训练V3版本的模型ch_PP-OCRv3_rec.yml报错，acc一直是0

[hds1999]

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：X86 windows11
• 版本号/Version：Paddle：paddle-gpu 2.5.1 PaddleOCR：PaddleOcrV2.7 问题相关组件/Related components：ch_PP-OCRv3_rec.yml
• Global: debug: false use_gpu: true epoch_num: 200 log_smooth_window: 20 print_batch_step: 10 save_model_dir: ./output/rec_ppocr_v3 save_epoch_step: 10 eval_batch_step: [0, 100] cal_metric_during_train: true pretrained_model: ./pretrain_models/ch_rec/best_accuracy.pdparams 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: &max_text_length 100 infer_mode: false use_space_char: true distributed: true save_res_path: ./output/rec/predicts_ppocrv3.txt

Optimizer: name: Adam beta1: 0.9 beta2: 0.999 lr: name: Cosine learning_rate: 0.0001 warmup_epoch: 5 regularizer: name: L2 factor: 3.0e-05

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 Head: fc_decay: 0.00001
• SARHead: enc_dim: 512 max_text_length: *max_text_length

Loss: name: MultiLoss loss_config_list:

• CTCLoss:
• SARLoss:

PostProcess:
name: CTCLabelDecode

Metric: name: RecMetric main_indicator: acc ignore_space: False

Train: dataset: name: SimpleDataSet data_dir: ./train_data/ch_rec/ ext_op_transform_idx: 1 label_file_list:

• ./train_data/ch_rec/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: 32 drop_last: true num_workers: 1 Eval: dataset: name: SimpleDataSet data_dir: ./train_data/ch_rec/ label_file_list:
• ./train_data/ch_rec/eval_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: 32 num_workers: 1
    • 运行指令/Command Code：python tools/train.py -c configs\rec\PP-OCRv3\ch_PP-OCRv3_rec.yml -o Global.use_gpu=True Global.save_model_dir=./output/inference Train.loader.batch_size_per_card=1
    • 完整报错/Complete Error Message： [2024/04/09 16:36:40] ppocr INFO: train with paddle 2.5.1 and device Place(gpu:0) [2024/04/09 16:36:40] ppocr INFO: Initialize indexs of datasets:['./train_data/ch_rec/train_list.txt'] list index out of range [2024/04/09 16:36:40] ppocr INFO: Initialize indexs of datasets:['./train_data/ch_rec/eval_list.txt'] [2024/04/09 16:36:43] ppocr INFO: train dataloader has 103 iters [2024/04/09 16:36:43] ppocr INFO: valid dataloader has 103 iters [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.conv1._conv.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.conv1._batch_norm.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.conv1._batch_norm.bias not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.conv1._batch_norm._mean not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.conv1._batch_norm._variance not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._depthwise_conv._conv.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._depthwise_conv._batch_norm.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._depthwise_conv._batch_norm.bias not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._depthwise_conv._batch_norm._mean not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._depthwise_conv._batch_norm._variance not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._pointwise_conv._conv.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._pointwise_conv._batch_norm.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._pointwise_conv._batch_norm.bias not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._pointwise_conv._batch_norm._mean not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.0._pointwise_conv._batch_norm._variance not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._depthwise_conv._conv.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._depthwise_conv._batch_norm.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._depthwise_conv._batch_norm.bias not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._depthwise_conv._batch_norm._mean not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._depthwise_conv._batch_norm._variance not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._pointwise_conv._conv.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._pointwise_conv._batch_norm.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._pointwise_conv._batch_norm.bias not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._pointwise_conv._batch_norm._mean not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.1._pointwise_conv._batch_norm._variance not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.2._depthwise_conv._conv.weight not in model [2024/04/09 16:36:43] ppocr WARNING: The pretrained params Teacher.backbone.block_list.2._depthwise_conv._batch_norm.weight not in model

[GreatV]

是不是把预训练模型载入错了

[hds1999]

是不是把预训练模型载入错了 用的配置文件： https://github.com/PaddlePaddle/PaddleOCR/blob/main/configs/rec/PP-OCRv3/ch_PP-OCRv3_rec.yml 预训练模型： https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/ch_PP-OCRv3_rec_train.tar

[TingquanGao]

使用 https://github.com/PaddlePaddle/PaddleOCR/blob/main/configs/rec/PP-OCRv3/ch_PP-OCRv3_rec_distillation.yml 训练配置，或者： 参考文档 https://github.com/PaddlePaddle/PaddleOCR/blob/main/doc/doc_ch/knowledge_distillation.md#215-%E8%92%B8%E9%A6%8F%E6%A8%A1%E5%9E%8B%E5%BE%AE%E8%B0%83 把 https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/ch_PP-OCRv3_rec_train.tar 中的student参数提取出来。

[TingquanGao]

长时间未回复，该issue已关闭，如仍有问题可以reopen或新开issue。

[nissansz]

使用 https://github.com/PaddlePaddle/PaddleOCR/blob/main/configs/rec/PP-OCRv3/ch_PP-OCRv3_rec_distillation.yml 训练配置，或者： 参考文档 https://github.com/PaddlePaddle/PaddleOCR/blob/main/doc/doc_ch/knowledge_distillation.md#215-%E8%92%B8%E9%A6%8F%E6%A8%A1%E5%9E%8B%E5%BE%AE%E8%B0%83 把 https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/ch_PP-OCRv3_rec_train.tar 中的student参数提取出来。

哪个svtr 配置文件速度最快，准确率最高？