search

Serge-weihao / CCNet-Pure-Pytorch

Criss-Cross Attention (2d&3d) for Semantic Segmentation in pure Pytorch with a faster and more precise implementation.
MIT License
183 stars 21 forks source link

Calculating loss #10

Open shrutishrestha opened 3 years ago

shrutishrestha commented 3 years ago

In the ccnet code. At the end of resnet there is out which gives an output of 33X33 sized image. But it is then passed into the self.criterion(outs, labels). But the labels are of original size i.e. 769*769. So, how can it be computed? I am getting the "CUDNN_STATUS_INTERNAL_ERROR" error. I have attached the same in the snapshot below.

ccnet
Serge-weihao commented 3 years ago

@shrutishrestha https://github.com/Serge-weihao/CCNet-Pure-Pytorch/blob/master/loss/criterion.py

Serge-weihao commented 3 years ago

@shrutishrestha I remember that the out stride is 8 and the output size is 92?