Read on Twitter
My #CVPR2021 review takeaways:
*Don’t oversell*
The CMT matching system gets pretty good these days. Your reviewers will be knowledgeable in your field. They will not be happy when you downplay their work and oversell your contributions.
*Don’t oversell*
The CMT matching system gets pretty good these days. Your reviewers will be knowledgeable in your field. They will not be happy when you downplay their work and oversell your contributions.
*Be transparent*
It’s a PLUS to discuss limitations and failure modes of your method. Don’t wait for reviewers to point them out.
It’s a PLUS to discuss limitations and failure modes of your method. Don’t wait for reviewers to point them out.
*Evaluation*
Make sure your evaluation is fair and accurate so that the results are credible. Check if your results match the results reported in previous papers. If not, explain why.
Make sure your evaluation is fair and accurate so that the results are credible. Check if your results match the results reported in previous papers. If not, explain why.
*Achieving the state of the art*
Don’t just show that your method outperforms others. Discuss, analyze, ablate, and help readers learn new things.
Don’t just show that your method outperforms others. Discuss, analyze, ablate, and help readers learn new things.
*Be kind*
The authors of the compared methods in your paper will be your reviewers. Treat them fairly.
The authors of the compared methods in your paper will be your reviewers. Treat them fairly.
*Double/dual submission*
Don’t think that it won’t get caught...
Don’t think that it won’t get caught...
