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)
https://paddlepaddle.github.io/PaddleOCR/
Apache License 2.0
44.05k stars 7.81k forks source link

很久没有增加新的模型了，能否把MixNet加进来？https://github.com/D641593/MixNet #12011

Closed cnszmike closed 6 months ago

cnszmike commented 6 months ago

背景

经过需求征集https://github.com/PaddlePaddle/PaddleOCR/issues/10334 和每周技术研讨会 https://github.com/PaddlePaddle/PaddleOCR/issues/10223 讨论，我们确定了XXXX任务。

解决步骤

  1. 根据开源代码进行网络结构、评估指标转换。代码链接：XXXX
  2. 结合论文复现指南，进行前反向对齐等操作，达到论文Table.1中的指标。
  3. 参考PR提交规范提交代码PR到ppocr中。
changdazhou commented 6 months ago

已经将您的需求记录，我们会进行讨论评估