WenmuZhou/PytorchOCR

Issues

• 预训练模型和config对应

  #297 opened 3 months ago by Mr-lonely0
  3

• 在icdar 2015数据集上训练rec模型，rec_acc能到0.99，但是在训练集上只有0.62左右

  #299 opened 2 months ago by Mr-lonely0
  4

• 建议合入表格识别SLANet：https://github.com/yutian929/SLANet-Pytorch

  #298 opened 3 months ago by yutian929
  3

• predicts = predicts['res']

  #288 opened 6 months ago by Alexander-code-365
  12

• cls_mv3 模型预训练权重

  #296 opened 3 months ago by GongYuxiu
  1

• 训练开始没几步就报nan

  #293 opened 6 months ago by maomaoyuchengzi
  1

• 训练 text detection不收敛，eval 的结果都是0

  #295 opened 8 months ago by Starethics
  1

• ch_PP-OCRv4_det_cml模块对齐精度不达标

  #294 opened 6 months ago by guanfuchen
  1

• 训练报错，求救大佬

  #290 opened 6 months ago by ggsimida-cw
  1

• Train

  #289 opened 6 months ago by kingloo2014
  5

• onnx2trt

  #286 opened 6 months ago by txy00001
  1

• 大佬你好 ，想问个关于模型大小的问题

  #284 opened 6 months ago by chenhengwang
  1

• 训练好的paddle模型，迁移到torch，又转onnx，进行推理为什么都识别不出来东西

  #283 opened 6 months ago by Ivansyf34
  1

• 为什么新样本首次推理速度都会慢？

  #279 opened a year ago by DemonDamon
  6

• 请教一下 mac m2 gpu 环境上的问题

  #291 opened 9 months ago by kwdchol
  1

• 您好，onnx模型在哪拿呢？

  #292 opened 9 months ago by June-Li
  0

• 关于paddleocr库代码中使用这个配置ch_PP-OCRv4_rec_distill.yml进行文本识别训练代码bug

  #287 opened a year ago by QQQTAO
  1

• 转trt

  #285 opened a year ago by txy00001
  1

• 检测模型pth转换为onnx报错，configs\det\ch_PP-OCRv3\ch_PP-OCRv3_det_student.yml，

  #282 opened a year ago by Ivansyf34
  0

• 文字识别模型的训练数据来源

  #280 opened a year ago by fourierer
  3

• predict部分需要三个模型是什么情况？

  #281 opened a year ago by Mr47121836
  2

• use the comand to run the train report the warning,any one have the same warning?

  #275 opened a year ago by erquren
  1

• cls_mv3 模型权重是不是忘了上传了？ 链接里没有

  #278 opened a year ago by GuangtaoLyu
  2

• ch_ppocr_mobile_v2.0_cls_train 的 config文件是不是没上传？

  #277 opened a year ago by GuangtaoLyu
  3

• 请问cls的模型有开源参数吗

  #273 opened a year ago by Arlen-yuzu
  2

• 训练ch_PP-OCRv3_rec不收敛，CTCLOSS到50就不降了

  #276 opened a year ago by erquren
  1

• 请问目前转换成onnx格式推理的功能可以使用了吗

  #272 opened a year ago by Arlen-yuzu
  3

• 在det模型中，如何调节检测框之间合并的间距阈值

  #274 opened a year ago by Arlen-yuzu
  0

• python tools/infer_rec.py -c configs/rec/PP-OCRv3/ch_PP-OCRv3_rec_distillation.yml

  #271 opened a year ago by 2018-Summer
  6

• 有没有Int版本的模型

  #268 opened a year ago by HGGshiwo
  0

• ValueError: num_samples should be a positive integer value, but got num_samples=0

  #269 opened a year ago by 2018-Summer
  0

• 如何在ocr识别的字典文件中添加上下标的训练素材

  #270 opened a year ago by fourierer
  0

• 您好，作者，请问可以加载paddle的检测模型进行微调训练吗？

  #265 opened a year ago by luoyq6
  4

• 识别模型训练时如何开启数据增强

  #266 opened a year ago by hellohahaw
  2

• pytorchOCR中的DBNet与文幕纯DBNet精度差别大

  #267 opened a year ago by ssvicnent
  0

• 有没有本项目crnn pth训练结果 转onnx的代码？

  #262 opened a year ago by nissansz
  0

• 大佬们你们训练的时候速度是多少？为什么我这里这么👨

  #263 opened a year ago by bianliuyang
  0

• 文本识别: RecLmdbDataset数据集存在严重bug

  #264 opened a year ago by wkailiu
  0

• CRNN中文训练不收敛，只有lr=0.005才勉强能慢慢收敛，非常慢

  #255 opened a year ago by huangxin168
  1

• MobileNetV3 small CRNN训练跑不满GPU

  #256 opened a year ago by huangxin168
  1

• 如何使用convnext swintransformer的预训练模型

  #257 opened a year ago by DenghuiXiao
  0

• det_infer 推理没转成rgb， 训练验证都是rgb 会导致结果不一致。。

  #250 opened a year ago by shantzhou
  1

• 识别网络大家用的多大的数据集呀？ 准确率能够达到多少呀？ 我训练mjsynth，只能够到83左右

  #253 opened a year ago by LYHTZHANGJIANG
  0

• loss:nan怎么解决？

  #260 opened 2 years ago by nissansz
  1

• 同样的数据，代码，在colab训练没有nan错误，在win10上就会出现nan错误，是怎么解决？

  #261 opened 2 years ago by nissansz
  1

• 没看到保存的pth文件

  #259 opened 2 years ago by nissansz
  1

• 启动训练报这个错误怎么解决？图片的label txt文件里，相对路径在哪里设置？

  #258 opened 2 years ago by nissansz
  2

• 使用提供的预训练模型进行文本检测出错

  #251 opened 2 years ago by LinXin04
  0

• 训练识别网络的问题验证准确率只有0.3

  #252 opened 2 years ago by LYHTZHANGJIANG
  1

• 为何训练时精度高，对训练集推理就全错？

  #249 opened 3 years ago by Tomhardy13
  1