Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

训练h5文件的对应顺序 #12

Open
zhanglaplace opened this issue Nov 17, 2017 · 7 comments
Open

训练h5文件的对应顺序 #12

zhanglaplace opened this issue Nov 17, 2017 · 7 comments

Comments

@zhanglaplace
Copy link

你好,请问下,训练的h5文件对应的是该图片的
xmin ymin xmax ymax x1 y1 x2 y2 .....x68, 68吗?
多谢了,想自己学习训练下

@panda-lab
Copy link
Owner

@zhanglaplace 是的

@zhanglaplace
Copy link
Author

@lsy17096535 嗯,谢谢,那坐标都是图像的绝对坐标吗,按照prototxt里面,图像resize为60*60后,这些point应该是相对值吧,具体的参考物又是什么呢,麻烦您了

@zhanglaplace
Copy link
Author

@lsy17096535 你好,我看了下测试的代码,网络输入是6060,但是预测的ratio是针对缩放前的人脸,
那么在训练的时候,需要对图像人脸都resize成相同的60
60吧;
还是训练图片是原图,和rect坐标,这样子网络的输出就会维度不一致。麻烦解答下,现在有点疑问

@panda-lab
Copy link
Owner

@zhanglaplace 训练的时候也要60x60

@zhanglaplace
Copy link
Author

@lsy17096535 训练的时候也要60*60的话,输入网络的就是一个face和68个point的相对坐标了,对于point超出face的只能标注为<-1或者>1的吧,然后h5的顺序应该是ImagePath x1 y1 x2 y2 .....应该就没有face坐标的事情了。多谢解答了

@arvrschool
Copy link

@zhanglaplace 你训练的数据集收敛了吗?效果如何?我使用了8w张训练,也不收敛,test loss一直在0.5左右,train loss 在 0-1之间。不确定什么原因,希望能够得到解答,谢谢

@angeladzl
Copy link

@arvrschool loss问题,微信交流下,Djiangnan007

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants