tfzhou / ContrastiveSeg

ICCV2021 (Oral) - Exploring Cross-Image Pixel Contrast for Semantic Segmentation
MIT License
663 stars 86 forks source link

About simple reproduction #34

Closed xianxuan-z closed 2 years ago

xianxuan-z commented 2 years ago

Hello, there are too many files in the project, I would like to ask you if I just want to try to reproduce the simple results of the camvid dataset combined with the pixel contrast learning method, what should I go to see about model creation, loss function, What about the data preprocessing, and the .py files and .sh files for training and testing? I hope you can take the time to give me some simple guidelines, which will benefit me a lot, thank you!

tfzhou commented 2 years ago

Hi @xianxuan-z , very good suggestion. However, I am sorry that I will not have too much time for this. It will be great if you can do this thing, and I also can provide supports for possible issues.