained: + freeze_params: false + return_all_feats: true + model_type: *model_type + algorithm: &algorithm "LayoutXLM" + Transform: + Backbone: + name: LayoutXLMForSer + pretrained: True + # one of base or vi + mode: vi + checkpoints: + num_classes: &num_classes 7 + Student: + pretrained: + freeze_params: false + return_all_feats: true + model_type: *model_type + algorithm: *algorithm + Transform: + Backbone: + name: LayoutXLMForSer + pretrained: True + # one of base or vi + mode: vi + checkpoints: + num_classes: *num_classes + + +Loss: + name: CombinedLoss + loss_config_list: + - DistillationVQASerTokenLayoutLMLoss: + weight: 1.0 + model_name_list: ["Student", "Teacher"] + key: backbone_out + num_classes: *num_classes + - DistillationSERDMLLoss: + weight: 1.0 + act: "softmax" + use_log: true + model_name_pairs: + - ["Student", "Teacher"] + key: backbone_out + - DistillationVQADistanceLoss: + weight: 0.5 + mode: "l2" + model_name_pairs: + - ["Student", "Teacher"] + key: hidden_states_5 + name: "loss_5" + - DistillationVQADistanceLoss: + weight: 0.5 + mode: "l2" + model_name_pairs: + - ["Student", "Teacher"] + key: hidden_states_8 + name: "loss_8" + + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + lr: + name: Linear + learning_rate: 0.00005 + epochs: *epoch_num + warmup_epoch: 10 + regularizer: + name: L2 + factor: 0.00000 + +PostProcess: + name: DistillationSerPostProcess + model_name: ["Student", "Teacher"] + key: backbone_out + class_path: &class_path train_data/XFUND/class_list_xfun.txt + +Metric: + name: DistillationMetric + base_metric_name: VQASerTokenMetric + main_indicator: hmean + key: "Student" + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_train/image + label_file_list: + - train_data/XFUND/zh_train/train.json + ratio_list: [ 1.0 ] + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + # one of [None, "tb-yx"] + order_method: &order_method "tb-yx" + - VQATokenPad: + max_seq_len: &max_seq_len 512 + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: True + drop_last: False + batch_size_per_card: 4 + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/XFUND/zh_val/image + label_file_list: + - train_data/XFUND/zh_val/val.json + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VQATokenLabelEncode: # Class handling label + contains_re: False + algorithm: *algorithm + class_path: *class_path + order_method: *order_method + - VQATokenPad: + max_seq_len: *max_seq_len + return_attention_mask: True + - VQASerTokenChunk: + max_seq_len: *max_seq_len + - Resize: + size: [224,224] + - NormalizeImage: + scale: 1 + mean: [ 123.675, 116.28, 103.53 ] + std: [ 58.395, 57.12, 57.375 ] + order: 'hwc' + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'image', 'labels'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 4 diff --git a/configs/rec/PP-OCRv3/ch_PP-OCRv3_rec.yml b/configs/rec/PP-OCRv3/ch_PP-OCRv3_rec.yml new file mode 100644 index 0000000000000000000000000000000000000000..a07d5057a8835a7d42592a51cf306089b652a135 --- /dev/null +++ b/configs/rec/PP-OCRv3/ch_PP-OCRv3_rec.yml @@ -0,0 +1,119 @@ +Global: + device: gpu + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/rec_ppocr_v3 + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/ch/word_1.jpg + character_dict_path: &character_dict_path ppocr/utils/ppocr_keys_v1.txt + max_text_length: &max_text_length 25 + use_space_char: &use_space_char true + +Export: + export_dir: + export_shape: [ 1, 3, 48, 320 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.001 + weight_decay: 3.0e-05 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 5 + +Architecture: + model_type: rec + algorithm: SVTR_LCNet + Transform: + Backbone: + name: MobileNetV1Enhance + scale: 0.5 + last_conv_stride: [1, 2] + last_pool_type: avg + last_pool_kernel_size: [2, 2] + Head: + name: MultiHead + head_list: + - CTCHead: + Neck: + name: svtr + dims: 64 + depth: 2 + hidden_dims: 120 + use_guide: True + - SARHead: + enc_dim: 512 + max_text_length: *max_text_length + +Loss: + name: MultiLoss + loss_config_list: + - CTCLoss: + - SARLoss: + +PostProcess: + name: CTCLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + ignore_space: False + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: + - ./train_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: + - RecResizeImg: + image_shape: [3, 48, 320] + - KeepKeys: + keep_keys: [image, label_ctc, label_sar, length, valid_ratio] + loader: + shuffle: true + batch_size_per_card: 128 + drop_last: true + num_workers: 4 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + ext_op_transform_idx: 1 + label_file_list: + - ./train_data/val_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - MultiLabelEncode: + - RecResizeImg: + image_shape: [3, 48, 320] + - KeepKeys: + keep_keys: [ image, label_ctc, label_sar, length, valid_ratio ] + loader: + shuffle: false + drop_last: false + batch_size_per_card: 128 + num_workers: 4 \ No newline at end of file diff --git a/configs/rec/PP-OCRv3/ch_PP-OCRv3_rec_distillation.yml b/configs/rec/PP-OCRv3/ch_PP-OCRv3_rec_distillation.yml new file mode 100644 index 0000000000000000000000000000000000000000..23321e038e5da996c17786ea2b0e773c73a074f5 --- /dev/null +++ b/configs/rec/PP-OCRv3/ch_PP-OCRv3_rec_distillation.yml @@ -0,0 +1,205 @@ +Global: + device: gpu + epoch_num: 800 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/rec_ppocr_v3_distillation + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/ch/word_1.jpg + character_dict_path: &character_dict_path ppocr/utils/ppocr_keys_v1.txt + max_text_length: &max_text_length 25 + use_space_char: &use_space_char true + +Export: + export_dir: + export_shape: [ 1, 3, 48, 320 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 3.0e-05 + amsgrad: true + +LRScheduler: + name: MultiStepLR + milestones: [700] + warmup_epoch: 5 + +Architecture: + model_type: &model_type "rec" + name: DistillationModel + algorithm: Distillation + Models: + Teacher: + pretrained: + freeze_params: false + return_all_feats: true + model_type: *model_type + algorithm: SVTR_LCNet + Transform: + Backbone: + name: MobileNetV1Enhance + scale: 0.5 + last_conv_stride: [1, 2] + last_pool_type: avg + last_pool_kernel_size: [2, 2] + Head: + name: MultiHead + head_list: + - CTCHead: + Neck: + name: svtr + dims: 64 + depth: 2 + hidden_dims: 120 + use_guide: True + - SARHead: + enc_dim: 512 + max_text_length: *max_text_length + Student: + pretrained: + freeze_params: false + return_all_feats: true + model_type: *model_type + algorithm: SVTR_LCNet + Transform: + Backbone: + name: MobileNetV1Enhance + scale: 0.5 + last_conv_stride: [1, 2] + last_pool_type: avg + last_pool_kernel_size: [2, 2] + Head: + name: MultiHead + head_list: + - CTCHead: + Neck: + name: svtr + dims: 64 + depth: 2 + hidden_dims: 120 + use_guide: True + Head: + fc_decay: 0.00001 + - SARHead: + enc_dim: 512 + max_text_length: *max_text_length +Loss: + name: CombinedLoss + loss_config_list: + - DistillationDMLLoss: + weight: 1.0 + act: "softmax" + use_log: true + model_name_pairs: + - ["Student", "Teacher"] + key: head_out + multi_head: True + dis_head: ctc + name: dml_ctc + - DistillationDMLLoss: + weight: 0.5 + act: "softmax" + use_log: true + model_name_pairs: + - ["Student", "Teacher"] + key: head_out + multi_head: True + dis_head: sar + name: dml_sar + - DistillationDistanceLoss: + weight: 1.0 + mode: "l2" + model_name_pairs: + - ["Student", "Teacher"] + key: backbone_out + - DistillationCTCLoss: + weight: 1.0 + model_name_list: ["Student", "Teacher"] + key: head_out + multi_head: True + - DistillationSARLoss: + weight: 1.0 + model_name_list: ["Student", "Teacher"] + key: head_out + multi_head: True + +PostProcess: + name: DistillationCTCLabelDecode + model_name: ["Student", "Teacher"] + key: head_out + multi_head: True + character_dict_path: *character_dict_path + use_space_char: *use_space_char + + +Metric: + name: DistillationMetric + base_metric_name: RecMetric + main_indicator: acc + key: "Student" + ignore_space: False + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data + ext_op_transform_idx: 1 + label_file_list: + - ./train_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: + - RecResizeImg: + image_shape: [3, 48, 320] + - KeepKeys: + keep_keys: + - image + - label_ctc + - label_sar + - length + - valid_ratio + loader: + shuffle: true + batch_size_per_card: 128 + drop_last: true + num_workers: 4 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: + - ./train_data/train_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - MultiLabelEncode: + - RecResizeImg: + image_shape: [3, 48, 320] + - KeepKeys: + keep_keys: + - image + - label_ctc + - label_sar + - length + - valid_ratio + loader: + shuffle: false + drop_last: false + batch_size_per_card: 128 + num_workers: 4 diff --git a/configs/rec/PP-OCRv4/ch_PP-OCRv4_rec.yml b/configs/rec/PP-OCRv4/ch_PP-OCRv4_rec.yml new file mode 100644 index 0000000000000000000000000000000000000000..8b1dd9f083841a88fda49be0a45cd466c222d74f --- /dev/null +++ b/configs/rec/PP-OCRv4/ch_PP-OCRv4_rec.yml @@ -0,0 +1,133 @@ +Global: + device: gpu + epoch_num: 200 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/rec_ppocr_v4 + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/ch/word_1.jpg + character_dict_path: &character_dict_path ppocr/utils/ppocr_keys_v1.txt + max_text_length: &max_text_length 25 + use_space_char: &use_space_char true + +Export: + export_dir: + export_shape: [ 1, 3, 48, 320 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.001 + weight_decay: 3.0e-05 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 5 + +Architecture: + model_type: rec + algorithm: SVTR_LCNet + Transform: + Backbone: + name: PPLCNetV3 + scale: 0.95 + Head: + name: MultiHead + head_list: + - CTCHead: + Neck: + name: svtr + dims: 120 + depth: 2 + hidden_dims: 120 + kernel_size: [1, 3] + use_guide: True + - NRTRHead: + nrtr_dim: 384 + max_text_length: *max_text_length + +Loss: + name: MultiLoss + loss_config_list: + - CTCLoss: + - NRTRLoss: + +PostProcess: + name: CTCLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: MultiScaleDataSet + ds_width: false + data_dir: ./train_data/ + ext_op_transform_idx: 1 + label_file_list: + - ./train_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 192 + 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: ./train_data + label_file_list: + - ./train_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: 128 + num_workers: 4 diff --git a/configs/rec/PP-OCRv4/ch_PP-OCRv4_rec_distill.yml b/configs/rec/PP-OCRv4/ch_PP-OCRv4_rec_distill.yml new file mode 100644 index 0000000000000000000000000000000000000000..fe8aaa2575cba779d911bdd06eec9bbfbc6504f3 --- /dev/null +++ b/configs/rec/PP-OCRv4/ch_PP-OCRv4_rec_distill.yml @@ -0,0 +1,200 @@ +Global: + device: gpu + epoch_num: 200 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/rec_ppocr_v4__distill + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/ch/word_1.jpg + character_dict_path: &character_dict_path ppocr/utils/ppocr_keys_v1.txt + max_text_length: &max_text_length 25 + use_space_char: &use_space_char true + +Export: + export_dir: + export_shape: [ 1, 3, 48, 320 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.001 + weight_decay: 3.0e-05 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 5 + +Architecture: + model_type: rec + name: DistillationModel + algorithm: Distillation + Models: + Teacher: + pretrained: + freeze_params: true + return_all_feats: true + model_type: rec + algorithm: SVTR + Transform: null + Backbone: + name: PPHGNet_small + Head: + name: MultiHead + head_list: + - CTCHead: + Neck: + name: svtr + dims: 120 + depth: 2 + hidden_dims: 120 + kernel_size: [1, 3] + use_guide: True + Head: + fc_decay: 0.00001 + - NRTRHead: + nrtr_dim: 384 + max_text_length: *max_text_length + Student: + pretrained: + freeze_params: false + return_all_feats: true + model_type: rec + algorithm: SVTR + Transform: null + Backbone: + name: PPLCNetV3 + scale: 0.95 + Head: + name: MultiHead + head_list: + - CTCHead: + Neck: + name: svtr + dims: 120 + depth: 2 + hidden_dims: 120 + kernel_size: [1, 3] + use_guide: True + - NRTRHead: + nrtr_dim: 384 + max_text_length: *max_text_length +Loss: + name: CombinedLoss + loss_config_list: + - DistillationDKDLoss: + weight: 0.1 + model_name_pairs: + - - Student + - Teacher + key: head_out + multi_head: true + alpha: 1.0 + beta: 2.0 + dis_head: nrtr + name: dkd + - DistillationCTCLoss: + weight: 1.0 + model_name_list: + - Student + key: head_out + multi_head: true + - DistillCTCLogits: + weight: 1.0 + reduction: mean + model_name_pairs: + - - Student + - Teacher + key: head_out + - DistillationNRTRLoss: + weight: 1.0 + smoothing: false + model_name_list: + - Student + key: head_out + multi_head: true + +PostProcess: + name: DistillationCTCLabelDecode + model_name: + - Student + key: head_out + multi_head: true + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: DistillationMetric + base_metric_name: RecMetric + main_indicator: acc + key: Student + ignore_space: false +Train: + dataset: + name: MultiScaleDataSet + ds_width: false + data_dir: ./train_data/ + ext_op_transform_idx: 1 + label_file_list: + - ./train_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 