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

Unlabeled training with datasets like megaface #6

Closed
mrgloom opened this Issue Oct 14, 2015 · 3 comments

Comments

Projects
None yet
2 participants
@mrgloom

mrgloom commented Oct 14, 2015

For larger datasets you can request http://megaface.cs.washington.edu/

@bamos

This comment has been minimized.

Collaborator

bamos commented Oct 14, 2015

Thanks for the reference! Training new models with megaface would be interesting, but the current training with triplet loss can't be used because the images aren't labeled.

@bamos bamos changed the title from Suggested Improvements to Unlabeled training with datasets like megaface Oct 14, 2015

@bamos bamos closed this Feb 17, 2016

@mrgloom

This comment has been minimized.

mrgloom commented Feb 18, 2016

What status of this ussue? On what data OpenFace trained?

@bamos

This comment has been minimized.

Collaborator

bamos commented Feb 18, 2016

We can't train on megaface directly since they don't provide labels. We train on CASIA+FaceScrub, the two largest labeled datasets I'm aware of.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment