Tramac / awesome-semantic-segmentation-pytorch

Semantic Segmentation on PyTorch (include FCN, PSPNet, Deeplabv3, Deeplabv3+, DANet, DenseASPP, BiSeNet, EncNet, DUNet, ICNet, ENet, OCNet, CCNet, PSANet, CGNet, ESPNet, LEDNet, DFANet)
Apache License 2.0
2.82k stars 581 forks source link

The jpu paramters not add to optim. #85

Open xing-yc opened 4 years ago

xing-yc commented 4 years ago

As the title say, I find the jpu paramters not add to optim. That makes the jpu module's paramters not change. But interenting things is: when I add the jpu paramters to optim, the results almost not change in mIoU and pixels acc. I not change other hyper-paramter, What a amazing result.

Tramac commented 4 years ago

It is possible, network with different modules may depend on different parameters, this paper maybe helpful for searching th optimal parameter.