192 + 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: ./train_data + label_file_list: + - ./train_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: 128 + num_workers: 4 diff --git a/configs/rec/PP-OCRv4/ch_PP-OCRv4_rec_hgnet.yml b/configs/rec/PP-OCRv4/ch_PP-OCRv4_rec_hgnet.yml new file mode 100644 index 0000000000000000000000000000000000000000..21536dbd462ea6526915cefdc2277259538e8588 --- /dev/null +++ b/configs/rec/PP-OCRv4/ch_PP-OCRv4_rec_hgnet.yml @@ -0,0 +1,134 @@ +Global: + device: gpu + epoch_num: 200 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/rec_ppocr_v4_hgnet + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: img + character_dict_path: &character_dict_path ppocr/utils/ppocr_keys_v1.txt + max_text_length: &max_text_length 25 + use_space_char: &use_space_char true + +Export: + export_dir: + export_shape: [ 1, 3, 48, 320 ] + dynamic_axes: [ 0, 2, 3 ] + + + +Optimizer: + name: Adam + lr: 0.001 + weight_decay: 3.0e-05 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 5 + +Architecture: + model_type: rec + algorithm: SVTR_HGNet + Transform: + Backbone: + name: PPHGNet_small + Head: + name: MultiHead + head_list: + - CTCHead: + Neck: + name: svtr + dims: 120 + depth: 2 + hidden_dims: 120 + kernel_size: [1, 3] + use_guide: True + - NRTRHead: + nrtr_dim: 384 + max_text_length: *max_text_length + +Loss: + name: MultiLoss + loss_config_list: + - CTCLoss: + - NRTRLoss: + +PostProcess: + name: CTCLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: MultiScaleDataSet + ds_width: false + data_dir: ./train_data/ + ext_op_transform_idx: 1 + label_file_list: + - ./train_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 128 + 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: ./train_data + label_file_list: + - ./train_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: 128 + num_workers: 4 diff --git a/configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml b/configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml new file mode 100644 index 0000000000000000000000000000000000000000..159a91edda7ec4e9532f150e9eb7f3945177995b --- /dev/null +++ b/configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml @@ -0,0 +1,110 @@ +Global: + debug: false + use_gpu: true + epoch_num: 800 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_mobile_pp-OCRv2 + save_epoch_step: 3 + eval_batch_step: [0, 2000] + cal_metric_during_train: true + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: false + infer_img: doc/imgs_words/ch/word_1.jpg + character_dict_path: ppocr/utils/ppocr_keys_v1.txt + max_text_length: 25 + infer_mode: false + use_space_char: true + distributed: true + save_res_path: ./output/rec/predicts_mobile_pp-OCRv2.txt + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Piecewise + decay_epochs : [700] + values : [0.001, 0.0001] + warmup_epoch: 5 + regularizer: + name: L2 + factor: 2.0e-05 + + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV1Enhance + scale: 0.5 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 64 + Head: + name: CTCHead + mid_channels: 96 + fc_decay: 0.00002 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: + - ./train_data/train_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - RecAug: + - CTCLabelEncode: + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: true + batch_size_per_card: 128 + drop_last: true + num_workers: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: + - ./train_data/val_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - CTCLabelEncode: + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: false + drop_last: false + batch_size_per_card: 128 + num_workers: 8 diff --git a/configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec_distillation.yml b/configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec_distillation.yml new file mode 100644 index 0000000000000000000000000000000000000000..83e834fd2871bde1d1c43072f8fa7da66d514d7a --- /dev/null +++ b/configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec_distillation.yml @@ -0,0 +1,160 @@ +Global: + debug: false + use_gpu: true + epoch_num: 800 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_pp-OCRv2_distillation + save_epoch_step: 3 + eval_batch_step: [0, 2000] + cal_metric_during_train: true + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: false + infer_img: doc/imgs_words/ch/word_1.jpg + character_dict_path: ppocr/utils/ppocr_keys_v1.txt + max_text_length: 25 + infer_mode: false + use_space_char: true + distributed: true + save_res_path: ./output/rec/predicts_pp-OCRv2_distillation.txt + amp_custom_black_list: ['matmul','matmul_v2','elementwise_add'] + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Piecewise + decay_epochs : [700] + values : [0.001, 0.0001] + warmup_epoch: 5 + regularizer: + name: L2 + factor: 2.0e-05 + +Architecture: + model_type: &model_type "rec" + name: DistillationModel + algorithm: Distillation + Models: + Teacher: + pretrained: + freeze_params: false + return_all_feats: true + model_type: *model_type + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV1Enhance + scale: 0.5 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 64 + Head: + name: CTCHead + mid_channels: 96 + fc_decay: 0.00002 + Student: + pretrained: + freeze_params: false + return_all_feats: true + model_type: *model_type + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV1Enhance + scale: 0.5 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 64 + Head: + name: CTCHead + mid_channels: 96 + fc_decay: 0.00002 + + +Loss: + name: CombinedLoss + loss_config_list: + - DistillationCTCLoss: + weight: 1.0 + model_name_list: ["Student", "Teacher"] + key: head_out + - DistillationDMLLoss: + weight: 1.0 + act: "softmax" + use_log: true + model_name_pairs: + - ["Student", "Teacher"] + key: head_out + - DistillationDistanceLoss: + weight: 1.0 + mode: "l2" + model_name_pairs: + - ["Student", "Teacher"] + key: backbone_out + +PostProcess: + name: DistillationCTCLabelDecode + model_name: ["Student", "Teacher"] + key: head_out + +Metric: + name: DistillationMetric + base_metric_name: RecMetric + main_indicator: acc + key: "Student" + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: + - ./train_data/train_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - RecAug: + - CTCLabelEncode: + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: true + batch_size_per_card: 128 + drop_last: true + num_sections: 1 + num_workers: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: + - ./train_data/val_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - CTCLabelEncode: + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: false + drop_last: false + batch_size_per_card: 128 + num_workers: 8 diff --git a/configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec_enhanced_ctc_loss.yml b/configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec_enhanced_ctc_loss.yml new file mode 100644 index 0000000000000000000000000000000000000000..a515cf5c24e9b12f98986cacda20736918f625c6 --- /dev/null +++ b/configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec_enhanced_ctc_loss.yml @@ -0,0 +1,124 @@ +Global: + debug: false + use_gpu: true + epoch_num: 800 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_mobile_pp-OCRv2_enhanced_ctc_loss + save_epoch_step: 3 + eval_batch_step: [0, 2000] + cal_metric_during_train: true + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: false + infer_img: doc/imgs_words/ch/word_1.jpg + character_dict_path: ppocr/utils/ppocr_keys_v1.txt + max_text_length: 25 + infer_mode: false + use_space_char: true + distributed: true + save_res_path: ./output/rec/predicts_mobile_pp-OCRv2_enhanced_ctc_loss.txt + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Piecewise + decay_epochs : [700] + values : [0.001, 0.0001] + warmup_epoch: 5 + regularizer: + name: L2 + factor: 2.0e-05 + + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV1Enhance + scale: 0.5 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 64 + Head: + name: CTCHead + mid_channels: 96 + fc_decay: 0.00002 + return_feats: true + +Loss: + name: CombinedLoss + loss_config_list: + - CTCLoss: + use_focal_loss: false + weight: 1.0 + - CenterLoss: + weight: 0.05 + num_classes: 6625 + feat_dim: 96 + center_file_path: + # you can also try to add ace loss on your own dataset + # - ACELoss: + # weight: 0.1 + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: + - ./train_data/train_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - RecAug: + - CTCLabelEncode: + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: + - image + - label + - length + - label_ace + loader: + shuffle: true + batch_size_per_card: 128 + drop_last: true + num_workers: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: + - ./train_data/val_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - CTCLabelEncode: + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: false + drop_last: false + batch_size_per_card: 128 + num_workers: 8 diff --git a/configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml b/configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml new file mode 100644 index 0000000000000000000000000000000000000000..7e72bc0249bfe7159d6aa2cbbe6a2e1c22bd000f --- /dev/null +++ b/configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml @@ -0,0 +1,100 @@ +Global: + use_gpu: true + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_chinese_common_v2.0 + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words/ch/word_1.jpg + # for data or label process + character_dict_path: ppocr/utils/ppocr_keys_v1.txt + max_text_length: 25 + infer_mode: False + use_space_char: True + save_res_path: ./output/rec/predicts_chinese_common_v2.0.txt + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 5 + regularizer: + name: 'L2' + factor: 0.00004 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: ResNet + layers: 34 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 256 + Head: + name: CTCHead + fc_decay: 0.00004 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/train_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RecAug: + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/val_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml b/configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml new file mode 100644 index 0000000000000000000000000000000000000000..a5ec0ce0d7f4e5c3be2a23ff27a020fe547763d7 --- /dev/null +++ b/configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml @@ -0,0 +1,102 @@ +Global: + use_gpu: true + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_chinese_lite_v2.0 + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words/ch/word_1.jpg + # for data or label process + character_dict_path: ppocr/utils/ppocr_keys_v1.txt + max_text_length: 25 + infer_mode: False + use_space_char: True + save_res_path: ./output/rec/predicts_chinese_lite_v2.0.txt + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + warmup_epoch: 5 + regularizer: + name: 'L2' + factor: 0.00001 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: [1, 2, 2, 2] + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 0.00001 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/train_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RecAug: + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: ["./train_data/val_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/generate_multi_language_configs.py b/configs/rec/multi_language/generate_multi_language_configs.py new file mode 100644 index 0000000000000000000000000000000000000000..9d586b62f37da6026bcc43e7535128c399a016a4 --- /dev/null +++ b/configs/rec/multi_language/generate_multi_language_configs.py @@ -0,0 +1,226 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import yaml +from argparse import ArgumentParser, RawDescriptionHelpFormatter +import os.path +import logging +logging.basicConfig(level=logging.INFO) + +support_list = { + 'it': 'italian', + 'xi': 'spanish', + 'pu': 'portuguese', + 'ru': 'russian', + 'ar': 'arabic', + 'ta': 'tamil', + 'ug': 'uyghur', + 'fa': 'persian', + 'ur': 'urdu', + 'rs': 'serbian latin', + 'oc': 'occitan', + 'rsc': 'serbian cyrillic', + 'bg': 'bulgarian', + 'uk': 'ukranian', + 'be': 'belarusian', + 'te': 'telugu', + 'ka': 'kannada', + 'chinese_cht': 'chinese tradition', + 'hi': 'hindi', + 'mr': 'marathi', + 'ne': 'nepali', +} + +latin_lang = [ + 'af', 'az', 'bs', 'cs', 'cy', 'da', 'de', 'es', 'et', 'fr', 'ga', 'hr', + 'hu', 'id', 'is', 'it', 'ku', 'la', 'lt', 'lv', 'mi', 'ms', 'mt', 'nl', + 'no', 'oc', 'pi', 'pl', 'pt', 'ro', 'rs_latin', 'sk', 'sl', 'sq', 'sv', + 'sw', 'tl', 'tr', 'uz', 'vi', 'latin' +] +arabic_lang = ['ar', 'fa', 'ug', 'ur'] +cyrillic_lang = [ + 'ru', 'rs_cyrillic', 'be', 'bg', 'uk', 'mn', 'abq', 'ady', 'kbd', 'ava', + 'dar', 'inh', 'che', 'lbe', 'lez', 'tab', 'cyrillic' +] +devanagari_lang = [ + 'hi', 'mr', 'ne', 'bh', 'mai', 'ang', 'bho', 'mah', 'sck', 'new', 'gom', + 'sa', 'bgc', 'devanagari' +] +multi_lang = latin_lang + arabic_lang + cyrillic_lang + devanagari_lang + +assert (os.path.isfile("./rec_multi_language_lite_train.yml") + ), "Loss basic configuration file rec_multi_language_lite_train.yml.