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

请教 #23

Open
gengyanlei opened this issue Nov 30, 2019 · 5 comments
Open

请教 #23

gengyanlei opened this issue Nov 30, 2019 · 5 comments

Comments

@gengyanlei
Copy link

作者,您好,请问您在训练数据时,有没有采用MTCNN 或者 Retinaface对人脸数据进行对齐裁剪呢?
期待您的回复,谢谢。

@vicwer
Copy link
Owner

vicwer commented Nov 30, 2019

如果你在推理时的流程是检测,关键点,对齐,年龄识别,那训练时候就需要对齐裁剪,如果推理时不做对齐,那训练时候就不需要裁剪,两者保持一致。检测时候用那个检测网络,年龄识别的人脸数据就用那个网络去做预处理。

@gengyanlei
Copy link
Author

@vicwer 如果不对齐的话,效果是不是差一些呢?

@gengyanlei
Copy link
Author

如果对齐了的话,那么您所提到的3个尺度是不是就不太适用了呢?一般对齐会紧贴着头进行裁剪,如果这样的话,需要预留一些空边。

@vicwer
Copy link
Owner

vicwer commented Dec 2, 2019

对齐之后裁剪之前扩充

@Hzzone
Copy link

Hzzone commented Feb 11, 2020

我觉得对齐会对性能有影响。原论文对齐之后裁剪三个尺度,保证人脸一定会裁到,而不对齐只 center crop 的话,就未必了。

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

3 participants