search

meetps / pytorch-semseg

Semantic Segmentation Architectures Implemented in PyTorch
https://meetshah.dev/semantic-segmentation/deep-learning/pytorch/visdom/2017/06/01/semantic-segmentation-over-the-years.html
MIT License
3.39k stars 796 forks source link

pspnet training #243

Open kingloo2014 opened 4 years ago

kingloo2014 commented 4 years ago

hello, when I trained the pspnet, I found the ouput of this network has 2 tensors. So, the loss function cannot accept this output and lead to crash. Could you tell me which loss function should be used in this network ?

yyyjia commented 4 years ago

When training, there are two loss functions, only one in the test phase.

Da-He commented 4 years ago

hello, when I trained the pspnet, I found the ouput of this network has 2 tensors. So, the loss function cannot accept this output and lead to crash. Could you tell me which loss function should be used in this network ?

When using PSPNet, it seems that you should use "multi_scale_cross_entropy" instead of "cross_entropy" in your config file.