Liang Qiao
Liang Qiao
目前暂时没有提供inference接口的实现。 你可以参考如果调用test.py的流程进行推理的，对新的需要预测的图片离线生成包含ocr结果的json，然后使用test.py脚本进行批量测试。
Could you provide the datalist you generated (you may send it by email)?
> Hi - thanks very much. I've just sent now I have not received the email yet. The file might exceed the capacity limit of the email. Could you please...
warp_ctc由于是一个第三方库，它的安装跟cuda版本有很大的关系，安装问题可以看看DAVAR-Lab-OCR/docs/FAQ.md中的相关说明能不能解决。 这个问题如果暂时无法解决，如果你不使用text_recognition中的CTC模型，可以简单的将相关的引用从__init__.py中去掉，不会影响其他模块的运营。
You may customize the input image scale by modifing the parameter in data pipeline. See 'demo/table_recognition/lgpma/configs/lgpma_base.py' (L164 controls the training scales, L182 controls the testing scales). The module implementation is...
Current repository mainly serves for research. It follows the architecture of mmdetection and is suited to the deployment components it provides. You can follow the guide of [mmdetection](https://mmdetection.readthedocs.io/en/v2.20.0/tutorials/pytorch2onnx.html)
You can download them follow the instruction in https://github.com/hikopensource/DAVAR-Lab-OCR/tree/main/demo/text_spotting/datalist
> IC13/IC15/TotalText train/CTW with transcript are all missing from that link I've just checked the link, and there are 20 datalists in the directory, including all the lists you mentioned....
MANGO is a pure single-staged end-to-end framework，where the recognizor is predicted based on the attention of the global feature map. The detector here is only used to do coarse location...
we will release this end-to-end model's config later