search

penincillin / DREAM

This is the public repository for our accepted CVPR 2018 paper "Pose-Robust Face Recognition via Deep Residual Equivariant Mapping"
http://mmlab.ie.cuhk.edu.hk/projects/DREAM/
BSD 2-Clause "Simplified" License
389 stars 98 forks source link

Have you ever tried on image with different pitch? #14

Closed inkzk closed 6 years ago

inkzk commented 6 years ago

I have been training DREAM on different pitch angle images for a long time, and it still seems not a feasible work. Have you ever tried on pitch angles?

penincillin commented 6 years ago

The same question you asked was also quest by one of reviewers. We believe our simple idea can also work on the pitch issue. However, the most pivot problem is the dataset. The yaw distribution on MS-Celeb-1M and the propose of CFP benchmark allow us to conduct a relatively solid experiment. If you wish to experiment on the pitch, you may build your benchmark and find a proper training data. Then you can test on our code or even propose your further method

penincillin commented 6 years ago

One of my colleagues tried our algorithm on pitch angle and gains performance enhancement.