文字识别后返回单字识别坐标目前paddleocr支持吗 #12227
Replies: 6 comments
-
支持的,现在PPStructure已有返回单字坐标的功能,可以参考PR: #10515 如果想在普通的识别中使用该功能的话,只需修改一下部分源码即可 |
Beta Was this translation helpful? Give feedback.
-
thanks,i will have a try |
Beta Was this translation helpful? Give feedback.
-
您好,如果想在普通的识别中使用该功能的话,有具体修改好的代码工程吗,或者可以说一下具体修改哪里吗 |
Beta Was this translation helpful? Give feedback.
-
我改了一下:
4.修改tools/infer/utility.py,增加了一个参数word_boxs,用于画单个字的矩形框
一定要修改ppocr2.6,由于改的时间较长,有点忘记改哪了,有问题请留言 参数如下: |
Beta Was this translation helpful? Give feedback.
-
好的谢谢您,有问题再请教您,万分感谢 |
Beta Was this translation helpful? Give feedback.
-
这种后处理方式,如果文字间隔比较大的话,单个文字定位就很差 |
Beta Was this translation helpful? Give feedback.
-
文字识别后返回单字识别坐标目前paddleocr支持吗
Beta Was this translation helpful? Give feedback.
All reactions