PPLCNetNew找不到對應模型

[cjwang0318]

• 系统环境/System Environment：Ubuntu
• 版本号/Version：Paddle v2.5： PaddleOCR： 问题相关组件/Related components： ppocr/modeling/necks/db_fpn.py
• 运行指令/Command Code：python tools/train.py -c ./ch_PP-OCRv4_det_cml.yml

使用ch_PP-OCRv4_det訓連模型進行fintune時，將 ch_PP-OCRv4_det_cml.yml 中的 Backbone name: PPLCNetNew 找不到對應模型，如果改成PPLCNetNewV3則會報錯 for i in range(len(in_channels)):TypeError: object of type 'int' has no len() 可以麻煩請教一下原因嗎？謝謝。

• 完整报错/Complete Error Message：

• Traceback (most recent call last): File "/home/cjwang/cjwang/PaddleOCR-release-2.7/tools/train.py", line 227, in main(config, device, logger, vdl_writer) File "/home/cjwang/cjwang/PaddleOCR-release-2.7/tools/train.py", line 135, in main model = build_model(config['Architecture']) File "/home/cjwang/cjwang/PaddleOCR-release-2.7/ppocr/modeling/architectures/init.py", line 34, in build_model arch = getattr(mod, name)(config) File "/home/cjwang/cjwang/PaddleOCR-release-2.7/ppocr/modeling/architectures/distillation_model.py", line 47, in init model = BaseModel(model_config) File "/home/cjwang/cjwang/PaddleOCR-release-2.7/ppocr/modeling/architectures/base_model.py", line 67, in init self.neck = build_neck(config['Neck']) File "/home/cjwang/cjwang/PaddleOCR-release-2.7/ppocr/modeling/necks/init.py", line 42, in build_neck module_class = eval(module_name)(**config) File "/home/cjwang/cjwang/PaddleOCR-release-2.7/ppocr/modeling/necks/db_fpn.py", line 240, in init for i in range(len(in_channels)): TypeError: object of type 'int' has no len()

[SWHL]

I meet the same problem.

[gl94]

不是“PPLCNetNewV3”而是“PPLCNetV3”

[Taichipeace]

不是“PPLCNetNewV3”而是“PPLCNetV3”

PPLCNetV3 我改成了这个，但是也是一样的报错。

[kadimakipp]

把PPLCNetNew改成PPLCNetV3，在添加一行det: true

[enemy1205]

添加det: true后， 训练会遇到另外报错

Traceback (most recent call last):
  File "/mnt/home/sp/projects/PaddleOCR/tools/train.py", line 228, in <module>
    main(config, device, logger, vdl_writer)
  File "/mnt/home/sp/projects/PaddleOCR/tools/train.py", line 199, in main
    program.train(config, train_dataloader, valid_dataloader, device, model,
  File "/mnt/home/sp/projects/PaddleOCR/tools/program.py", line 334, in train
    post_result = post_process_class(preds, batch[1])
  File "/mnt/home/sp/projects/PaddleOCR/ppocr/postprocess/db_postprocess.py", line 275, in __call__
    results[k] = self.post_process(predicts[k], shape_list=shape_list)
  File "/mnt/home/sp/projects/PaddleOCR/ppocr/postprocess/db_postprocess.py", line 229, in __call__
    src_h, src_w, ratio_h, ratio_w = shape_list[batch_index]
ValueError: too many values to unpack (expected 4)
I1013 19:03:27.982057 679392 tcp_store.cc:273] receive shutdown event and so quit from MasterDaemon run loop

调试进入 ppocr/postprocess/db_postprocess.py文件中229lines

        for batch_index in range(pred.shape[0]):
            src_h, src_w, ratio_h, ratio_w = shape_list[batch_index]
            if self.dilation_kernel is not None:
                mask = cv2.dilate(
                    np.array(segmentation[batch_index]).astype(np.uint8),
                    self.dilation_kernel)

发现shappe_list是一个 batch_size * 640(height) *640(width) 的数组， 显然下标访问后取值无法返回src_h, src_w, ratio_h, ratio_w

[kadimakipp]

https://github.com/PaddlePaddle/PaddleOCR/issues/5710

[ITccyhub]

Traceback (most recent call last): File "tools/train.py", line 227, in main(config, device, logger, vdl_writer) File "tools/train.py", line 135, in main model = build_model(config['Architecture']) File "D:\PaddleOCR\PaddleOCR-release-2.7\ppocr\modeling\architectures_init_.py", line 34, in build_model arch = getattr(mod, name)(config) File "D:\PaddleOCR\PaddleOCR-release-2.7\ppocr\modeling\architectures\distillation_model.py", line 47, in init model = BaseModel(model_config) File "D:\PaddleOCR\PaddleOCR-release-2.7\ppocr\modeling\architectures\base_model.py", line 67, in init self.neck = build_neck(config['Neck']) File "D:\PaddleOCR\PaddleOCR-release-2.7\ppocr\modeling\necks_init_.py", line 42, in build_neck module_class = eval(module_name)(**config) File "D:\PaddleOCR\PaddleOCR-release-2.7\ppocr\modeling\necks\db_fpn.py", line 240, in init for i in range(len(in_channels)): TypeError: object of type 'int' has no len()

