rec模型，同部首易错字，是放到一行进行训练好，还是2个一组分开多行进行训练好？例如：奇绮畸崎椅倚

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

38.99k stars 7.32k forks source link

rec模型，同部首易错字，是放到一行进行训练好，还是2个一组分开多行进行训练好？例如：奇绮畸崎椅倚 #11993

Closed nissansz closed 2 weeks ago

nissansz commented 3 weeks ago

请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem

系统环境/System Environment：
版本号/Version：Paddle： PaddleOCR：2.7.5 问题相关组件/Related components：
运行指令/Command Code：
完整报错/Complete Error Message：

同部首易错字，是放到一行进行训练好，还是2个一组分开多行进行训练好？例如：奇绮畸崎椅倚

Sunting78 commented 2 weeks ago

您好，搭配语义比较好不建议把单字放在一起