Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

proposal对齐问题 #33

Open
he010103 opened this issue Sep 19, 2019 · 7 comments
Open

proposal对齐问题 #33

he010103 opened this issue Sep 19, 2019 · 7 comments

Comments

@he010103
Copy link

代码中有一步是对 global_feat 和 排序的part_feat concat,然后进行200类的分类。但是不同图片的part_feat是按照得分来排序的,不是按照身体部位来排序的,所以存在特征不对齐的问题。有可能图a是: 头,腿,胸,身体,图b是胸,头,腿,身体。我这里尝试把concat换成add会训练更稳定一点,性能有小的提升!

@RYYAI
Copy link

RYYAI commented Oct 11, 2019

请问有多大的提升?

@he010103
Copy link
Author

@RYYAI 能到88%

@ZF4444
Copy link

ZF4444 commented Oct 16, 2019

@he010103 你好,想向你请教一下你说的add操作是简单的几个part和global的2048维特征向量相加吗?谢谢!

@he010103
Copy link
Author

@ZF1044404254 我这边是直接相加的,比起性能小幅提高,更重要的是训练波动更小。

@ZF4444
Copy link

ZF4444 commented Oct 17, 2019

@he010103 谢谢你的分享

@liu826250634
Copy link

@he010103 能把代码贴上来么,谢谢了。

@atnegam
Copy link

atnegam commented Apr 17, 2022

@liu826250634 你试过吗,有效果吗?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants