Visual-Attention-Network / SegNeXt

Official Pytorch implementations for "SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation" (NeurIPS 2022)
Apache License 2.0
795 stars 85 forks source link

The code implementation does not match the picture in the paper #22

Closed pb585 closed 2 years ago

pb585 commented 2 years ago

Thank you for your job, I have a question about code. In backbones/mscan.py file, there is a "shortcut" operation in SpatialAttention module but there is a "drop_path" in Block module, It should work the same as “shortcut”。 The first shortcut operation I didn't find in the paper。

MenghaoGuo commented 2 years ago

We follow the setting of VAN : https://arxiv.org/pdf/2202.09741.pdf. Here is a trick in VAN experiment, I do not know whether it is important in segnext.