Software that can generate photos from paintings, turn horses into zebras, perform style transfer, and more.

Image-to-image translation with conditional adversarial nets

Densely Connected Convolutional Networks, In CVPR 2017 (Best Paper Award).

TripletLoss used in Google's FaceNet paper

Implementation of our paper - Recurrent Convolutional Network for Video-based Person Re-Identification

Code for our ICCV 2017 paper -- Jointly Attentive Spatial-Temporal Pooling Networks for Video-based Person Re-Identification