Computer Vision News - July 2020

49 How to Write a Good Review system is a bit more complex than “just filling out the form” , and not every reviewer is aware of its intricacies. Yet, there currently is no mechanism in the vision community to educate the new generation of reviewers that we so desperately need. This is why Torsten and Laura decided to organize a tutorial for CVPR on “How to write a good review?”. There probably is no single formula for writing a review as each review should be uniquely tailored to the paper it describes and will depend on what the reviewer finds important. The tutorial reflected this by having a team of excellent speakers, both junior and senior, share their own best practices for the full “lifetime” of a submission: how to write a good paper, a review, a rebuttal, and finally, how it is all used to decide whether to accept or reject a paper. The tutorial was opened by Bill Freeman, who talked about how to write a paper. Communicating the main ideas and contributions of your paper is a hard task, and often failure to do so clearly leads to poor ratings by reviewers. Thus, spending a significant amount of time on writing the paper, with a focus on making it as easy as possible to read, is highly important and it is better to publish a few very good papers than many less- impactful ones. The next session focused on writing reviews. Rick Szeliski used his experience in both computer graphics and computer vision to compare the review processes in both communities, highlighting that SIGGRAPH gives reviewers much clearer instructions. Another main focus of his talk was that we as reviewers owe it to the authors to carefully review their papers, just as we expect that other reviewers carefully read our own submissions . "… critical thinking, constructive input, empathy" Best of CVPR 2020 Laura Leal-Taixé