search

PaddlePaddle / PaddleOCR

Awesome multilingual OCR toolkits based on PaddlePaddle (practical ultra lightweight OCR system, support 80+ languages recognition, provide data annotation and synthesis tools, support training and deployment among server, mobile, embedded and IoT devices)
Apache License 2.0
39.75k stars 7.38k forks source link

问题 #12069

Closed 51057913 closed 1 month ago

51057913 commented 1 month ago

我使用的图片是 table 使用代码 table_engine = PPStructure(show_log=True,lang='en',recovery=True) 得到空表格，去掉recovery参数报错 InvalidArgumentError: When step > 0, end should be greater than start, but received end = 0, start = 501. [Hint: Expected end >= start, but received end:0 < start:501.] (at ../paddle/phi/kernels/funcs/slice_utils.h:78) [operator < set_value > error] 使用中文图片和中文模型，layout设置为True,读不出来表格，更提不出文字。 result = table_engine(img) 返回为空 layout设置为False,根据源代码，ocr自动为false,自动全版面设置为表格，也提不出文字。

layout_model_dir换成中文推理模型picodet_lcnet_x1_0_fgd_layout_cdla_infer,也识别不出来表格和文字。 请问大佬们有什么解决方案？

GreatV commented 1 month ago

试试这里 https://github.com/PaddlePaddle/PaddleOCR/blob/main/ppstructure/docs/quickstart.md#22-python%E8%84%9A%E6%9C%AC%E4%BD%BF%E7%94%A8

51057913 commented 1 month ago

试试这里 https://github.com/PaddlePaddle/PaddleOCR/blob/main/ppstructure/docs/quickstart.md#22-python%E8%84%9A%E6%9C%AC%E4%BD%BF%E7%94%A8

谢谢，最终初步找出来是paddle的安装问题，以为run_check就够了，结果导入paddle打出来的tensor都是0。

UserWangZz commented 1 month ago

感谢讨论❤️❤️❤️