search

mx54039q / fnm

Unsupervised Face Normalization with Extreme Pose and Expression in the Wild (CVPR2019).
126 stars 30 forks source link

About Train #19

Closed cLEARLoVE00 closed 4 years ago

cLEARLoVE00 commented 4 years ago

Hello，Author，thank you open source code！About how to train model I want to know the front list and profile list how to bulid,do you have some examples to show?

mx54039q commented 4 years ago

Thanks for your interest. The experiment settings (e.g. dataset construct) are listed in detail in our paper. You can get it in the Experiment Section.

cLEARLoVE00 commented 4 years ago

I have some question，I cannot understand your paper what is normal dataset,is front face or profile and front face?can you tell me how to set my normal dataset?

mx54039q commented 4 years ago

Sec 4.1 in our paper, it is saying that "In unconstrained experiment, the non-normal face set contains 297,369 face images from unconstrained dataset CASIA-WebFace [28], while the normal face set contains 5,000 face images (front pose and 20 illuminations of 250 identities) from Multi-PIE"