\ +You can download it from \ +https://github.com/PaddlePaddle/PaddleOCR/tree/dygraph/configs/rec/multi_language/" + +global_config = yaml.load( + open("./rec_multi_language_lite_train.yml", 'rb'), Loader=yaml.Loader) +project_path = os.path.abspath(os.path.join(os.getcwd(), "../../../")) + + +class ArgsParser(ArgumentParser): + def __init__(self): + super(ArgsParser, self).__init__( + formatter_class=RawDescriptionHelpFormatter) + self.add_argument( + "-o", "--opt", nargs='+', help="set configuration options") + self.add_argument( + "-l", + "--language", + nargs='+', + help="set language type, support {}".format(support_list)) + self.add_argument( + "--train", + type=str, + help="you can use this command to change the train dataset default path" + ) + self.add_argument( + "--val", + type=str, + help="you can use this command to change the eval dataset default path" + ) + self.add_argument( + "--dict", + type=str, + help="you can use this command to change the dictionary default path" + ) + self.add_argument( + "--data_dir", + type=str, + help="you can use this command to change the dataset default root path" + ) + + def parse_args(self, argv=None): + args = super(ArgsParser, self).parse_args(argv) + args.opt = self._parse_opt(args.opt) + args.language = self._set_language(args.language) + return args + + def _parse_opt(self, opts): + config = {} + if not opts: + return config + for s in opts: + s = s.strip() + k, v = s.split('=') + config[k] = yaml.load(v, Loader=yaml.Loader) + return config + + def _set_language(self, type): + lang = type[0] + assert (type), "please use -l or --language to choose language type" + assert( + lang in support_list.keys() or lang in multi_lang + ),"the sub_keys(-l or --language) can only be one of support list: \n{},\nbut get: {}, " \ + "please check your running command".format(multi_lang, type) + if lang in latin_lang: + lang = "latin" + elif lang in arabic_lang: + lang = "arabic" + elif lang in cyrillic_lang: + lang = "cyrillic" + elif lang in devanagari_lang: + lang = "devanagari" + global_config['Global'][ + 'character_dict_path'] = 'ppocr/utils/dict/{}_dict.txt'.format(lang) + global_config['Global'][ + 'output_dir'] = './output/rec_{}_lite'.format(lang) + global_config['Train']['dataset'][ + 'label_file_list'] = ["train_data/{}_train.txt".format(lang)] + global_config['Eval']['dataset'][ + 'label_file_list'] = ["train_data/{}_val.txt".format(lang)] + global_config['Global']['character_type'] = lang + assert ( + os.path.isfile( + os.path.join(project_path, global_config['Global'][ + 'character_dict_path'])) + ), "Loss default dictionary file {}_dict.txt.You can download it from \ +https://github.com/PaddlePaddle/PaddleOCR/tree/dygraph/ppocr/utils/dict/".format( + lang) + return lang + + +def merge_config(config): + """ + Merge config into global config. + Args: + config (dict): Config to be merged. + Returns: global config + """ + for key, value in config.items(): + if "." not in key: + if isinstance(value, dict) and key in global_config: + global_config[key].update(value) + else: + global_config[key] = value + else: + sub_keys = key.split('.') + assert ( + sub_keys[0] in global_config + ), "the sub_keys can only be one of global_config: {}, but get: {}, please check your running command".format( + global_config.keys(), sub_keys[0]) + cur = global_config[sub_keys[0]] + for idx, sub_key in enumerate(sub_keys[1:]): + if idx == len(sub_keys) - 2: + cur[sub_key] = value + else: + cur = cur[sub_key] + + +def loss_file(path): + assert ( + os.path.exists(path) + ), "There is no such file:{},Please do not forget to put in the specified file".format( + path) + + +if __name__ == '__main__': + FLAGS = ArgsParser().parse_args() + merge_config(FLAGS.opt) + save_file_path = 'rec_{}_lite_train.yml'.format(FLAGS.language) + if os.path.isfile(save_file_path): + os.remove(save_file_path) + + if FLAGS.train: + global_config['Train']['dataset']['label_file_list'] = [FLAGS.train] + train_label_path = os.path.join(project_path, FLAGS.train) + loss_file(train_label_path) + if FLAGS.val: + global_config['Eval']['dataset']['label_file_list'] = [FLAGS.val] + eval_label_path = os.path.join(project_path, FLAGS.val) + loss_file(eval_label_path) + if FLAGS.dict: + global_config['Global']['character_dict_path'] = FLAGS.dict + dict_path = os.path.join(project_path, FLAGS.dict) + loss_file(dict_path) + if FLAGS.data_dir: + global_config['Eval']['dataset']['data_dir'] = FLAGS.data_dir + global_config['Train']['dataset']['data_dir'] = FLAGS.data_dir + data_dir = os.path.join(project_path, FLAGS.data_dir) + loss_file(data_dir) + + with open(save_file_path, 'w') as f: + yaml.dump( + dict(global_config), f, default_flow_style=False, sort_keys=False) + logging.info("Project path is :{}".format(project_path)) + logging.info("Train list path set to :{}".format(global_config['Train'][ + 'dataset']['label_file_list'][0])) + logging.info("Eval list path set to :{}".format(global_config['Eval'][ + 'dataset']['label_file_list'][0])) + logging.info("Dataset root path set to :{}".format(global_config['Eval'][ + 'dataset']['data_dir'])) + logging.info("Dict path set to :{}".format(global_config['Global'][ + 'character_dict_path'])) + logging.info("Config file set to :configs/rec/multi_language/{}". + format(save_file_path)) diff --git a/configs/rec/multi_language/rec_arabic_lite_train.yml b/configs/rec/multi_language/rec_arabic_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..27e9ef6869b47d66347ce07856f73e73ec9f7fd7 --- /dev/null +++ b/configs/rec/multi_language/rec_arabic_lite_train.yml @@ -0,0 +1,110 @@ +Global: + use_gpu: true + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_arabic_lite + save_epoch_step: 3 + eval_batch_step: + - 0 + - 2000 + cal_metric_during_train: true + pretrained_model: null + checkpoints: null + save_inference_dir: null + use_visualdl: false + infer_img: null + character_dict_path: ppocr/utils/dict/arabic_dict.txt + max_text_length: 25 + infer_mode: false + use_space_char: true +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: L2 + factor: 1.0e-05 +Architecture: + model_type: rec + algorithm: CRNN + Transform: null + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: + - 1 + - 2 + - 2 + - 2 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 1.0e-05 +Loss: + name: CTCLoss +PostProcess: + name: CTCLabelDecode +Metric: + name: RecMetric + main_indicator: acc +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: + - train_data/arabic_train.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - RecAug: null + - CTCLabelEncode: null + - RecResizeImg: + image_shape: + - 3 + - 32 + - 320 + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: true + batch_size_per_card: 256 + drop_last: true + num_workers: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: + - train_data/arabic_val.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - CTCLabelEncode: null + - RecResizeImg: + image_shape: + - 3 + - 32 + - 320 + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: false + drop_last: false + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/rec_cyrillic_lite_train.yml b/configs/rec/multi_language/rec_cyrillic_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..2626bcd6ad7ea2479d279ff9712f3f9bfd5fef64 --- /dev/null +++ b/configs/rec/multi_language/rec_cyrillic_lite_train.yml @@ -0,0 +1,110 @@ +Global: + use_gpu: true + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_cyrillic_lite + save_epoch_step: 3 + eval_batch_step: + - 0 + - 2000 + cal_metric_during_train: true + pretrained_model: null + checkpoints: null + save_inference_dir: null + use_visualdl: false + infer_img: null + character_dict_path: ppocr/utils/dict/cyrillic_dict.txt + max_text_length: 25 + infer_mode: false + use_space_char: true +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: L2 + factor: 1.0e-05 +Architecture: + model_type: rec + algorithm: CRNN + Transform: null + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: + - 1 + - 2 + - 2 + - 2 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 1.0e-05 +Loss: + name: CTCLoss +PostProcess: + name: CTCLabelDecode +Metric: + name: RecMetric + main_indicator: acc +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: + - train_data/cyrillic_train.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - RecAug: null + - CTCLabelEncode: null + - RecResizeImg: + image_shape: + - 3 + - 32 + - 320 + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: true + batch_size_per_card: 256 + drop_last: true + num_workers: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: + - train_data/cyrillic_val.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - CTCLabelEncode: null + - RecResizeImg: + image_shape: + - 3 + - 32 + - 320 + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: false + drop_last: false + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/rec_devanagari_lite_train.yml b/configs/rec/multi_language/rec_devanagari_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..3b445f8828a70b94702b492b2bcb5d6f467219e2 --- /dev/null +++ b/configs/rec/multi_language/rec_devanagari_lite_train.yml @@ -0,0 +1,110 @@ +Global: + use_gpu: true + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_devanagari_lite + save_epoch_step: 3 + eval_batch_step: + - 0 + - 2000 + cal_metric_during_train: true + pretrained_model: null + checkpoints: null + save_inference_dir: null + use_visualdl: false + infer_img: null + character_dict_path: ppocr/utils/dict/devanagari_dict.txt + max_text_length: 25 + infer_mode: false + use_space_char: true +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: L2 + factor: 1.0e-05 +Architecture: + model_type: rec + algorithm: CRNN + Transform: null + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: + - 1 + - 2 + - 2 + - 2 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 1.0e-05 +Loss: + name: CTCLoss +PostProcess: + name: CTCLabelDecode +Metric: + name: RecMetric + main_indicator: acc +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: + - train_data/devanagari_train.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - RecAug: null + - CTCLabelEncode: null + - RecResizeImg: + image_shape: + - 3 + - 32 + - 320 + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: true + batch_size_per_card: 256 + drop_last: true + num_workers: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: + - train_data/devanagari_val.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - CTCLabelEncode: null + - RecResizeImg: + image_shape: + - 3 + - 32 + - 320 + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: false + drop_last: false + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/rec_en_number_lite_train.yml b/configs/rec/multi_language/rec_en_number_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..c29f736b104d902a8cc0d0ef18fab23fb94ecda2 --- /dev/null +++ b/configs/rec/multi_language/rec_en_number_lite_train.yml @@ -0,0 +1,101 @@ +Global: + use_gpu: True + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_en_number_lite + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + # if pretrained_model is saved in static mode, load_static_weights must set to True + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + # for data or label process + character_dict_path: ppocr/utils/en_dict.txt + max_text_length: 25 + infer_mode: False + use_space_char: True + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0.00001 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: [1, 2, 2, 2] + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 0.00001 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/train_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RecAug: + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/eval_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/rec_french_lite_train.yml b/configs/rec/multi_language/rec_french_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..7a78e04c5bca45f031c5e5fd5bb3b3e619733e30 --- /dev/null +++ b/configs/rec/multi_language/rec_french_lite_train.yml @@ -0,0 +1,101 @@ +Global: + use_gpu: True + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_french_lite + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + # if pretrained_model is saved in static mode, load_static_weights must set to True + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + # for data or label process + character_dict_path: ppocr/utils/dict/french_dict.txt + max_text_length: 25 + infer_mode: False + use_space_char: False + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0.00001 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: [1, 2, 2, 2] + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 0.00001 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/train_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RecAug: + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/eval_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/rec_german_lite_train.yml b/configs/rec/multi_language/rec_german_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..fe27f6bcf1c54fff83f807b9e1051d7a6f5fc3e9 --- /dev/null +++ b/configs/rec/multi_language/rec_german_lite_train.yml @@ -0,0 +1,101 @@ +Global: + use_gpu: True + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_german_lite + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + # if pretrained_model is saved in static mode, load_static_weights must set to True + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + # for data or label process + character_dict_path: ppocr/utils/dict/german_dict.txt + max_text_length: 25 + infer_mode: False + use_space_char: False + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0.00001 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: [1, 2, 2, 2] + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 0.00001 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/train_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RecAug: + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/eval_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/rec_japan_lite_train.yml b/configs/rec/multi_language/rec_japan_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..30776ded03bfa0d335d8681814a3f9c333f0b391 --- /dev/null +++ b/configs/rec/multi_language/rec_japan_lite_train.yml @@ -0,0 +1,101 @@ +Global: + use_gpu: True + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_japan_lite + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + # if pretrained_model is saved in static mode, load_static_weights must set to True + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + # for data or label process + character_dict_path: ppocr/utils/dict/japan_dict.txt + max_text_length: 25 + infer_mode: False + use_space_char: False + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0.00001 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: [1, 2, 2, 2] + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 0.00001 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/train_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RecAug: + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/eval_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/rec_korean_lite_train.yml b/configs/rec/multi_language/rec_korean_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..1131dc433587414174791368aca96a3a2be73398 --- /dev/null +++ b/configs/rec/multi_language/rec_korean_lite_train.yml @@ -0,0 +1,101 @@ +Global: + use_gpu: True + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_korean_lite + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + # if pretrained_model is saved in static mode, load_static_weights must set to True + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + # for data or label process + character_dict_path: ppocr/utils/dict/korean_dict.txt + max_text_length: 25 + infer_mode: False + use_space_char: False + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0.00001 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: [1, 2, 2, 2] + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 0.00001 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/train_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RecAug: + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ + label_file_list: ["./train_data/eval_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/rec_latin_lite_train.yml b/configs/rec/multi_language/rec_latin_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..f7cdcfbb79867077894327fe02495c589b6c854d --- /dev/null +++ b/configs/rec/multi_language/rec_latin_lite_train.yml @@ -0,0 +1,110 @@ +Global: + use_gpu: true + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_latin_lite + save_epoch_step: 3 + eval_batch_step: + - 0 + - 2000 + cal_metric_during_train: true + pretrained_model: null + checkpoints: null + save_inference_dir: null + use_visualdl: false + infer_img: null + character_dict_path: ppocr/utils/dict/latin_dict.txt + max_text_length: 25 + infer_mode: false + use_space_char: true +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: L2 + factor: 1.0e-05 +Architecture: + model_type: rec + algorithm: CRNN + Transform: null + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: + - 1 + - 2 + - 2 + - 2 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 1.0e-05 +Loss: + name: CTCLoss +PostProcess: + name: CTCLabelDecode +Metric: + name: RecMetric + main_indicator: acc +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: + - train_data/latin_train.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - RecAug: null + - CTCLabelEncode: null + - RecResizeImg: + image_shape: + - 3 + - 32 + - 320 + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: true + batch_size_per_card: 256 + drop_last: true + num_workers: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: + - train_data/latin_val.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - CTCLabelEncode: null + - RecResizeImg: + image_shape: + - 3 + - 32 + - 320 + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: false + drop_last: false + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/multi_language/rec_multi_language_lite_train.yml b/configs/rec/multi_language/rec_multi_language_lite_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..bc2cd607486bad2ce38d0c7c05f1413503d8205e --- /dev/null +++ b/configs/rec/multi_language/rec_multi_language_lite_train.yml @@ -0,0 +1,103 @@ +Global: + use_gpu: True + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec_multi_language_lite + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + # if pretrained_model is saved in static mode, load_static_weights must set to True + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + # for data or label process + character_dict_path: + # Set the language of training, if set, select the default dictionary file + character_type: + max_text_length: 25 + infer_mode: False + use_space_char: True + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Cosine + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0.00001 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: small + small_stride: [1, 2, 2, 2] + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 48 + Head: + name: CTCHead + fc_decay: 0.00001 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: ["./train_data/train_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RecAug: + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: train_data/ + label_file_list: ["./train_data/val_list.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 320] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/rec_d28_can.yml b/configs/rec/rec_d28_can.yml new file mode 100644 index 0000000000000000000000000000000000000000..dca72173009a5e77b86a4f6b2eacf3f2c7e6922f --- /dev/null +++ b/configs/rec/rec_d28_can.yml @@ -0,0 +1,122 @@ +Global: + use_gpu: True + epoch_num: 240 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/can/ + save_epoch_step: 1 + # evaluation is run every 1105 iterations (1 epoch)(batch_size = 8) + eval_batch_step: [0, 1105] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/datasets/crohme_demo/hme_00.jpg + # for data or label process + character_dict_path: ppocr/utils/dict/latex_symbol_dict.txt + max_text_length: 36 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_can.txt + +Optimizer: + name: Momentum + momentum: 0.9 + clip_norm_global: 100.0 + lr: + name: TwoStepCosine + learning_rate: 0.01 + warmup_epoch: 1 + weight_decay: 0.0001 + +Architecture: + model_type: rec + algorithm: CAN + in_channels: 1 + Transform: + Backbone: + name: DenseNet + growthRate: 24 + reduction: 0.5 + bottleneck: True + use_dropout: True + input_channel: 1 + Head: + name: CANHead + in_channel: 684 + out_channel: 111 + max_text_length: 36 + ratio: 16 + attdecoder: + is_train: True + input_size: 256 + hidden_size: 256 + encoder_out_channel: 684 + dropout: True + dropout_ratio: 0.5 + word_num: 111 + counting_decoder_out_channel: 111 + attention: + attention_dim: 512 + word_conv_kernel: 1 + +Loss: + name: CANLoss + +PostProcess: + name: CANLabelDecode + +Metric: + name: CANMetric + main_indicator: exp_rate + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/CROHME/training/images/ + label_file_list: ["./train_data/CROHME/training/labels.txt"] + transforms: + - DecodeImage: + channel_first: False + - NormalizeImage: + mean: [0,0,0] + std: [1,1,1] + order: 'hwc' + - GrayImageChannelFormat: + inverse: True + - CANLabelEncode: + lower: False + - KeepKeys: + keep_keys: ['image', 'label'] + loader: + shuffle: True + batch_size_per_card: 8 + drop_last: False + num_workers: 4 + collate_fn: DyMaskCollator + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/CROHME/evaluation/images/ + label_file_list: ["./train_data/CROHME/evaluation/labels.txt"] + transforms: + - DecodeImage: + channel_first: False + - NormalizeImage: + mean: [0,0,0] + std: [1,1,1] + order: 'hwc' + - GrayImageChannelFormat: + inverse: True + - CANLabelEncode: + lower: False + - KeepKeys: + keep_keys: ['image', 'label'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 1 + num_workers: 4 + collate_fn: DyMaskCollator diff --git a/configs/rec/rec_efficientb3_fpn_pren.yml b/configs/rec/rec_efficientb3_fpn_pren.yml new file mode 100644 index 0000000000000000000000000000000000000000..2043c84ef414113a18bee2a68efcaba47e397a6b --- /dev/null +++ b/configs/rec/rec_efficientb3_fpn_pren.yml @@ -0,0 +1,92 @@ +Global: + use_gpu: True + epoch_num: 8 + log_smooth_window: 20 + print_batch_step: 5 + output_dir: ./output/rec/pren_new + save_epoch_step: 3 + # evaluation is run every 2000 iterations after the 4000th iteration + eval_batch_step: [4000, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words/ch/word_1.jpg + # for data or label process + character_dict_path: + max_text_length: &max_text_length 25 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_pren.txt + +Optimizer: + name: Adadelta + lr: + name: Piecewise + decay_epochs: [2, 5, 7] + values: [0.5, 0.1, 0.01, 0.001] + +Architecture: + model_type: rec + algorithm: PREN + in_channels: 3 + Backbone: + name: EfficientNetb3_PREN + Neck: + name: PRENFPN + n_r: 5 + d_model: 384 + max_len: *max_text_length + dropout: 0.1 + Head: + name: PRENHead + +Loss: + name: PRENLoss + +PostProcess: + name: PRENLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: + img_mode: BGR + channel_first: False + - PRENLabelEncode: + - RecAug: + - PRENResizeImg: + image_shape: [64, 256] # h,w + - KeepKeys: + keep_keys: ['image', 'label'] + loader: + shuffle: True + batch_size_per_card: 128 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: + img_mode: BGR + channel_first: False + - PRENLabelEncode: + - PRENResizeImg: + image_shape: [64, 256] # h,w + - KeepKeys: + keep_keys: ['image', 'label'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 64 + num_workers: 8 diff --git a/configs/rec/rec_icdar15_train.yml b/configs/rec/rec_icdar15_train.yml new file mode 100644 index 0000000000000000000000000000000000000000..ab7505c520e73dbeb573def5773fa1740f7f6866 --- /dev/null +++ b/configs/rec/rec_icdar15_train.yml @@ -0,0 +1,97 @@ +Global: + use_gpu: true + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/ic15/ + save_epoch_step: 3 + # evaluation is run every 2000 iterations + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: ./ + use_visualdl: False + infer_img: doc/imgs_words_en/word_10.png + # for data or label process + character_dict_path: ppocr/utils/en_dict.txt + max_text_length: 25 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_ic15.txt + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + learning_rate: 0.0005 + regularizer: + name: 'L2' + factor: 0 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 96 + Head: + name: CTCHead + fc_decay: 0 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ic15_data/ + label_file_list: ["./train_data/ic15_data/rec_gt_train.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ic15_data + label_file_list: ["./train_data/ic15_data/rec_gt_test.