FusionSeg: Learning to Combine Motion and Appearance for Fully Automatic Segmentation of Generic Objects in Videos | IEEE Conference Publication | IEEE Xplore