Closed isLiuHao closed 2 years ago
请提供详细的配置文件和运行指令
配置文件和运行指令如下:
Global: algorithm: CRNN use_gpu: true epoch_num: 1000 log_smooth_window: 20 print_batch_step: 10 save_model_dir: output/rec_CRNN save_epoch_step: 3 eval_batch_step: 2000 train_batch_size_per_card: 2048 test_batch_size_per_card: 2048 image_shape: [3, 32, 320] max_text_length: 25 character_type: ch character_dict_path: ./ppocr/utils/lian-med-dict.txt loss_type: ctc distort: false use_space_char: true reader_yml: ./configs/rec/rec_chinese_reader.yml pretrain_weights: checkpoints: save_inference_dir: infer_img: Architecture: function: ppocr.modeling.architectures.rec_model,RecModel Backbone: function: ppocr.modeling.backbones.rec_resnet_vd,ResNet layers: 34 Head: function: ppocr.modeling.heads.rec_ctc_head,CTCPredict encoder_type: rnn SeqRNN: hidden_size: 256 Loss: function: ppocr.modeling.losses.rec_ctc_loss,CTCLoss Optimizer: function: ppocr.optimizer,AdamDecay base_lr: 0.001 beta1: 0.9 beta2: 0.999
python3 tools/train.py \ -c configs/rec/rec_r34_vd_none_bilstm_ctc.yml \ -o Global.character_dict_path=./ppocr/utils/lian-med-dict.txt \ Global.save_model_dir=./output/rec_r34_vd_none_bilstm_ctc
建议使用最新版release2.4版本哦,现在主推动态图,develop版本不再维护更新了 https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.4/configs/rec/rec_r34_vd_none_bilstm_ctc.yml
建议使用最新版release2.4版本哦,现在主推动态图,develop版本不再维护更新了 https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.4/configs/rec/rec_r34_vd_none_bilstm_ctc.yml
2.4版本是可以直接训练静态图是吗?
先训练动态图,动态图方便训练调试,然后可以通过转换为静态图来进行预测部署。
您好,我使用的rec_r34_vd_none_bilstm_ctc.yml配置在V100上进行训练,数据集为360W的《中文文档文字识别》。训练了100个epoch,acc仍未0.006左右,请问一下这种情况正常吗?还有没有比较继续训练?