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iou-aware分支在inference时的问题 #16

Open
hnzjuer opened this issue Dec 11, 2020 · 2 comments
Open

关于iou-aware分支在inference时的问题 #16

hnzjuer opened this issue Dec 11, 2020 · 2 comments

Comments

@hnzjuer
Copy link

hnzjuer commented Dec 11, 2020

你好,我想问下在ppyolo中添加的iou-aware分支,在inference时,原论文中的iou confidence是通过α=0.4次方的形式与class的confidence的0.6次方相乘,得到inference时的confidence。
但是在代码中,yolodet/models/heads/ppyolo.py中的get_eval_bboxes函数:
new_obj = torch.pow(obj, (1 - self.iou_aware_factor)) * torch.pow(ioua, self.iou_aware_factor)
实现论文的new_obj以后还进行了-log(1/new_obj-1)的操作。
我这边实现了iou-aware后假设原new_obj为0.9,但是经过-log(1/new_obj-1)后续操作得到的confidence数值是2.19,造成了识别的confidence的范围不在0到1以内。
想请教下为什么代码中这样实现,还是说在其他地方有将其重新约束到(0,1)的操作?

@wuzhihao7788
Copy link
Owner

该部分代码主要多的参考百度paddlepaddle对于ppyolo的论文浮现。在loss方法中有对值的约束。注:paddlepaddle对ppyolo的浮现,在网络结构上与原论文有所不同

@wenchima
Copy link

wenchima commented Dec 1, 2021

I have the similar problem. When I enabled the α=0.4, it seems the output AP would be hurt severely. Do you have any idea how to fix this?

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

3 participants