Compositional Video Prediction

Yufei Ye, Maneesh Singh, Abhinav Gupta +1

We present an approach for pixel-level future prediction given an input image of a scene. We observe that a scene is comprised of distinct entities that undergo motion and present…