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

关于样本分配的顺序问题(或者说训练为什么可以收敛?) #113

Open
feiyuhuahuo opened this issue Mar 22, 2023 · 1 comment

Comments

@feiyuhuahuo
Copy link

感谢您的工作,这个问题想请教下:
IAM去预测100个实例,但是样本分配时直接使用cost来匹配,也就是说没有基于位置的先验,那IAM是如何决定预测的顺序的呢?具体哪些通道去预测图片的哪个目标,这个是如何决定的呢? 或者换个角度讲,训练一开始 , 用来二分匹配的cost是随机的,这导致IAM一开始以随机的顺序去学习,等到下个epoch时,IAM学习的顺序就可能变了,这样的话模型也能顺利收敛吗?

@feiyuhuahuo
Copy link
Author

又思考了下这个问题,其实这个和DETR的样本分配差不多,训练一开始,没有先验,收敛可能会受到阻碍,但一直训练下去,最终还是能收敛的。网络最终自己会学习到和目标位置相关的表示,训练时手工assignment可能没必要。 手工设计的assignment可能就是不如网络自己学习到的表示,而且还不够end-to-end.

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

1 participant