CCNet-Pure-Pytorch icon indicating copy to clipboard operation
CCNet-Pure-Pytorch copied to clipboard

Published 20 hours ago • verified publisher icon Serge-weihao
→

Metadata

Criss-Cross Attention (2d&3d) for Semantic Segmentation in pure Pytorch with a faster and more precise implementation.

Results 14 CCNet-Pure-Pytorch issues
Sort by recently updated
recently updated
newest added

关于网络最后的return[x, x_dsn]

5 comment

您好，感谢您的代码实现。有些疑问想请教您一下： 1. 网络最后的return[x, x_dsn]是什么意思，返回一个list?在论文中，最后不该是一个cat吗？ 2. 网络最后的输出，如果不经过上采样的话，如何恢复原图大小？

yearing1017 avatar yearing1017

precision

13 comment

have you compared the moiu between your cc.py and the original code on cityscapes ??? becauese i have test your cc.py,and .....

swjtulinxi avatar swjtulinxi

为什么CCNet输入和输出总是一样的呢？

2 comment

![图片](https://github.com/Serge-weihao/CCNet-Pure-Pytorch/assets/66677296/5dc27034-6fc7-4130-8c3e-0ce833a46a35)

lijing-coder avatar lijing-coder

About 3D CC Attention module

can anyone explain what is different between CrissCrossAttention3D, CrissCrossAttention3D1, CrissCrossAttention3D2 ?? Thanks

adchentc avatar adchentc

← Metadata

181
Stars
20
Forks
Watchers

Owner

verified publisher icon Serge-weihao

Metadata

Criss-Cross Attention (2d&3d) for Semantic Segmentation in pure Pytorch with a faster and more precise implementation.

Back