[2022 spring] ICCV 2021 LabOR(20213320)

[sky0701]

좋은 리뷰 감사합니다. PPL와 SPL 사이의 성능 차이가 어디서 기인하는지 궁금합니다. 두 방법에서 실질적으로 라벨링 하는 영역의 면적이 같을때에도 성능 차이가 생길지 궁금하고, 이에 대해 어떻게 생각하시는지 궁금합니다.

[wonjeongchoi]

Thank you for your good review with some useful notes in figures.
I think your review has good structure and detailed explanation.

However, sentences in your korean version review are so difficult to understand for me.
I guess it is a translated version of english review by using some translator systems..
The overall sentences are quite unnatural and they often have needless punctuation marks.
First of all, I highly recommend you to revise your korean version review with understanable 'korean' sentences.

In the details of methods section(3.2), it would be better if some additional explanations of each loss function are provided.
There are quite a lot of losses and each loss has distinctive goal to train model. How about providing what they do and how they work?

Thank you