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 4 diff --git a/configs/rec/rec_mtb_nrtr.yml b/configs/rec/rec_mtb_nrtr.yml new file mode 100644 index 0000000000000000000000000000000000000000..e622d514ed9cc15967b5c6ce3090c314d49944c6 --- /dev/null +++ b/configs/rec/rec_mtb_nrtr.yml @@ -0,0 +1,99 @@ +Global: + device: gpu + epoch_num: 21 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/nrtr + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path ppocr/utils/EN_symbol_dict.txt + max_text_length: &max_text_length 25 + use_space_char: &use_space_char False + +Export: + export_dir: + export_shape: [ 1, 1, 32, 100 ] + dynamic_axes: [] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 0 + +LRScheduler: + name: CosineAnnealingLR + warmup_epoch: 2 + + +Architecture: + model_type: rec + algorithm: NRTR + in_channels: 1 + Transform: + Backbone: + name: MTB + cnn_num: 2 + Head: + name: Transformer + d_model: 512 + num_encoder_layers: 6 + beam_size: -1 # When Beam size is greater than 0, it means to use beam search when evaluation. + + +Loss: + name: CELoss + smoothing: True + +PostProcess: + name: NRTRLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - NRTRLabelEncode: # Class handling label + - GrayRecResizeImg: + image_shape: [100, 32] # W H + resize_type: PIL # PIL or OpenCV + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 512 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - NRTRLabelEncode: # Class handling label + - GrayRecResizeImg: + image_shape: [100, 32] # W H + resize_type: PIL # PIL or OpenCV + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 4 diff --git a/configs/rec/rec_mv3_none_bilstm_ctc.yml b/configs/rec/rec_mv3_none_bilstm_ctc.yml new file mode 100644 index 0000000000000000000000000000000000000000..83827adcba1a4dd2d3e55712c8b8567e4447adf7 --- /dev/null +++ b/configs/rec/rec_mv3_none_bilstm_ctc.yml @@ -0,0 +1,96 @@ +Global: + device: gpu + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/rec_mv3_none_bilstm_ctc + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path + max_text_length: &max_text_length 25 + use_space_char: &use_space_char False + + +Export: + export_dir: + export_shape: [ 1, 3, 32, 100 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 0.0 + +LRScheduler: + name: ConstLR + warmup_epoch: 0 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 96 + Head: + name: CTCHead + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 4 diff --git a/configs/rec/rec_mv3_none_none_ctc.yml b/configs/rec/rec_mv3_none_none_ctc.yml new file mode 100644 index 0000000000000000000000000000000000000000..a9b0a87c4b4bcdcc1aabbf604103e86fbb7ec0b3 --- /dev/null +++ b/configs/rec/rec_mv3_none_none_ctc.yml @@ -0,0 +1,96 @@ +Global: + device: gpu + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/mv3_none_none_ctc + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path + max_text_length: &max_text_length 25 + use_space_char: &use_space_char False + +Export: + export_dir: + export_shape: [ 1, 3, 32, 100 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 0 + +LRScheduler: + name: ConstLR + warmup_epoch: 0 + + +Architecture: + model_type: rec + algorithm: Rosetta + Transform: + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + Neck: + name: SequenceEncoder + encoder_type: reshape + Head: + name: CTCHead + fc_decay: 0.0004 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/rec_mv3_tps_bilstm_att.yml b/configs/rec/rec_mv3_tps_bilstm_att.yml new file mode 100644 index 0000000000000000000000000000000000000000..fe6ac5d47a7c340ced7c85f18f3e335696f2550d --- /dev/null +++ b/configs/rec/rec_mv3_tps_bilstm_att.yml @@ -0,0 +1,102 @@ +Global: + device: gpu + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/rec_mv3_tps_bilstm_att + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path + max_text_length: &max_text_length 25 + use_space_char: &use_space_char False + + +Export: + export_dir: + export_shape: [ 1, 3, 32, 100 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 0.00001 + +LRScheduler: + name: ConstLR + warmup_epoch: 0 + +Architecture: + model_type: rec + algorithm: RARE + Transform: + name: TPS + num_fiducial: 20 + loc_lr: 0.1 + model_name: small + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 96 + Head: + name: AttentionHead + hidden_size: 96 + + +Loss: + name: AttentionLoss + +PostProcess: + name: AttnLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - AttnLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - AttnLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 1 diff --git a/configs/rec/rec_mv3_tps_bilstm_ctc.yml b/configs/rec/rec_mv3_tps_bilstm_ctc.yml new file mode 100644 index 0000000000000000000000000000000000000000..3358466bc5b46ceb322ac4747081af6651a6b7dd --- /dev/null +++ b/configs/rec/rec_mv3_tps_bilstm_ctc.yml @@ -0,0 +1,99 @@ +Global: + device: gpu + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/rec_mv3_tps_bilstm_ctc + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path + max_text_length: &max_text_length 25 + use_space_char: &use_space_char False + +Export: + export_dir: + export_shape: [ 1, 3, 32, 100 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 0.0 + +LRScheduler: + name: ConstLR + warmup_epoch: 0 + +Architecture: + model_type: rec + algorithm: STARNet + Transform: + name: TPS + num_fiducial: 20 + loc_lr: 0.1 + model_name: small + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 96 + Head: + name: CTCHead + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 4 diff --git a/configs/rec/rec_r31_robustscanner.yml b/configs/rec/rec_r31_robustscanner.yml new file mode 100644 index 0000000000000000000000000000000000000000..157734a1c2f10be590f8b8c37cd00f2601434228 --- /dev/null +++ b/configs/rec/rec_r31_robustscanner.yml @@ -0,0 +1,107 @@ +Global: + use_gpu: true + epoch_num: 5 + log_smooth_window: 20 + print_batch_step: 20 + output_dir: ./output/rec/rec_r31_robustscanner/ + save_epoch_step: 1 + # evaluation is run every 2000 iterations + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words_en/word_10.png + # for data or label process + character_dict_path: ppocr/utils/dict90.txt + max_text_length: &max_text_length 40 + infer_mode: False + use_space_char: False + rm_symbol: True + save_res_path: ./output/rec/predicts_robustscanner.txt + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Piecewise + decay_epochs: [3, 4] + values: [0.001, 0.0001, 0.00001] + regularizer: + name: 'L2' + factor: 0 + +Architecture: + model_type: rec + algorithm: RobustScanner + Transform: + Backbone: + name: ResNet31 + init_type: KaimingNormal + Head: + name: RobustScannerHead + enc_outchannles: 128 + hybrid_dec_rnn_layers: 2 + hybrid_dec_dropout: 0 + position_dec_rnn_layers: 2 + start_idx: 91 + mask: True + padding_idx: 92 + encode_value: False + max_text_length: *max_text_length + +Loss: + name: SARLoss + +PostProcess: + name: SARLabelDecode + +Metric: + name: RecMetric + is_filter: True + + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SARLabelEncode: # Class handling label + - RobustScannerRecResizeImg: + image_shape: [3, 48, 48, 160] # h:48 w:[48,160] + width_downsample_ratio: 0.25 + max_text_length: *max_text_length + - KeepKeys: + keep_keys: ['image', 'label', 'valid_ratio', 'word_positons'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 64 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/evaluation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SARLabelEncode: # Class handling label + - RobustScannerRecResizeImg: + image_shape: [3, 48, 48, 160] + max_text_length: *max_text_length + width_downsample_ratio: 0.25 + - KeepKeys: + keep_keys: ['image', 'label', 'valid_ratio', 'word_positons'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 64 + num_workers: 4 + diff --git a/configs/rec/rec_r31_sar.yml b/configs/rec/rec_r31_sar.yml new file mode 100644 index 0000000000000000000000000000000000000000..c0e5ed031267906ebdd0297f3061c6a5ed6b4cba --- /dev/null +++ b/configs/rec/rec_r31_sar.yml @@ -0,0 +1,95 @@ +Global: + device: gpu + epoch_num: 5 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/sar + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path ppocr/utils/dict90.txt + max_text_length: &max_text_length 30 + use_space_char: &use_space_char False + rm_symbol: &rm_symbol True + +Export: + export_dir: + export_shape: [ 1, 3, 48, 160 ] + dynamic_axes: [] + +Optimizer: + name: Adam + lr: 0.001 + weight_decay: 0 + +LRScheduler: + name: MultiStepLR + milestones: [3,4] + warmup_epoch: 0 + +Architecture: + model_type: rec + algorithm: SAR + Transform: + Backbone: + name: ResNet31 + Head: + name: SARHead + +Loss: + name: SARLoss + +PostProcess: + name: SARLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + rm_symbol: *rm_symbol + +Metric: + name: RecMetric + + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SARLabelEncode: # Class handling label + - SARRecResizeImg: + image_shape: [3, 48, 48, 160] # h:48 w:[48,160] + width_downsample_ratio: 0.25 + - KeepKeys: + keep_keys: ['image', 'label', 'valid_ratio'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 64 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SARLabelEncode: # Class handling label + - SARRecResizeImg: + image_shape: [3, 48, 48, 160] + width_downsample_ratio: 0.25 + - KeepKeys: + keep_keys: ['image', 'label', 'valid_ratio'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 64 + num_workers: 4 + diff --git a/configs/rec/rec_r32_gaspin_bilstm_att.yml b/configs/rec/rec_r32_gaspin_bilstm_att.yml new file mode 100644 index 0000000000000000000000000000000000000000..5ebdea136687f205f485ad9db08ddca9a482331d --- /dev/null +++ b/configs/rec/rec_r32_gaspin_bilstm_att.yml @@ -0,0 +1,116 @@ +Global: + use_gpu: True + epoch_num: 6 + log_smooth_window: 50 + print_batch_step: 50 + output_dir: ./output/rec/rec_r32_gaspin_bilstm_att/ + save_epoch_step: 3 + # evaluation is run every 2000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words_en/word_10.png + # for data or label process + character_dict_path: ./ppocr/utils/dict/spin_dict.txt + max_text_length: 25 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_r32_gaspin_bilstm_att.txt + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + lr: + name: Piecewise + decay_epochs: [3, 4, 5] + values: [0.001, 0.0003, 0.00009, 0.000027] + clip_norm: 5 + +Architecture: + model_type: rec + algorithm: SPIN + in_channels: 1 + Transform: + name: GA_SPIN + offsets: True + default_type: 6 + loc_lr: 0.1 + stn: True + Backbone: + name: ResNet32 + out_channels: 512 + Neck: + name: SequenceEncoder + encoder_type: cascadernn + hidden_size: 256 + out_channels: [256, 512] + with_linear: True + Head: + name: SPINAttentionHead + hidden_size: 256 + + +Loss: + name: SPINAttentionLoss + ignore_index: 0 + +PostProcess: + name: SPINLabelDecode + use_space_char: False + + +Metric: + name: RecMetric + main_indicator: acc + is_filter: True + +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ic15_data/ + label_file_list: ["./train_data/ic15_data/rec_gt_train.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SPINLabelEncode: # Class handling label + - SPINRecResizeImg: + image_shape: [100, 32] + interpolation : 2 + mean: [127.5] + std: [127.5] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 8 + drop_last: True + num_workers: 4 + +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data/ic15_data + label_file_list: ["./train_data/ic15_data/rec_gt_test.txt"] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SPINLabelEncode: # Class handling label + - SPINRecResizeImg: + image_shape: [100, 32] + interpolation : 2 + mean: [127.5] + std: [127.5] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 8 + num_workers: 2 diff --git a/configs/rec/rec_r34_vd_none_bilstm_ctc.yml b/configs/rec/rec_r34_vd_none_bilstm_ctc.yml new file mode 100644 index 0000000000000000000000000000000000000000..74a3aee10b995e3ff687faeac000f31910d3fa79 --- /dev/null +++ b/configs/rec/rec_r34_vd_none_bilstm_ctc.yml @@ -0,0 +1,95 @@ +Global: + device: gpu + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/r34_vd_none_bilstm_ctc + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path + max_text_length: &max_text_length 25 + use_space_char: &use_space_char False + +Export: + export_dir: + export_shape: [ 1, 3, 32, 100 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 0.0 + +LRScheduler: + name: ConstLR + warmup_epoch: 0 + +Architecture: + model_type: rec + algorithm: CRNN + Transform: + Backbone: + name: ResNet + layers: 34 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 256 + Head: + name: CTCHead + fc_decay: 0 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 4 diff --git a/configs/rec/rec_r34_vd_none_none_ctc.yml b/configs/rec/rec_r34_vd_none_none_ctc.yml new file mode 100644 index 0000000000000000000000000000000000000000..ea20f63759fae7df39555489e6109ae194a902b8 --- /dev/null +++ b/configs/rec/rec_r34_vd_none_none_ctc.yml @@ -0,0 +1,94 @@ +Global: + device: gpu + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/r34_vd_none_none_ctc + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path + max_text_length: &max_text_length 25 + use_space_char: &use_space_char False + + +Export: + export_dir: + export_shape: [ 1, 3, 32, 100 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 0.0 + +LRScheduler: + name: ConstLR + warmup_epoch: 0 + +Architecture: + model_type: rec + algorithm: Rosetta + Backbone: + name: ResNet + layers: 34 + Neck: + name: SequenceEncoder + encoder_type: reshape + Head: + name: CTCHead + fc_decay: 0.0004 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 4 diff --git a/configs/rec/rec_r34_vd_tps_bilstm_att.yml b/configs/rec/rec_r34_vd_tps_bilstm_att.yml new file mode 100644 index 0000000000000000000000000000000000000000..c7303692a572e409b600622a563c87a7a93c3ddd --- /dev/null +++ b/configs/rec/rec_r34_vd_tps_bilstm_att.yml @@ -0,0 +1,100 @@ +Global: + device: gpu + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/r34_vd_tps_bilstm_att + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path + max_text_length: &max_text_length 25 + use_space_char: &use_space_char False + +Export: + export_dir: + export_shape: [ 1, 3, 32, 100 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 0.0 + +LRScheduler: + name: ConstLR + warmup_epoch: 0 + +Architecture: + model_type: rec + algorithm: RARE + Transform: + name: TPS + num_fiducial: 20 + loc_lr: 0.1 + model_name: large + Backbone: + name: ResNet + layers: 34 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 256 #96 + Head: + name: AttentionHead # AttentionHead + hidden_size: 256 # + l2_decay: 0.00001 + +Loss: + name: