search

jeya-maria-jose / UNeXt-pytorch

Official Pytorch Code base for "UNeXt: MLP-based Rapid Medical Image Segmentation Network", MICCAI 2022
https://jeya-maria-jose.github.io/UNext-web/
MIT License
459 stars 76 forks source link

Performance on multi-class segmentation. #8

Closed Capchenxi closed 2 years ago

Capchenxi commented 2 years ago

Hello, Thanks for your excellent idea on shifted mlp module. I have some questions on other segmentation task other than the medical segmentation. I noticed that the code you provide is mainly focusing on 2-class segmentation task. Is there any possible that you tried the UneXt structure on multi-class task like Cityscapes? I have tried this structure on my own task, which is a 12-class segmentation task for parking lot scene, but it performs poorly. So I just wonder if the UneXt is experimented on Cityscape dataset or other multi-class dataset and how it performs? Thanks.

jeya-maria-jose commented 2 years ago

Hi, I have not experimented UNext with Cityscapes. My guess is that we might need more representative power to work on more complex datasets.

Capchenxi commented 2 years ago

Hi, I have not experimented UNext with Cityscapes. My guess is that we might need more representative power to work on more complex datasets.

Thanks for your prompt reply. I'm now trying to change the backbone of unext into other complex structures. Hope it helps.