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

测试集输出结果全部是一样的 #36

Open
ustcluq opened this issue Mar 22, 2018 · 7 comments
Open

测试集输出结果全部是一样的 #36

ustcluq opened this issue Mar 22, 2018 · 7 comments

Comments

@ustcluq
Copy link

ustcluq commented Mar 22, 2018

如果不用预训练模型,初始化也设成了
weight_filler {
type: "xavier"
variance_norm: AVERAGE
}
但是loss降到2再也不降,对测试集来说输出的人脸特征点全部一样。

请问这是什么原因呢?谢谢!

@Happythinker
Copy link

@ustcluq 请问您是怎么解决这个问题的呢?

@Happythinker
Copy link

@ustcluq 我也是这个问题

@qiexing
Copy link
Owner

qiexing commented Jan 2, 2019

@Happythinker
weight_filler {
type: "xavier"
}
直接这样?我当时改成xavier就可以了

@Happythinker
Copy link

@qiexing 谢谢您的回复,我修改一下再试试, 请问您训练的最终loss大概为多少呢

@Happythinker
Copy link

@qiexing
weight_filler {
type: "xavier"
variance_norm: AVERAGE
}
是需要在您的train文件添加这些参数吗?还需要修改其他的吗?

@Happythinker
Copy link

@qiexing
得到的结果,无论人脸状态怎么样,关键点还是呈现正脸分布
请问有其他的建议可以改善,达到你的训练效果吗?

@StarIsFalling
Copy link

你们是不是修改了输入维度?我尝试着输入96x96的时候,也有这种问题,输出值完全一样。但输入改成了楼主的3x224x224,就不同脸有不同输出了

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