Diss Back

[yu-changqian]

Hello,

I am the author of this paper BiSeNet.

First, thanks for your awesome practice. The tricks mentioned in the README file are our common tricks in the implementation of semantic segmentation algorithms.

However, I don't agree with your "diss" part, because the result you mentioned is the performance compared with other non-real-time algorithms. Actually, this paper, the several-month hard work of me and my collaborators, mainly focuses on the real-time scenario, which is also our original motivation. It is precise because of the limit of the real-time scenario, we design the two-branch architecture. In the non-real-time scene, without consideration of the computation resource, we didn't need to design this type of architecture. Furthermore, for the real-time result, we didn't use the evaluation tricks, like multi-scale and flip testing. If you read this paper carefully, I think you would not make these mistakes. Finally, due to the request of the reviewers and to validate the effectiveness of this architecture, we also make a detailed comparison with other non-real-time algorithms.

Next, I admit that this paper is not a significant improvement. However, the advance of the research community depends on the hard work of each researcher. I think each of us, as a qualified researcher, should show enough tolerance and respect to these efforts. Of course, I also sincerely hope each researcher, including you and me, can make major progress to push forward the whole community.

Finally, send one quote I just saw tonight to you: "Be kind, always. Everyone you meet is fighting a battle you know nothing about."

[CoinCheung]

Hi,

The diss is not a serious one, and not target on your contribution to the community. Readme is also not a formal place to publish options like articles.

I appreciate your paper and the idea is inspiring. The paper is accepted, which obviously verifies the contribution and hard work. Maybe people are more easily persuaded by my 'rap' than I thought. I agree with you that researchers should be tolerant and encourage different opinions.

I did not expect this repo and the readme would even draw the attention of the author. Anyway, if you feel it is not proper, leave a message and I will make a statement in the readme to clarify the motivation of my 'rap'.

PS: you went to bed really late, which is not good for your health. Maybe you should take more care of your health, it is as important as research :)

[mengzhibin]

Hello, @ycszen @CoinCheung , I found something wrong, when changed sync bn to apex bn in your code @CoinCheung , miou will reduce to 74 miou. But torchseg of @ycszen is using sync bn of apex. I dont know what's the difference of inplace bn and apex sync bn. Thanks you two anyway for your work.

[ShihuaHuang95]

When you use a deeper and larger backbone (e.g. R101), things are different and maybe the Spatial Path is necessary to provide detailed spatial information as it has small receptive filed. By the way, thanks for the re-implementation @CoinCheung.

[MengzhangLI]

When you use a deeper and larger backbone (e.g. R101), things are different and maybe the Spatial Path is necessary to provide detailed spatial information as it has small receptive filed. By the way, thanks for the re-implementation @CoinCheung.

In my opinion, res101 would made it useless because it is proposed for real-time scenery.