Fetching the paper…

GRAtt-VIS: Gated Residual Attention for Auto Rectifying Video Instance Segmentation · Around