[2022 spring] Arxiv 2022 VRT (20213143)

[wonjeongchoi]

Fist of all, thank you for your great and detailed review of VRT method. Actually, this is my first time to get to know the transformer used in video vision task. It's very interesting and I appreciate you for letting me know this powerful and recent method.

I have some comments to your review.

1. In the problem definition part, it would be better if some detailed explanations about the covered task such as video deblurring, super resolution are provided. I expect that those are similar to image tasks but I wonder how it is different in video task.
2. Do the LQ and HQ mean Low-Quality and High-Quality?
3. In the method part, it is quite hard to understand the Mutual Attention section. It would be better if you explain that part with more understandable sentences and wordings.

Thank you!

[Junhyeon-Park]

좋은 리뷰 감사합니다. 내용 정리가 매우 잘되어 있고, Video Restoration에 사용된 Attention 관련 내용을 자세히 정리 해주셨습니다. 특히 Method 내용이 차례대로 정리가 매우 잘 되어 있어 좋았습니다. Experiment의 Comparison의 Figure 6의 크기 조절 혹은 다른 시각적 결과도 추가되면 더 좋을 것 같습니다.