yxgeee / pytorch-fpn

Feature Pyramid Networks written by Pytorch

Hi there 👋

• 🌱 I’m currently a Principal Researcher at Tencent ARC Lab.
• 🔭 I’m currently working on vision and multimodal foundation models.
• 👯 I’m looking for self-motivated interns to collaborate on related research topics.
• 📫 Reach me at my homepage.

pytorch-fpn's Issues

关于新建工程

您好，好像您这个工程是自己新建的，而不是直接从ruotianluo那里fork过来的。因为我看您的branch中只有两条分支，所以我猜测这个工程应该您自己新建的吧？
请问这种不需要fork但又能加载其他开源工程的做法是怎么实现的？

about the result

I wonder the result after u add FPN.
Did that work or not? What's your result by using VOC_2007_test?
Thanks for your reply~

I can't download the pretrained model resnet.

MAP

why my MAP is only 46?

Why bbox loss is always 0?

I have trained FPN and the bbox loss is always 0, except the first one hundrand iterations.
I have not found the problem.
Could you please give me some advice? Thank you.

loss

When I train FPN, what if loss does not decrease at a certain value？

怎样添加FPN到faster rcnn中

你好，我想把FPN添加到faster rcnn中去，但是我不是很清楚应该修改faster rcnn的哪些内容，你能告诉我应该怎么修改吗，万分感谢！