最新的模型中文识别和文字检测需要改进 #57
gk966988
started this conversation in
Show and tell
Replies: 1 comment 3 replies
-
|
文字识别和检测目前是交由PaddleOCR来完成的，可以先尝试使用更大的权重 |
Beta Was this translation helpful? Give feedback.
3 replies
-
|
不是呀，没有使用混合模式。最新的模型不是能够同时识别中文、英文和公式吗 |
Beta Was this translation helpful? Give feedback.
-
|
同样的疑问 |
Beta Was this translation helpful? Give feedback.
-
|
目前的识别模型确实能同时识别中英文，但是效果可能没有用段落识别来的好 |
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
原图
识别结果：
感觉中文识别效果不是很好，而且原图下方的有些文字好像没有检测到
Beta Was this translation helpful? Give feedback.
All reactions