NeurIPS 2020

SOLOv2: Dynamic and Fast Instance Segmentation

Meta Review

The reviewers found that this paper provides excellent results for image segmentation, building off of previous work. There was some disagreement among the reviewers on a few areas. The first of these is novelty -- always a tricky subject -- where some reviewers noted that the work appears to be incremental methodologically. However, given that the results provide a novel contribution to the field, I think that this is not a particularly strong drawback. There were questions raised about the clarity of the paper. I believe that these have been reasonably addressed by the author response -- and I have also read the paper myself to judge clarity, and feel that it is acceptable. That said, I do expect the authors to fully incorporate the feedback from the reviewers to further improve the paper, as promised in their author response.