PaddleOCR
PaddleOCR copied to clipboard
使用ch_PP-OCRv4_rec_distill.yml 训练时SimpleDataSet 重复报错 KeyError: 'valid_ratio'
请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem
- 系统环境/System Environment:windows 10
- 版本号/Version:Paddle: 2.4.2.post116 PaddleOCR: 2.7.3问题相关组件/Related components:
- 运行指令/Command Code:
- 完整报错/Complete Error Message: [2024/05/22 14:06:59] ppocr INFO: Architecture : [2024/05/22 14:06:59] ppocr INFO: Models : [2024/05/22 14:06:59] ppocr INFO: Student : [2024/05/22 14:06:59] ppocr INFO: Backbone : [2024/05/22 14:06:59] ppocr INFO: name : PPLCNetV3 [2024/05/22 14:06:59] ppocr INFO: scale : 0.95 [2024/05/22 14:06:59] ppocr INFO: Head : [2024/05/22 14:06:59] ppocr INFO: head_list : [2024/05/22 14:06:59] ppocr INFO: CTCHead : [2024/05/22 14:06:59] ppocr INFO: Head : [2024/05/22 14:06:59] ppocr INFO: fc_decay : 1e-05 [2024/05/22 14:06:59] ppocr INFO: Neck : [2024/05/22 14:06:59] ppocr INFO: depth : 2 [2024/05/22 14:06:59] ppocr INFO: dims : 120 [2024/05/22 14:06:59] ppocr INFO: hidden_dims : 120 [2024/05/22 14:06:59] ppocr INFO: kernel_size : [1, 3] [2024/05/22 14:06:59] ppocr INFO: name : svtr [2024/05/22 14:06:59] ppocr INFO: use_guide : True [2024/05/22 14:06:59] ppocr INFO: NRTRHead : [2024/05/22 14:06:59] ppocr INFO: max_text_length : 25 [2024/05/22 14:06:59] ppocr INFO: nrtr_dim : 384 [2024/05/22 14:06:59] ppocr INFO: name : MultiHead [2024/05/22 14:06:59] ppocr INFO: Transform : None [2024/05/22 14:06:59] ppocr INFO: algorithm : SVTR [2024/05/22 14:06:59] ppocr INFO: freeze_params : False [2024/05/22 14:06:59] ppocr INFO: model_type : rec [2024/05/22 14:06:59] ppocr INFO: pretrained : None [2024/05/22 14:06:59] ppocr INFO: return_all_feats : True [2024/05/22 14:06:59] ppocr INFO: Teacher : [2024/05/22 14:06:59] ppocr INFO: Backbone : [2024/05/22 14:06:59] ppocr INFO: depth : [3, 6, 3] [2024/05/22 14:06:59] ppocr INFO: embed_dim : [64, 128, 256] [2024/05/22 14:06:59] ppocr INFO: img_size : [48, 320] [2024/05/22 14:06:59] ppocr INFO: last_stage : False [2024/05/22 14:06:59] ppocr INFO: local_mixer : [[5, 5], [5, 5], [5, 5]] [2024/05/22 14:06:59] ppocr INFO: mixer : ['Conv', 'Conv', 'Conv', 'Conv', 'Conv', 'Conv', 'Global', 'Global', 'Global', 'Global', 'Global', 'Global'] [2024/05/22 14:06:59] ppocr INFO: name : SVTRNet [2024/05/22 14:06:59] ppocr INFO: num_heads : [2, 4, 8] [2024/05/22 14:06:59] ppocr INFO: out_channels : 192 [2024/05/22 14:06:59] ppocr INFO: out_char_num : 40 [2024/05/22 14:06:59] ppocr INFO: patch_merging : Conv [2024/05/22 14:06:59] ppocr INFO: prenorm : True [2024/05/22 14:06:59] ppocr INFO: Head : [2024/05/22 14:06:59] ppocr INFO: head_list : [2024/05/22 14:06:59] ppocr INFO: CTCHead : [2024/05/22 14:06:59] ppocr INFO: Head : [2024/05/22 14:06:59] ppocr INFO: fc_decay : 1e-05 [2024/05/22 14:06:59] ppocr INFO: Neck : [2024/05/22 14:06:59] ppocr INFO: depth : 2 [2024/05/22 14:06:59] ppocr INFO: dims : 120 [2024/05/22 14:06:59] ppocr INFO: hidden_dims : 120 [2024/05/22 14:06:59] ppocr INFO: kernel_size : [1, 3] [2024/05/22 14:06:59] ppocr INFO: name : svtr [2024/05/22 14:06:59] ppocr INFO: use_guide : True [2024/05/22 14:06:59] ppocr INFO: NRTRHead : [2024/05/22 14:06:59] ppocr INFO: max_text_length : 25 [2024/05/22 14:06:59] ppocr INFO: nrtr_dim : 384 [2024/05/22 14:06:59] ppocr INFO: name : MultiHead [2024/05/22 14:06:59] ppocr INFO: Transform : None [2024/05/22 14:06:59] ppocr INFO: algorithm : SVTR [2024/05/22 14:06:59] ppocr INFO: freeze_params : True [2024/05/22 14:06:59] ppocr INFO: model_type : rec [2024/05/22 14:06:59] ppocr INFO: pretrained : None [2024/05/22 14:06:59] ppocr INFO: return_all_feats : True [2024/05/22 14:06:59] ppocr INFO: algorithm : Distillation [2024/05/22 14:06:59] ppocr INFO: model_type : rec [2024/05/22 14:06:59] ppocr INFO: name : DistillationModel [2024/05/22 14:06:59] ppocr INFO: Eval : [2024/05/22 14:06:59] ppocr INFO: dataset : [2024/05/22 14:06:59] ppocr INFO: data_dir : H:/PythonCode/PaddleOCR-main/train_data/yanhe_small3_test [2024/05/22 14:06:59] ppocr INFO: label_file_list : ['H:/PythonCode/PaddleOCR-main/train_data/yanhe_small3_test/rec_gt.txt'] [2024/05/22 14:06:59] ppocr INFO: name : SimpleDataSet [2024/05/22 14:06:59] ppocr INFO: transforms : [2024/05/22 14:06:59] ppocr INFO: DecodeImage : [2024/05/22 14:06:59] ppocr INFO: channel_first : False [2024/05/22 14:06:59] ppocr INFO: img_mode : BGR [2024/05/22 14:06:59] ppocr INFO: MultiLabelEncode : [2024/05/22 14:06:59] ppocr INFO: gtc_encode : NRTRLabelEncode [2024/05/22 14:06:59] ppocr INFO: RecResizeImg : [2024/05/22 14:06:59] ppocr INFO: image_shape : [3, 48, 320] [2024/05/22 14:06:59] ppocr INFO: KeepKeys : [2024/05/22 14:06:59] ppocr INFO: keep_keys : ['image', 'label_ctc', 'label_gtc', 'length', 'valid_ratio'] [2024/05/22 14:06:59] ppocr INFO: loader : [2024/05/22 14:06:59] ppocr INFO: batch_size_per_card : 128 [2024/05/22 14:06:59] ppocr INFO: drop_last : False [2024/05/22 14:06:59] ppocr INFO: num_workers : 4 [2024/05/22 14:06:59] ppocr INFO: shuffle : False [2024/05/22 14:06:59] ppocr INFO: Global : [2024/05/22 14:06:59] ppocr INFO: cal_metric_during_train : True [2024/05/22 14:06:59] ppocr INFO: character_dict_path : ppocr/utils/ppocr_keys_v1.txt [2024/05/22 14:06:59] ppocr INFO: checkpoints : None [2024/05/22 14:06:59] ppocr INFO: debug : False [2024/05/22 14:06:59] ppocr INFO: distributed : False [2024/05/22 14:06:59] ppocr INFO: epoch_num : 100 [2024/05/22 14:06:59] ppocr INFO: eval_batch_step : [0, 500] [2024/05/22 14:06:59] ppocr INFO: infer_img : doc/imgs_words/ch/word_1.jpg [2024/05/22 14:06:59] ppocr INFO: infer_mode : False [2024/05/22 14:06:59] ppocr INFO: log_smooth_window : 20 [2024/05/22 14:06:59] ppocr INFO: max_text_length : 25 [2024/05/22 