search

Chasel-Tsui / mmdet-rfla

ECCV22: RFLA
MIT License
254 stars 23 forks source link

如何快速移植 #1

Closed sdreamforchen closed 3 weeks ago

sdreamforchen commented 2 years ago

校友，您好。想请问下，我想移植label assign部分到我的pytorch代码里面。想请问下，我应该关注哪些代码或者python文件？ Thanks！

Chasel-Tsui commented 2 years ago

校友，您好。想请问下，我想移植label assign部分到我的pytorch代码里面。想请问下，我应该关注哪些代码或者python文件？ Thanks！

双阶段算法主要关注mmdet/core/bbox/assigner下的assigner文件，mmdet/core/bbox/iou_calculators下的距离计算文件，以及mmdet/core/anchor下的rf生成文件；单阶段除这些外还有mmdet/models/dense_heads下相应的detection head

davidsvy commented 7 months ago

Hello @sdreamforchen. Have you succeeded adding RFLA to your codebase? If so, can you please tell me how you estimated the receptive field (either theoretical or effective) for intermediate layers of different architectures?