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)
请问各位大佬,我现在有一些长宽比超过20的图像,例如6400*280,这种图像在做det的时候应该如何设置训练参数呢,感觉检测的效果很差