video super-resolution (VSR) aims to restore a sequence of high-resolution
(HR) frames from their low-resolution (LR) counterparts. Although some progress
has been made, there are grand challenges to effectively utilize temporal
dependency in entire video sequences. Existing approaches