14:06:59] ppocr INFO: pretrained_model : H:/PythonCode/PaddleOCR-main/PretrainedMode/rec/ch_PP-OCRv4_rec_train/student [2024/05/22 14:06:59] ppocr INFO: print_batch_step : 1 [2024/05/22 14:06:59] ppocr INFO: save_epoch_step : 100 [2024/05/22 14:06:59] ppocr INFO: save_inference_dir : None [2024/05/22 14:06:59] ppocr INFO: save_model_dir : ./output/rec_dkd_400w_svtr_ctc_lcnet_blank_dkd0.1/ [2024/05/22 14:06:59] ppocr INFO: save_res_path : ./output/rec/predicts_ppocrv3.txt [2024/05/22 14:06:59] ppocr INFO: use_gpu : True [2024/05/22 14:06:59] ppocr INFO: use_space_char : True [2024/05/22 14:06:59] ppocr INFO: use_visualdl : False [2024/05/22 14:06:59] ppocr INFO: Loss : [2024/05/22 14:06:59] ppocr INFO: loss_config_list : [2024/05/22 14:06:59] ppocr INFO: DistillationDKDLoss : [2024/05/22 14:06:59] ppocr INFO: alpha : 1.0 [2024/05/22 14:06:59] ppocr INFO: beta : 2.0 [2024/05/22 14:06:59] ppocr INFO: dis_head : gtc [2024/05/22 14:06:59] ppocr INFO: key : head_out [2024/05/22 14:06:59] ppocr INFO: model_name_pairs : [['Student', 'Teacher']] [2024/05/22 14:06:59] ppocr INFO: multi_head : True [2024/05/22 14:06:59] ppocr INFO: name : dkd [2024/05/22 14:06:59] ppocr INFO: weight : 0.1 [2024/05/22 14:06:59] ppocr INFO: DistillationCTCLoss : [2024/05/22 14:06:59] ppocr INFO: key : head_out [2024/05/22 14:06:59] ppocr INFO: model_name_list : ['Student'] [2024/05/22 14:06:59] ppocr INFO: multi_head : True [2024/05/22 14:06:59] ppocr INFO: weight : 1.0 [2024/05/22 14:06:59] ppocr INFO: DistillationNRTRLoss : [2024/05/22 14:06:59] ppocr INFO: key : head_out [2024/05/22 14:06:59] ppocr INFO: model_name_list : ['Student'] [2024/05/22 14:06:59] ppocr INFO: multi_head : True [2024/05/22 14:06:59] ppocr INFO: smoothing : False [2024/05/22 14:06:59] ppocr INFO: weight : 1.0 [2024/05/22 14:06:59] ppocr INFO: DistillCTCLogits : [2024/05/22 14:06:59] ppocr INFO: key : head_out [2024/05/22 14:06:59] ppocr INFO: model_name_pairs : [['Student', 'Teacher']] [2024/05/22 14:06:59] ppocr INFO: reduction : mean [2024/05/22 14:06:59] ppocr INFO: weight : 1.0 [2024/05/22 14:06:59] ppocr INFO: name : CombinedLoss [2024/05/22 14:06:59] ppocr INFO: Metric : [2024/05/22 14:06:59] ppocr INFO: base_metric_name : RecMetric [2024/05/22 14:06:59] ppocr INFO: ignore_space : False [2024/05/22 14:06:59] ppocr INFO: key : Student [2024/05/22 14:06:59] ppocr INFO: main_indicator : acc [2024/05/22 14:06:59] ppocr INFO: name : DistillationMetric [2024/05/22 14:06:59] ppocr INFO: Optimizer : [2024/05/22 14:06:59] ppocr INFO: beta1 : 0.9 [2024/05/22 14:06:59] ppocr INFO: beta2 : 0.999 [2024/05/22 14:06:59] ppocr INFO: lr : [2024/05/22 14:06:59] ppocr INFO: learning_rate : 0.001 [2024/05/22 14:06:59] ppocr INFO: name : Cosine [2024/05/22 14:06:59] ppocr INFO: warmup_epoch : 2 [2024/05/22 14:06:59] ppocr INFO: name : Adam [2024/05/22 14:06:59] ppocr