search

charlesCXK / TorchSemiSeg

[CVPR 2021] CPS: Semi-Supervised Semantic Segmentation with Cross Pseudo Supervision
MIT License
504 stars 74 forks source link

The performance of default voc8.res50v3+.CPS #35

Closed yangkai798 closed 3 years ago

yangkai798 commented 3 years ago

thank you for release the code. we run the default config as your public in voc8.res50v3+.CPS, the best performance is 72.76, a little lower than the 73.20 in paper. have any other tricks? the default epoch is 34, the paper is 60? but your released log, 73.28 is also use 34 epoch.

charlesCXK commented 3 years ago

Hi, we train this model for 34 epochs with no additional tricks. Please try to use the Docker we provide to run these experiments (https://github.com/charlesCXK/TorchSemiSeg/issues/27).

yangkai798 commented 3 years ago

thank you. But I have installed with conda env as the doc. the version is same. have any other random factor? I have no permission to use nvidia-docker in cluster. the code and config is same between docker and this repo?

yangkai798 commented 3 years ago

thanks , its OK. I run it again, the performance is 73.32.

charlesCXK commented 3 years ago

The docker only provides the environment to run the experiments and no code is stored in it. Since you have reproduced the result, I will close this issue. Welcome to re-open it if you have any questions.