FEELVOS: Fast End-To-End Embedding Learning for Video Object Segmentation | IEEE Conference Publication | IEEE Xplore