Loss calculation is too slow

speedinghzl / DSRG

Weakly-Supervised Semantic Segmentation Network with Deep Seeded Region Growing (CVPR 2018).

MIT License

251 stars 36 forks source link

Loss calculation is too slow #18

Closed yassouali closed 5 years ago

yassouali commented 5 years ago

Hi,

First of all, thank you very much for providing the implementation of your paper. I am trying to recreate the results in PyTorch, but I have very slow training time due to the consistency loss computation, which uses pydensecrf, and it takes a lot of time for a moderately sized images (~ 356). Do you have any recommendations on how to speed things up, and can you please provide the time it took to train the network for in your case.

Thank you very much.

speedinghzl commented 5 years ago

Please refer to GPU version dense CRF. https://github.com/heiwang1997/DenseCRF. Hope it helps.

yassouali commented 5 years ago

Thank you very much, I see that the code is in C++, so I need to port it to python is that right, anyhow, thank you.