Improving Smiling Detection with Race and Gender Diversity

From a Google talk @ NIPS 2017:

Recent progress in deep learning has been accompanied by a growing concern for whether models are fair for users, with equally good performance across different demographics. (...) We measure race and gender inclusion in the context of smiling detection, and introduce a method for improving smiling detection across demographic groups. (...) Our best-performing model defines a new state-of-the art for smiling detection, reaching 91% on the Faces of the World dataset.


Want to receive more content like this in your inbox?