search

patrikhuber / superviseddescent

C++11 implementation of the supervised descent optimisation method
http://patrikhuber.github.io/superviseddescent/
Apache License 2.0
402 stars 188 forks source link

Confidence score #24

Closed golunov closed 8 years ago

golunov commented 8 years ago

Hi, Patrik!

I'm fascinated by your work!

Did you see confidence score in Intraface face alignment? Is it possible to implement it here?

I'll appreciate your answer. Thanks.

patrikhuber commented 8 years ago

Hi,

Thanks!

Intraface do not describe how they do the confidence. We tried stuff like train a classifier using the HOG features from training or a regressor using the distance measure to the ground truth, and it all kind of worked, but I did not have the time to fully train it. If it's of interest to you, it's rather easy to train though. If you want to do it with a classifier, you'll need to hook up superviseddescent to something like liblinear, libsvm, or tiny-cnn. But just using the regression framework in superviseddescent also didn't work too bad.

golunov commented 8 years ago

Thanks for reply!

I've tried to learn svm classifier and it worked. I've got TPR~0.78 | FAR=0 and TPR~0.95 | FAR=0.01. I think it is not the best results, but it's enough for my research.

patrikhuber commented 8 years ago

Cool! That sounds good. I'll close this issue, feel free to continue/reopen if you have questions or want to share anything.