AttentionLoss + +PostProcess: + name: AttnLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - AttnLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - AttnLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/rec_r34_vd_tps_bilstm_ctc.yml b/configs/rec/rec_r34_vd_tps_bilstm_ctc.yml new file mode 100644 index 0000000000000000000000000000000000000000..4c860ce23abbbed8faf568a660b182aa9f34579e --- /dev/null +++ b/configs/rec/rec_r34_vd_tps_bilstm_ctc.yml @@ -0,0 +1,99 @@ +Global: + device: gpu + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/r34_vd_tps_bilstm_ctc + eval_epoch_step: [0, 1] + cal_metric_during_train: true + pretrained_model: + checkpoints: + use_tensorboard: false + infer_mode: false + infer_img: doc/imgs_words/en/word_1.png + character_dict_path: &character_dict_path + max_text_length: &max_text_length 25 + use_space_char: &use_space_char False + +Export: + export_dir: + export_shape: [ 1, 3, 32, 100 ] + dynamic_axes: [ 0, 2, 3 ] + +Optimizer: + name: Adam + lr: 0.0005 + weight_decay: 0.0 + +LRScheduler: + name: ConstLR + warmup_epoch: 0 + +Architecture: + model_type: rec + algorithm: STARNet + Transform: + name: TPS + num_fiducial: 20 + loc_lr: 0.1 + model_name: large + Backbone: + name: ResNet + layers: 34 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 256 + Head: + name: CTCHead + fc_decay: 0 + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + character_dict_path: *character_dict_path + use_space_char: *use_space_char + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 4 diff --git a/configs/rec/rec_r45_abinet.yml b/configs/rec/rec_r45_abinet.yml new file mode 100644 index 0000000000000000000000000000000000000000..8b3452599f068517250e9a8f2b9c9d9e4b2ad9dc --- /dev/null +++ b/configs/rec/rec_r45_abinet.yml @@ -0,0 +1,100 @@ +Global: + use_gpu: True + epoch_num: 10 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/r45_abinet/ + save_epoch_step: 1 + # evaluation is run every 2000 iterations + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: ./pretrain_models/abinet_vl_pretrained + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words_en/word_10.png + # for data or label process + character_dict_path: + character_type: en + max_text_length: 25 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_abinet.txt + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.99 + clip_norm: 20.0 + lr: + name: Piecewise + decay_epochs: [6] + values: [0.0001, 0.00001] + regularizer: + name: 'L2' + factor: 0. + +Architecture: + model_type: rec + algorithm: ABINet + in_channels: 3 + Transform: + Backbone: + name: ResNet45 + Head: + name: ABINetHead + use_lang: True + iter_size: 3 + + +Loss: + name: CELoss + ignore_index: &ignore_index 100 # Must be greater than the number of character classes + +PostProcess: + name: ABINetLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - ABINetRecAug: + - ABINetLabelEncode: # Class handling label + ignore_index: *ignore_index + - ABINetRecResizeImg: + image_shape: [3, 32, 128] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 96 + drop_last: True + num_workers: 4 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/evaluation/ + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - ABINetLabelEncode: # Class handling label + ignore_index: *ignore_index + - ABINetRecResizeImg: + image_shape: [3, 32, 128] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 4 diff --git a/configs/rec/rec_r45_visionlan.yml b/configs/rec/rec_r45_visionlan.yml new file mode 100644 index 0000000000000000000000000000000000000000..6236138de229d76589aa5d8ff5110f5d0ebbee66 --- /dev/null +++ b/configs/rec/rec_r45_visionlan.yml @@ -0,0 +1,106 @@ +Global: + use_gpu: true + epoch_num: 8 + log_smooth_window: 200 + print_batch_step: 200 + output_dir: ./output/rec/r45_visionlan + save_epoch_step: 1 + # evaluation is run every 2000 iterations + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: True + infer_img: doc/imgs_words/en/word_2.png + # for data or label process + character_dict_path: + max_text_length: &max_text_length 25 + training_step: &training_step LA + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_visionlan.txt + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + clip_norm: 20.0 + group_lr: true + training_step: *training_step + lr: + name: Piecewise + decay_epochs: [6] + values: [0.0001, 0.00001] + regularizer: + name: 'L2' + factor: 0 + +Architecture: + model_type: rec + algorithm: VisionLAN + Transform: + Backbone: + name: ResNet45 + strides: [2, 2, 2, 1, 1] + Head: + name: VLHead + n_layers: 3 + n_position: 256 + n_dim: 512 + max_text_length: *max_text_length + training_step: *training_step + +Loss: + name: VLLoss + mode: *training_step + weight_res: 0.5 + weight_mas: 0.5 + +PostProcess: + name: VLLabelDecode + +Metric: + name: RecMetric + is_filter: true + + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - ABINetRecAug: + - VLLabelEncode: # Class handling label + - VLRecResizeImg: + image_shape: [3, 64, 256] + - KeepKeys: + keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 220 + drop_last: True + num_workers: 4 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: RGB + channel_first: False + - VLLabelEncode: # Class handling label + - VLRecResizeImg: + image_shape: [3, 64, 256] + - KeepKeys: + keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 64 + num_workers: 4 + diff --git a/configs/rec/rec_r50_fpn_srn.yml b/configs/rec/rec_r50_fpn_srn.yml new file mode 100644 index 0000000000000000000000000000000000000000..a18a32a1daf56fb7ad0b7cf199f240802f2c16e2 --- /dev/null +++ b/configs/rec/rec_r50_fpn_srn.yml @@ -0,0 +1,106 @@ +Global: + use_gpu: True + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 5 + output_dir: ./output/rec/srn_new + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 5000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words/ch/word_1.jpg + # for data or label process + character_dict_path: + max_text_length: 25 + num_heads: 8 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_srn.txt + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + clip_norm: 10.0 + lr: + learning_rate: 0.0001 + +Architecture: + model_type: rec + algorithm: SRN + in_channels: 1 + Transform: + Backbone: + name: ResNetFPN + Head: + name: SRNHead + max_text_length: 25 + num_heads: 8 + num_encoder_TUs: 2 + num_decoder_TUs: 4 + hidden_dims: 512 + +Loss: + name: SRNLoss + +PostProcess: + name: SRNLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SRNLabelEncode: # Class handling label + - SRNRecResizeImg: + image_shape: [1, 64, 256] + - KeepKeys: + keep_keys: ['image', + 'label', + 'length', + 'encoder_word_pos', + 'gsrm_word_pos', + 'gsrm_slf_attn_bias1', + 'gsrm_slf_attn_bias2'] # dataloader will return list in this order + loader: + shuffle: False + batch_size_per_card: 64 + drop_last: False + num_workers: 4 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SRNLabelEncode: # Class handling label + - SRNRecResizeImg: + image_shape: [1, 64, 256] + - KeepKeys: + keep_keys: ['image', + 'label', + 'length', + 'encoder_word_pos', + 'gsrm_word_pos', + 'gsrm_slf_attn_bias1', + 'gsrm_slf_attn_bias2'] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 32 + num_workers: 4 diff --git a/configs/rec/rec_resnet_rfl_att.yml b/configs/rec/rec_resnet_rfl_att.yml new file mode 100644 index 0000000000000000000000000000000000000000..ed316547e81e5a2c7aba44fa84b752d68ba6ca2a --- /dev/null +++ b/configs/rec/rec_resnet_rfl_att.yml @@ -0,0 +1,112 @@ +Global: + use_gpu: True + epoch_num: 6 + log_smooth_window: 20 + print_batch_step: 50 + output_dir: ./output/rec/rec_resnet_rfl_att/ + save_epoch_step: 1 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 5000] + cal_metric_during_train: True + pretrained_model: ./pretrain_models/rec_resnet_rfl_visual/best_accuracy.pdparams + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words_en/word_10.png + # for data or label process + character_dict_path: + max_text_length: 25 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/rec_resnet_rfl.txt + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + weight_decay: 0.0 + clip_norm_global: 5.0 + lr: + name: Piecewise + decay_epochs : [3, 4, 5] + values : [0.001, 0.0003, 0.00009, 0.000027] + +Architecture: + model_type: rec + algorithm: RFL + in_channels: 1 + Transform: + name: TPS + num_fiducial: 20 + loc_lr: 1.0 + model_name: large + Backbone: + name: ResNetRFL + use_cnt: True + use_seq: True + Neck: + name: RFAdaptor + use_v2s: True + use_s2v: True + Head: + name: RFLHead + in_channels: 512 + hidden_size: 256 + batch_max_legnth: 25 + out_channels: 38 + use_cnt: True + use_seq: True + +Loss: + name: RFLLoss + # ignore_index: 0 + +PostProcess: + name: RFLLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RFLLabelEncode: # Class handling label + - RFLRecResizeImg: + image_shape: [1, 32, 100] + padding: false + interpolation: 2 + - KeepKeys: + keep_keys: ['image', 'label', 'length', 'cnt_label'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 64 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RFLLabelEncode: # Class handling label + - RFLRecResizeImg: + image_shape: [1, 32, 100] + padding: false + interpolation: 2 + - KeepKeys: + keep_keys: ['image', 'label', 'length', 'cnt_label'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/rec_resnet_rfl_visual.yml b/configs/rec/rec_resnet_rfl_visual.yml new file mode 100644 index 0000000000000000000000000000000000000000..8c2dac94903f82ac954886943b179158e9f93425 --- /dev/null +++ b/configs/rec/rec_resnet_rfl_visual.yml @@ -0,0 +1,110 @@ +Global: + use_gpu: True + epoch_num: 6 + log_smooth_window: 20 + print_batch_step: 50 + output_dir: ./output/rec/rec_resnet_rfl_visual/ + save_epoch_step: 1 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 5000] + cal_metric_during_train: False + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words_en/word_10.png + # for data or label process + character_dict_path: + max_text_length: 25 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/rec_resnet_rfl_visual.txt + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + weight_decay: 0.0 + clip_norm_global: 5.0 + lr: + name: Piecewise + decay_epochs : [3, 4, 5] + values : [0.001, 0.0003, 0.00009, 0.000027] + +Architecture: + model_type: rec + algorithm: RFL + in_channels: 1 + Transform: + name: TPS + num_fiducial: 20 + loc_lr: 1.0 + model_name: large + Backbone: + name: ResNetRFL + use_cnt: True + use_seq: False + Neck: + name: RFAdaptor + use_v2s: False + use_s2v: False + Head: + name: RFLHead + in_channels: 512 + hidden_size: 256 + batch_max_legnth: 25 + out_channels: 38 + use_cnt: True + use_seq: False +Loss: + name: RFLLoss + +PostProcess: + name: RFLLabelDecode + +Metric: + name: CNTMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RFLLabelEncode: # Class handling label + - RFLRecResizeImg: + image_shape: [1, 32, 100] + padding: false + interpolation: 2 + - KeepKeys: + keep_keys: ['image', 'label', 'length', 'cnt_label'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 64 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/evaluation + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - RFLLabelEncode: # Class handling label + - RFLRecResizeImg: + image_shape: [1, 32, 100] + padding: false + interpolation: 2 + - KeepKeys: + keep_keys: ['image', 'label', 'length', 'cnt_label'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/rec_resnet_stn_bilstm_att.yml b/configs/rec/rec_resnet_stn_bilstm_att.yml new file mode 100644 index 0000000000000000000000000000000000000000..7afd8fe33f3e9639c931faa554dfcc0626f9e293 --- /dev/null +++ b/configs/rec/rec_resnet_stn_bilstm_att.yml @@ -0,0 +1,108 @@ +Global: + use_gpu: True + epoch_num: 6 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/seed + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words_en/word_10.png + # for data or label process + character_dict_path: ppocr/utils/EN_symbol_dict.txt + max_text_length: 100 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_seed.txt + + +Optimizer: + name: Adadelta + weight_deacy: 0.0 + momentum: 0.9 + lr: + name: Piecewise + decay_epochs: [4, 5] + values: [1.0, 0.1, 0.01] + regularizer: + name: 'L2' + factor: 2.0e-05 + + +Architecture: + model_type: rec + algorithm: SEED + Transform: + name: STN_ON + tps_inputsize: [32, 64] + tps_outputsize: [32, 100] + num_control_points: 20 + tps_margins: [0.05,0.05] + stn_activation: none + Backbone: + name: ResNet_ASTER + Head: + name: AsterHead # AttentionHead + sDim: 512 + attDim: 512 + max_len_labels: 100 + +Loss: + name: AsterLoss + +PostProcess: + name: SEEDLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + is_filter: True + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - Fasttext: + path: "./cc.en.300.bin" + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SEEDLabelEncode: # Class handling label + - RecResizeImg: + character_dict_path: + image_shape: [3, 64, 256] + padding: False + - KeepKeys: + keep_keys: ['image', 'label', 'length', 'fast_label'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 6 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/evaluation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SEEDLabelEncode: # Class handling label + - RecResizeImg: + character_dict_path: + image_shape: [3, 64, 256] + padding: False + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: True + batch_size_per_card: 256 + num_workers: 4 diff --git a/configs/rec/rec_satrn.yml b/configs/rec/rec_satrn.yml new file mode 100644 index 0000000000000000000000000000000000000000..473b2097ded3a7930abadc012f7bfa1642f31a7b --- /dev/null +++ b/configs/rec/rec_satrn.yml @@ -0,0 +1,115 @@ +Global: + use_gpu: true + epoch_num: 5 + log_smooth_window: 20 + print_batch_step: 50 + output_dir: ./output/rec/rec_satrn/ + save_epoch_step: 1 + # evaluation is run every 5000 iterations + eval_batch_step: [0, 5000] + cal_metric_during_train: False + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: + # for data or label process + character_dict_path: ppocr/utils/dict90.txt + max_text_length: 25 + infer_mode: False + use_space_char: False + rm_symbol: True + save_res_path: ./output/rec/predicts_satrn.txt + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: Piecewise + decay_epochs: [3, 4] + values: [0.0003, 0.00003, 0.000003] + regularizer: + name: 'L2' + factor: 0 + +Architecture: + model_type: rec + algorithm: SATRN + Backbone: + name: ShallowCNN + in_channels: 3 + hidden_dim: 256 + Head: + name: SATRNHead + enc_cfg: + n_layers: 6 + n_head: 8 + d_k: 32 + d_v: 32 + d_model: 256 + n_position: 100 + d_inner: 1024 + dropout: 0.1 + dec_cfg: + n_layers: 6 + d_embedding: 256 + n_head: 8 + d_model: 256 + d_inner: 1024 + d_k: 32 + d_v: 32 + max_seq_len: 25 + start_idx: 91 + +Loss: + name: SATRNLoss + +PostProcess: + name: SATRNLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SATRNLabelEncode: # Class handling label + - SVTRRecResizeImg: + image_shape: [3, 32, 100] + padding: False + - KeepKeys: + keep_keys: ['image', 'label', 'valid_ratio'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 128 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/evaluation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SATRNLabelEncode: # Class handling label + - SVTRRecResizeImg: + image_shape: [3, 32, 100] + padding: False + - KeepKeys: + keep_keys: ['image', 'label', 'valid_ratio'] # dataloader will return list in this order + + loader: + shuffle: False + drop_last: False + batch_size_per_card: 128 + num_workers: 4 + diff --git a/configs/rec/rec_svtrnet.yml b/configs/rec/rec_svtrnet.yml new file mode 100644 index 0000000000000000000000000000000000000000..2df87de33c53be0388d41c063efba4d35f16ea8f --- /dev/null +++ b/configs/rec/rec_svtrnet.yml @@ -0,0 +1,118 @@ +Global: + use_gpu: True + epoch_num: 20 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/svtr/ + save_epoch_step: 1 + # evaluation is run every 2000 iterations after the 0th iteration + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words_en/word_10.png + # for data or label process + character_dict_path: + character_type: en + max_text_length: 25 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_svtr_tiny.txt + d2s_train_image_shape: [3, 64, 256] + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.99 + epsilon: 1.e-8 + weight_decay: 0.05 + no_weight_decay_name: norm pos_embed + one_dim_param_no_weight_decay: True + lr: + name: Cosine + learning_rate: 0.0005 + warmup_epoch: 2 + +Architecture: + model_type: rec + algorithm: SVTR + Transform: + name: STN_ON + tps_inputsize: [32, 64] + tps_outputsize: [32, 100] + num_control_points: 20 + tps_margins: [0.05,0.05] + stn_activation: none + Backbone: + name: SVTRNet + img_size: [32, 100] + out_char_num: 25 # W//4 or W//8 or W/12 + out_channels: 192 + patch_merging: 'Conv' + embed_dim: [64, 128, 256] + depth: [3, 6, 3] + num_heads: [2, 4, 8] + mixer: ['Local','Local','Local','Local','Local','Local','Global','Global','Global','Global','Global','Global'] + local_mixer: [[7, 11], [7, 11], [7, 11]] + last_stage: True + prenorm: False + Neck: + name: SequenceEncoder + encoder_type: reshape + Head: + name: CTCHead + +Loss: + name: CTCLoss + +PostProcess: + name: CTCLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - SVTRRecAug: + aug_type: 0 # or 1 + - CTCLabelEncode: # Class handling label + - SVTRRecResizeImg: + image_shape: [3, 64, 256] + padding: False + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 512 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/evaluation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - CTCLabelEncode: # Class handling label + - SVTRRecResizeImg: + image_shape: [3, 64, 256] + padding: False + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 2 diff --git a/configs/rec/rec_svtrnet_ch.yml b/configs/rec/rec_svtrnet_ch.yml new file mode 100644 index 0000000000000000000000000000000000000000..a5bbad8e145af8c58558344cc9576c5c7bbdad44 --- /dev/null +++ b/configs/rec/rec_svtrnet_ch.yml @@ -0,0 +1,156 @@ +Global: + use_gpu: true + epoch_num: 100 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/svtr_ch_all/ + save_epoch_step: 10 + eval_batch_step: + - 0 + - 2000 + cal_metric_during_train: true + pretrained_model: null + checkpoints: null + save_inference_dir: null + use_visualdl: false + infer_img: doc/imgs_words/ch/word_1.jpg + character_dict_path: ppocr/utils/ppocr_keys_v1.txt + max_text_length: 25 + infer_mode: false + use_space_char: true + save_res_path: ./output/rec/predicts_svtr_tiny_ch_all.txt + d2s_train_image_shape: [3, 32, 320] +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.99 + epsilon: 1.0e-08 + weight_decay: 0.05 + no_weight_decay_name: norm pos_embed + one_dim_param_no_weight_decay: true + lr: + name: Cosine + learning_rate: 0.0005 + warmup_epoch: 2 +Architecture: + model_type: rec + algorithm: SVTR + Transform: null + Backbone: + name: SVTRNet + img_size: + - 32 + - 320 + out_char_num: 40 # W//4 or W//8 or W/12 + out_channels: 96 + patch_merging: Conv + embed_dim: + - 64 + - 128 + - 256 + depth: + - 3 + - 6 + - 3 + num_heads: + - 2 + - 4 + - 8 + mixer: + - Local + - Local + - Local + - Local + - Local + - Local + - Global + - Global + - Global + - Global + - Global + - Global + local_mixer: + - - 7 + - 11 + - - 7 + - 11 + - - 7 + - 11 + last_stage: true + prenorm: false + Neck: + name: SequenceEncoder + encoder_type: reshape + Head: + name: CTCHead +Loss: + name: CTCLoss +PostProcess: + name: CTCLabelDecode +Metric: + name: RecMetric + main_indicator: acc +Train: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: + - ./train_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: + - 32 + - 320 + - 3 + - RecAug: null + - CTCLabelEncode: null + - SVTRRecResizeImg: + image_shape: + - 3 + - 32 + - 320 + padding: true + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: true + batch_size_per_card: 256 + drop_last: true + num_workers: 8 +Eval: + dataset: + name: SimpleDataSet + data_dir: ./train_data + label_file_list: + - ./train_data/val_list.txt + transforms: + - DecodeImage: + img_mode: BGR + channel_first: false + - CTCLabelEncode: null + - SVTRRecResizeImg: + image_shape: + - 3 + - 32 + - 320 + padding: true + - KeepKeys: + keep_keys: + - image + - label + - length + loader: + shuffle: false + drop_last: false + batch_size_per_card: 256 + num_workers: 2 +profiler_options: null diff --git a/configs/rec/rec_vitstr_none_ce.yml b/configs/rec/rec_vitstr_none_ce.yml new file mode 100644 index 0000000000000000000000000000000000000000..345040f409a4c1bbd1eb67403b925a1009be6065 --- /dev/null +++ b/configs/rec/rec_vitstr_none_ce.yml @@ -0,0 +1,102 @@ +Global: + use_gpu: True + epoch_num: 20 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/rec/vitstr_none_ce/ + save_epoch_step: 1 + # evaluation is run every 2000 iterations after the 0th iteration# + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words_en/word_10.png + # for data or label process + character_dict_path: ppocr/utils/EN_symbol_dict.txt + max_text_length: 25 + infer_mode: False + use_space_char: False + save_res_path: ./output/rec/predicts_vitstr.txt + + +Optimizer: + name: Adadelta + epsilon: 1.e-8 + rho: 0.95 + clip_norm: 5.0 + lr: + learning_rate: 1.0 + +Architecture: + model_type: rec + algorithm: ViTSTR + in_channels: 1 + Transform: + Backbone: + name: ViTSTR + scale: tiny + Neck: + name: SequenceEncoder + encoder_type: reshape + Head: + name: CTCHead + +Loss: + name: CELoss + with_all: True + ignore_index: &ignore_index 0 # Must be zero or greater than the number of character classes + +PostProcess: + name: ViTSTRLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - ViTSTRLabelEncode: # Class handling label + ignore_index: *ignore_index + - GrayRecResizeImg: + image_shape: [224, 224] # W H + resize_type: PIL # PIL or OpenCV + inter_type: 'Image.BICUBIC' + scale: false + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 48 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/evaluation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - ViTSTRLabelEncode: # Class handling label + ignore_index: *ignore_index + - GrayRecResizeImg: + image_shape: [224, 224] # W H + resize_type: PIL # PIL or OpenCV + inter_type: 'Image.BICUBIC' + scale: false + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 2 diff --git a/configs/sr/sr_telescope.yml b/configs/sr/sr_telescope.yml new file mode 100644 index 0000000000000000000000000000000000000000..47950bb166ac6186204bb5b19224c78d51c59860 --- /dev/null +++ b/configs/sr/sr_telescope.yml @@ -0,0 +1,84 @@ +Global: + use_gpu: true + epoch_num: 100 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/sr/sr_telescope/ + save_epoch_step: 3 + # evaluation is run every 2000 iterations + eval_batch_step: [0, 1000] + cal_metric_during_train: False + pretrained_model: + checkpoints: + save_inference_dir: ./output/sr/sr_telescope/infer + use_visualdl: False + infer_img: doc/imgs_words_en/word_52.png + # for data or label process + character_dict_path: + max_text_length: 100 + infer_mode: False + use_space_char: False + save_res_path: ./output/sr/predicts_telescope.txt + +Optimizer: + name: Adam + beta1: 0.5 + beta2: 0.999 + clip_norm: 0.25 + lr: + learning_rate: 0.0001 + +Architecture: + model_type: sr + algorithm: Telescope + Transform: + name: TBSRN + STN: True + infer_mode: False + +Loss: + name: TelescopeLoss + confuse_dict_path: ./ppocr/utils/dict/confuse.pkl + + +PostProcess: + name: None + +Metric: + name: SRMetric + main_indicator: all + +Train: + dataset: + name: LMDBDataSetSR + data_dir: ./train_data/TextZoom/train + transforms: + - SRResize: + imgH: 32 + imgW: 128 + down_sample_scale: 2 + - KeepKeys: + keep_keys: ['img_lr', 'img_hr', 'label'] # dataloader will return list in this order + loader: + shuffle: False + batch_size_per_card: 16 + drop_last: True + num_workers: 4 + +Eval: + dataset: + name: LMDBDataSetSR + data_dir: ./train_data/TextZoom/test + transforms: + - SRResize: + imgH: 32 + imgW: 128 + down_sample_scale: 2 + - KeepKeys: + keep_keys: ['img_lr', 'img_hr', 'label'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 16 + num_workers: 4 + diff --git a/configs/sr/sr_tsrn_transformer_strock.yml b/configs/sr/sr_tsrn_transformer_strock.yml new file mode 100644 index 0000000000000000000000000000000000000000..a26ebb179143383200688e26d166444b7a1ed88a --- /dev/null +++ b/configs/sr/sr_tsrn_transformer_strock.yml @@ -0,0 +1,85 @@ +Global: + use_gpu: true + epoch_num: 500 + log_smooth_window: 20 + print_batch_step: 10 + output_dir: ./output/sr/sr_tsrn_transformer_strock/ + save_epoch_step: 3 + # evaluation is run every 2000 iterations + eval_batch_step: [0, 1000] + cal_metric_during_train: False + pretrained_model: + checkpoints: + save_inference_dir: sr_output + use_visualdl: False + infer_img: doc/imgs_words_en/word_52.png + # for data or label process + character_dict_path: ./train_data/srdata/english_decomposition.txt + max_text_length: 100 + infer_mode: False + use_space_char: False + save_res_path: ./output/sr/predicts_gestalt.txt + +Optimizer: + name: Adam + beta1: 0.5 + beta2: 0.999 + clip_norm: 0.25 + lr: + learning_rate: 0.0001 + +Architecture: + model_type: sr + algorithm: Gestalt + Transform: + name: TSRN + STN: True + infer_mode: False + +Loss: + name: StrokeFocusLoss + character_dict_path: ./train_data/srdata/english_decomposition.txt + +PostProcess: + name: None + +Metric: + name: SRMetric + main_indicator: all + +Train: + dataset: + name: LMDBDataSetSR + data_dir: ./train_data/srdata/train + transforms: + - SRResize: + imgH: 32 + imgW: 128 + down_sample_scale: 2 + - SRLabelEncode: # Class handling label + - KeepKeys: + keep_keys: ['img_lr', 'img_hr', 'length', 'input_tensor', 'label'] # dataloader will return list in this order + loader: + shuffle: False + batch_size_per_card: 16 + drop_last: True + num_workers: 4 + +Eval: + dataset: + name: LMDBDataSetSR + data_dir: ./train_data/srdata/test + transforms: + - SRResize: + imgH: 32 + imgW: 128 + down_sample_scale: 2 + - SRLabelEncode: # Class handling label + - KeepKeys: + keep_keys: ['img_lr', 'img_hr','length', 'input_tensor', 'label'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 16 + num_workers: 4 + diff --git a/configs/table/SLANet.yml b/configs/table/SLANet.yml new file mode 100644 index 0000000000000000000000000000000000000000..1f9e15ebb094b284609253da6c68bfb42fd4c2b6 --- /dev/null +++ b/configs/table/SLANet.yml @@ -0,0 +1,145 @@ +Global: + use_gpu: true + epoch_num: 100 + log_smooth_window: 20 + print_batch_step: 20 + output_dir: ./output/SLANet + save_epoch_step: 400 + # evaluation is run every 1000 iterations after the 0th iteration + eval_batch_step: [0, 1000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: ./output/SLANet/infer + use_visualdl: False + infer_img: ppstructure/docs/table/table.jpg + # for data or label process + character_dict_path: ppocr/utils/dict/table_structure_dict.txt + character_type: en + max_text_length: &max_text_length 500 + box_format: &box_format 'xyxy' # 'xywh', 'xyxy', 'xyxyxyxy' + infer_mode: False + use_sync_bn: True + save_res_path: 'output/infer' + d2s_train_image_shape: [3, -1, -1] + amp_custom_white_list: ['concat', 'elementwise_sub', 'set_value'] + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + clip_norm: 5.0 + lr: + name: Piecewise + learning_rate: 0.001 + decay_epochs : [40, 50] + values : [0.001, 0.0001, 0.00005] + regularizer: + name: 'L2' + factor: 0.00000 + +Architecture: + model_type: table + algorithm: SLANet + Backbone: + name: PPLCNet + scale: 1.0 + pretrained: true + use_ssld: true + Neck: + name: CSPPAN + out_channels: 96 + Head: + name: SLAHead + hidden_size: 256 + max_text_length: *max_text_length + loc_reg_num: &loc_reg_num 4 + +Loss: + name: SLALoss + structure_weight: 1.0 + loc_weight: 2.0 + loc_loss: smooth_l1 + +PostProcess: + name: TableLabelDecode + merge_no_span_structure: &merge_no_span_structure True + +Metric: + name: TableMetric + main_indicator: acc + compute_bbox_metric: False + loc_reg_num: *loc_reg_num + box_format: *box_format + +Train: + dataset: + name: PubTabDataSet + data_dir: train_data/table/pubtabnet/train/ + label_file_list: [train_data/table/pubtabnet/PubTabNet_2.0.0_train.jsonl] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - TableLabelEncode: + learn_empty_box: False + merge_no_span_structure: *merge_no_span_structure + replace_empty_cell_token: False + loc_reg_num: *loc_reg_num + max_text_length: *max_text_length + - TableBoxEncode: + in_box_format: *box_format + out_box_format: *box_format + - ResizeTableImage: + max_len: 488 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - PaddingTableImage: + size: [488, 488] + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'structure', 'bboxes', 'bbox_masks', 'shape' ] + loader: + shuffle: True + batch_size_per_card: 48 + drop_last: True + num_workers: 1 + +Eval: + dataset: + name: PubTabDataSet + data_dir: train_data/table/pubtabnet/val/ + label_file_list: [train_data/table/pubtabnet/PubTabNet_2.0.0_val.jsonl] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - TableLabelEncode: + learn_empty_box: False + merge_no_span_structure: *merge_no_span_structure + replace_empty_cell_token: False + loc_reg_num: *loc_reg_num + max_text_length: *max_text_length + - TableBoxEncode: + in_box_format: *box_format + out_box_format: *box_format + - ResizeTableImage: + max_len: 488 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - PaddingTableImage: + size: [488, 488] + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'structure', 'bboxes', 'bbox_masks', 'shape' ] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 48 + num_workers: 1 diff --git a/configs/table/SLANet_ch.yml b/configs/table/SLANet_ch.yml new file mode 100644 index 0000000000000000000000000000000000000000..5d8a13b99b02447872d73c81870544492305fa40 --- /dev/null +++ b/configs/table/SLANet_ch.yml @@ -0,0 +1,141 @@ +Global: + use_gpu: True + epoch_num: 400 + log_smooth_window: 20 + print_batch_step: 20 + output_dir: ./output/SLANet_ch + save_epoch_step: 400 + # evaluation is run every 331 iterations after the 0th iteration + eval_batch_step: [0, 331] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: ./output/SLANet_ch/infer + use_visualdl: False + infer_img: ppstructure/docs/table/table.jpg + # for data or label process + character_dict_path: ppocr/utils/dict/table_structure_dict_ch.txt + character_type: en + max_text_length: &max_text_length 500 + box_format: &box_format xyxyxyxy # 'xywh', 'xyxy', 'xyxyxyxy' + infer_mode: False + use_sync_bn: True + save_res_path: output/infer + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + clip_norm: 5.0 + lr: + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0.00000 + +Architecture: + model_type: table + algorithm: SLANet + Backbone: + name: PPLCNet + scale: 1.0 + pretrained: True + use_ssld: True + Neck: + name: CSPPAN + out_channels: 96 + Head: + name: SLAHead + hidden_size: 256 + max_text_length: *max_text_length + loc_reg_num: &loc_reg_num 8 + +Loss: + name: SLALoss + structure_weight: 1.0 + loc_weight: 2.0 + loc_loss: smooth_l1 + +PostProcess: + name: TableLabelDecode + merge_no_span_structure: &merge_no_span_structure True + +Metric: + name: TableMetric + main_indicator: acc + compute_bbox_metric: False + loc_reg_num: *loc_reg_num + box_format: *box_format + del_thead_tbody: True + +Train: + dataset: + name: PubTabDataSet + data_dir: train_data/table/train/ + label_file_list: [train_data/table/train.txt] + transforms: + - DecodeImage: + img_mode: BGR + channel_first: False + - TableLabelEncode: + learn_empty_box: False + merge_no_span_structure: *merge_no_span_structure + replace_empty_cell_token: False + loc_reg_num: *loc_reg_num + max_text_length: *max_text_length + - TableBoxEncode: + in_box_format: *box_format + out_box_format: *box_format + - ResizeTableImage: + max_len: 488 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - PaddingTableImage: + size: [488, 488] + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'structure', 'bboxes', 'bbox_masks', 'shape' ] + loader: + shuffle: True + batch_size_per_card: 48 + drop_last: True + num_workers: 1 + +Eval: + dataset: + name: PubTabDataSet + data_dir: train_data/table/val/ + label_file_list: [train_data/table/val.txt] + transforms: + - DecodeImage: + img_mode: BGR + channel_first: False + - TableLabelEncode: + learn_empty_box: False + merge_no_span_structure: *merge_no_span_structure + replace_empty_cell_token: False + loc_reg_num: *loc_reg_num + max_text_length: *max_text_length + - TableBoxEncode: + in_box_format: *box_format + out_box_format: *box_format + - ResizeTableImage: + max_len: 488 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - PaddingTableImage: + size: [488, 488] + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'structure', 'bboxes', 'bbox_masks', 'shape' ] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 48 + num_workers: 1 diff --git a/configs/table/table_master.yml b/configs/table/table_master.yml new file mode 100644 index 0000000000000000000000000000000000000000..5df9d28c3e6915811daf4f724705dee6eefd8256 --- /dev/null +++ b/configs/table/table_master.yml @@ -0,0 +1,145 @@ +Global: + use_gpu: true + epoch_num: 17 + log_smooth_window: 20 + print_batch_step: 100 + output_dir: ./output/table_master/ + save_epoch_step: 17 + eval_batch_step: [0, 6259] + cal_metric_during_train: true + pretrained_model: null + checkpoints: + save_inference_dir: output/table_master/infer + use_visualdl: false + infer_img: ppstructure/docs/table/table.jpg + save_res_path: ./output/table_master + character_dict_path: ppocr/utils/dict/table_master_structure_dict.txt + infer_mode: false + max_text_length: &max_text_length 500 + box_format: &box_format 'xywh' # 'xywh', 'xyxy', 'xyxyxyxy' + d2s_train_image_shape: [3, 480, 480] + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + name: MultiStepDecay + learning_rate: 0.001 + milestones: [12, 15] + gamma: 0.1 + warmup_epoch: 0.02 + regularizer: + name: L2 + factor: 0.0 + +Architecture: + model_type: table + algorithm: TableMaster + Backbone: + name: TableResNetExtra + gcb_config: + ratio: 0.0625 + headers: 1 + att_scale: False + fusion_type: channel_add + layers: [False, True, True, True] + layers: [1,2,5,3] + Head: + name: TableMasterHead + hidden_size: 512 + headers: 8 + dropout: 0 + d_ff: 2024 + max_text_length: *max_text_length + loc_reg_num: &loc_reg_num 4 + +Loss: + name: TableMasterLoss + ignore_index: 42 # set to len of dict + 3 + +PostProcess: + name: TableMasterLabelDecode + box_shape: pad + merge_no_span_structure: &merge_no_span_structure True + +Metric: + name: TableMetric + main_indicator: acc + compute_bbox_metric: False + box_format: *box_format + +Train: + dataset: + name: PubTabDataSet + data_dir: train_data/table/pubtabnet/train/ + label_file_list: [train_data/table/pubtabnet/PubTabNet_2.0.0_train.jsonl] + transforms: + - DecodeImage: + img_mode: BGR + channel_first: False + - TableMasterLabelEncode: + learn_empty_box: False + merge_no_span_structure: *merge_no_span_structure + replace_empty_cell_token: True + loc_reg_num: *loc_reg_num + max_text_length: *max_text_length + - ResizeTableImage: + max_len: 480 + resize_bboxes: True + - PaddingTableImage: + size: [480, 480] + - TableBoxEncode: + in_box_format: *box_format + out_box_format: *box_format + - NormalizeImage: + scale: 1./255. + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: hwc + - ToCHWImage: null + - KeepKeys: + keep_keys: [image, structure, bboxes, bbox_masks, shape] + loader: + shuffle: True + batch_size_per_card: 10 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: PubTabDataSet + data_dir: train_data/table/pubtabnet/val/ + label_file_list: [train_data/table/pubtabnet/PubTabNet_2.0.0_val.jsonl] + transforms: + - DecodeImage: + img_mode: BGR + channel_first: False + - TableMasterLabelEncode: + learn_empty_box: False + merge_no_span_structure: *merge_no_span_structure + replace_empty_cell_token: True + loc_reg_num: *loc_reg_num + max_text_length: *max_text_length + - ResizeTableImage: + max_len: 480 + resize_bboxes: True + - PaddingTableImage: + size: [480, 480] + - TableBoxEncode: + in_box_format: *box_format + out_box_format: *box_format + - NormalizeImage: + scale: 1./255. + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: hwc + - ToCHWImage: null + - KeepKeys: + keep_keys: [image, structure, bboxes, bbox_masks, shape] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 10 + num_workers: 8 \ No newline at end of file diff --git a/configs/table/table_mv3.yml b/configs/table/table_mv3.yml new file mode 100644 index 0000000000000000000000000000000000000000..417eec97d986862de7fab87bf34f338ee1e53bdb --- /dev/null +++ b/configs/table/table_mv3.yml @@ -0,0 +1,129 @@ +Global: + use_gpu: true + epoch_num: 400 + log_smooth_window: 20 + print_batch_step: 5 + output_dir: ./output/table_mv3/ + save_epoch_step: 400 + # evaluation is run every 400 iterations after the 0th iteration + eval_batch_step: [0, 400] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: ppstructure/docs/table/table.jpg + save_res_path: output/table_mv3 + # for data or label process + character_dict_path: ppocr/utils/dict/table_structure_dict.txt + character_type: en + max_text_length: &max_text_length 500 + box_format: &box_format 'xyxy' # 'xywh', 'xyxy', 'xyxyxyxy' + infer_mode: False + amp_custom_black_list: ['matmul_v2','elementwise_add'] + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + clip_norm: 5.0 + lr: + learning_rate: 0.001 + regularizer: + name: 'L2' + factor: 0.00000 + +Architecture: + model_type: table + algorithm: TableAttn + Backbone: + name: MobileNetV3 + scale: 1.0 + model_name: small + disable_se: true + Head: + name: TableAttentionHead + hidden_size: 256 + max_text_length: *max_text_length + loc_reg_num: &loc_reg_num 4 + +Loss: + name: TableAttentionLoss + structure_weight: 100.0 + loc_weight: 10000.0 + +PostProcess: + name: TableLabelDecode + +Metric: + name: TableMetric + main_indicator: acc + compute_bbox_metric: false # cost many time, set False for training + +Train: + dataset: + name: PubTabDataSet + data_dir: train_data/table/pubtabnet/train/ + label_file_list: [train_data/table/pubtabnet/PubTabNet_2.0.0_train.jsonl] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - TableLabelEncode: + learn_empty_box: False + merge_no_span_structure: False + replace_empty_cell_token: False + loc_reg_num: *loc_reg_num + max_text_length: *max_text_length + - TableBoxEncode: + - ResizeTableImage: + max_len: 488 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - PaddingTableImage: + size: [488, 488] + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'structure', 'bboxes', 'bbox_masks', 'shape' ] + loader: + shuffle: True + batch_size_per_card: 48 + drop_last: True + num_workers: 1 + +Eval: + dataset: + name: PubTabDataSet + data_dir: train_data/table/pubtabnet/val/ + label_file_list: [train_data/table/pubtabnet/PubTabNet_2.0.0_val.jsonl] + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - TableLabelEncode: + learn_empty_box: False + merge_no_span_structure: False + replace_empty_cell_token: False + loc_reg_num: *loc_reg_num + max_text_length: *max_text_length + - TableBoxEncode: + - ResizeTableImage: + max_len: 488 + - NormalizeImage: + scale: 1./255. + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: 'hwc' + - PaddingTableImage: + size: [488, 488] + - ToCHWImage: + - KeepKeys: + keep_keys: [ 'image', 'structure', 'bboxes', 'bbox_masks', 'shape' ] + loader: + shuffle: False + drop_last: False + batch_size_per_card: 48 + num_workers: 1