Learning depth-sensitive conditional random fields for semantic segmentation of RGB-D images | IEEE Conference Publication | IEEE Xplore