affiliation not provided to SSRN
Video compression, Deformable convolution, Motion correction, Self-attention, Conditional encoding