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

NCNN & SphereFace #153

Open
JackyWangCN opened this issue Sep 29, 2017 · 3 comments
Open

NCNN & SphereFace #153

JackyWangCN opened this issue Sep 29, 2017 · 3 comments

Comments

@JackyWangCN
Copy link

有这方面模型结合的示例吗? 或者 dlib?
想知道一下相关测试的性能效果。

@Pinnh
Copy link

Pinnh commented Apr 9, 2018

I only test the speed of sphereface20 may be, cost about 360ms to extract feature
the accuracy seems get better,some other model may get worse

@xxspurs
Copy link

xxspurs commented Jun 21, 2018

I have trouble on converting my sphere face model of caffe to ncnn. It can run, but it gives out completely different result and can not classify correctly. May I see your code? Thank you.

@Pinnh
Copy link

Pinnh commented Jun 22, 2018

There are many check points of correct classification, model convert is one respect, it's hard to say where issue happens.

My suggestion is: check your face align first(which face points are detected by mtcnn), if you only test one same input between PC and Android skip this, then check your preprocess ,you may lost substract_mean_normalize, then check your output especially ncnn output.

You can show your code here or you can get detail help.

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