NeurIPS 2020

Make One-Shot Video Object Segmentation Efficient Again

Meta Review

All four reviewers feel borderline about the paper. Everyone was positive about the idea of combining MAML-inspired meta-learning for one-shot video object segmentation. Reviewers shared concerns that they would have like to see more analysis by the paper to offer more insight and understanding to the gains offered by the proposed method. The authors are recommended to take advantage of the feedback from the reviewers to revise their paper for the camera-ready.