Coarse-to-Fine Semantic Video Segmentation Using Supervoxel Trees | IEEE Conference Publication | IEEE Xplore