zhangying

Thanks for your sharing!

Same question！ Good work for your paper: )

please see the issue https://github.com/dbolya/yolact/issues/334, hope it can help you : )

I think the training time depends on your GPU, hahaha~ My training data is about 2000 images, I use the resnet101 backbone, 200 epoch is ok for my situation~

@hyunas1996 Since I can't see your code, I guess there are two possible reasons for your problems: - 1.Try to comment the three lines code, I think the three lines...

same question

@LDOUBLEV 大佬， 想问下使用paddle lite在x86上部署，目前v3 det 和 rec都已经跑通，但如果运行system的预测就会报double free的错，我打了下log， **发现是predictor_crnn->Run() 这个操作在识别第二个boxes 的文本行报错，进行了一下debug** （1）system输入的图像由多行文本行换成单行文本行的图像，不报错，程序正常运行 （2）把ocr_db_crnn.cc中以下代码boxes[i]换成boxes[boxes.size() -2]，也不会报错，程序正常运行 ` for (int i = boxes.size() - 1; i >= 0; i--) { auto preprocess_start =...