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)
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
有个问题,crnn训练时,如果一个图片训练结果不正确,是不会写到模型的吧?只有正确结果才会更新到模型里?