INFO: regularizer : [2024/05/22 14:06:59] ppocr INFO: factor : 3e-05 [2024/05/22 14:06:59] ppocr INFO: name : L2 [2024/05/22 14:06:59] ppocr INFO: PostProcess : [2024/05/22 14:06:59] ppocr INFO: key : head_out [2024/05/22 14:06:59] ppocr INFO: model_name : ['Student'] [2024/05/22 14:06:59] ppocr INFO: multi_head : True [2024/05/22 14:06:59] ppocr INFO: name : DistillationCTCLabelDecode [2024/05/22 14:06:59] ppocr INFO: Train : [2024/05/22 14:06:59] ppocr INFO: dataset : [2024/05/22 14:06:59] ppocr INFO: data_dir : H:/PythonCode/PaddleOCR-main/train_data/yanhe_small3_train [2024/05/22 14:06:59] ppocr INFO: label_file_list : ['H:/PythonCode/PaddleOCR-main/train_data/yanhe_small3_train/rec_gt.txt'] [2024/05/22 14:06:59] ppocr INFO: name : SimpleDataSet [2024/05/22 14:06:59] ppocr INFO: ratio_list : [1.0] [2024/05/22 14:06:59] ppocr INFO: transforms : [2024/05/22 14:06:59] ppocr INFO: DecodeImage : [2024/05/22 14:06:59] ppocr INFO: channel_first : False [2024/05/22 14:06:59] ppocr INFO: img_mode : BGR [2024/05/22 14:06:59] ppocr INFO: RecAug : None [2024/05/22 14:06:59] ppocr INFO: MultiLabelEncode : [2024/05/22 14:06:59] ppocr INFO: gtc_encode : NRTRLabelEncode [2024/05/22 14:06:59] ppocr INFO: KeepKeys : [2024/05/22 14:06:59] ppocr INFO: keep_keys : ['image', 'label_ctc', 'label_gtc', 'length', 'valid_ratio'] [2024/05/22 14:06:59] ppocr INFO: loader : [2024/05/22 14:06:59] ppocr INFO: batch_size_per_card : 2 [2024/05/22 14:06:59] ppocr INFO: drop_last : True [2024/05/22 14:06:59] ppocr INFO: num_workers : 8 [2024/05/22 14:06:59] ppocr INFO: shuffle : True [2024/05/22 14:06:59] ppocr INFO: use_shared_memory : True [2024/05/22 14:06:59] ppocr INFO: profiler_options : None [2024/05/22 14:06:59] ppocr INFO: train with paddle 2.4.2 and device Place(gpu:0) [2024/05/22 14:06:59] ppocr INFO: Initialize indexs of datasets:['H:/PythonCode/PaddleOCR-main/train_data/yanhe_small3_train/rec_gt.txt'] list index out of range [2024/05/22 14:06:59] ppocr INFO: Initialize indexs of datasets:['H:/PythonCode/PaddleOCR-main/train_data/yanhe_small3_test/rec_gt.txt'] W0522 14:06:59.953976 1384 gpu_resources.cc:61] Please NOTE: device: 0, GPU Compute Capability: 6.1, Driver API Version: 11.6, Runtime API Version: 11.6 W0522 14:06:59.958967 1384 gpu_resources.cc:91] device: 0, cuDNN Version: 8.4. [2024/05/22 14:07:02] ppocr INFO: train dataloader has 78 iters [2024/05/22 14:07:02] ppocr INFO: valid dataloader has 1 iters
重复报错信息: Traceback (most recent call last): File "D:\Work\hionNowWork\PaddleOCR\ppocr\data\simple_dataset.py", line 159, in getitem outs = transform(data, self.ops) File "D:\Work\hionNowWork\PaddleOCR\ppocr\data\imaug_init_.py", line 56, in transform data = op(data) File "D:\Work\hionNowWork\PaddleOCR\ppocr\data\imaug\operators.py", line 133, in call data_list.append(data[key]) KeyError: 'valid_ratio'