Human pose based video compression via forward-referencing using deep learning | IEEE Conference Publication | IEEE Xplore