[ITccyhub]

an'zh按照这个改了还是报错

[ITccyhub]

加上 det: true 直接

[sylarwcy]

把PPLCNetNew改成PPLCNetV3，在添加一行det: true

这个是正解

[sylarwcy]

只有设定了det: true，才能走输出序列这个条件分支。后面RSEFPN解析in_channels时，按序列来解，才不会报错。

[KongXCai]

添加det: true后， 训练会遇到另外报错

Traceback (most recent call last):
  File "/mnt/home/sp/projects/PaddleOCR/tools/train.py", line 228, in <module>
    main(config, device, logger, vdl_writer)
  File "/mnt/home/sp/projects/PaddleOCR/tools/train.py", line 199, in main
    program.train(config, train_dataloader, valid_dataloader, device, model,
  File "/mnt/home/sp/projects/PaddleOCR/tools/program.py", line 334, in train
    post_result = post_process_class(preds, batch[1])
  File "/mnt/home/sp/projects/PaddleOCR/ppocr/postprocess/db_postprocess.py", line 275, in __call__
    results[k] = self.post_process(predicts[k], shape_list=shape_list)
  File "/mnt/home/sp/projects/PaddleOCR/ppocr/postprocess/db_postprocess.py", line 229, in __call__
    src_h, src_w, ratio_h, ratio_w = shape_list[batch_index]
ValueError: too many values to unpack (expected 4)
I1013 19:03:27.982057 679392 tcp_store.cc:273] receive shutdown event and so quit from MasterDaemon run loop

调试进入 ppocr/postprocess/db_postprocess.py文件中229lines

        for batch_index in range(pred.shape[0]):
            src_h, src_w, ratio_h, ratio_w = shape_list[batch_index]
            if self.dilation_kernel is not None:
                mask = cv2.dilate(
                    np.array(segmentation[batch_index]).astype(np.uint8),
                    self.dilation_kernel)

发现shappe_list是一个 batch_size * 640(height) *640(width) 的数组， 显然下标访问后取值无法返回src_h, src_w, ratio_h, ratio_w

我也遇到了跟你一样的问题，请问你解决了吗？

[Sundragon1993]

@KongXCai just set: cal_metric_during_train: false

[Double97828]

把PPLCNetNew改成PPLCNetV3，在添加一行det: true

这个student1和student2都要做修改么，我两个都修改了以后，有非常多层的参数无法加载不到模型中

[koala-cat]

把PPLCNetNew改成PPLCNetV3，在添加一行det: true

这个student1和student2都要做修改么，我两个都修改了以后，有非常多层的参数无法加载不到模型中

解决了么？我也是PPLCNetNew改成PPLCNetV3之后，student1和student2都加了 det: true，也试过单独加 student1，结果还是报错

[nikooooooooooooooooooo]

配置加上还是报错，大佬解决了吗？ 预训练模型：ch_PP-OCRv4_det_train

[mattheliu]

@564142183 你可以试一下 set: cal_metric_during_train: false

[nikooooooooooooooooooo]

@mattheliu 感谢大佬回复，把cal_metric_during_train改成false后训练还是报错

[2024-05-11 09:12:21,664] [ WARNING] dataloader_iter.py:707 - DataLoader 1 workers exit unexpectedly, pids: 5673
Traceback (most recent call last):
  File "/PaddleOCR/tools/train.py", line 255, in <module>
    main(config, device, logger, vdl_writer, seed)
  File "/PaddleOCR/tools/train.py", line 208, in main
    program.train(
  File "/PaddleOCR/tools/program.py", line 304, in train
    for idx, batch in enumerate(train_dataloader):
  File "/usr/local/lib/python3.10/dist-packages/paddle/io/dataloader/dataloader_iter.py", line 826, in __next__
    self._reader.read_next_list()[0]
SystemError: (Fatal) Blocking queue is killed because the data reader raises an exception.
  [Hint: Expected killed_ != true, but received killed_:1 == true:1.] (at ../paddle/fluid/operators/reader/blocking_queue.h:171)

[nikooooooooooooooooooo]

问题已解决，就是训练完的效果不太理想，附项目地址：https://aistudio.baidu.com/projectdetail/7900653?sUid=3018256&shared=1&ts=1716019542135