cvlab-yonsei/lba

An official PyTorch implementation of the paper "Learning by Aligning: Visible-Infrared Person Re-identification using Cross-Modal Correspondences", ICCV 2021.

lba's Issues

Problems about the performance

I have run your source code on both SYSU and RegDB datasets, but I didn't get the performance of your paper.
So I want to know how to set the hyper-parameter to get the performance of your paper?

Question about the training speed

Thanks for your work.

When I tried to reproduce your results with an Nvidia 2080Ti (as recommended by the paper), however, the training speed seemed very slow. It nearly took 20 minutes for each epoch on SYSU-MM01, which mismatched with the reported 8 hours training time.

I have already used cuda for acceleration. Thus, I wonder how did this happen. Thank you.

When running "train. Py", there is a problem on line 132 of the "model. Py" file:

When running "train. Py", there is a problem on line（loss = torch.mean(comask_pos * self.criterion(feat, feat_recon_pos, feat_recon_neg))） 132 of the "model. Py" file:
Traceback：RuntimeError: The size of tensor a (9) must match the size of tensor b (18) at non-singleton dimension 3

Visualization problem

Hello，
Thanks for your great work, I am wondering about the visualization part, use mask and comask matrix in SYSU-MM01 dataset.
Can I get some details about the steps of your visualization method?
Thank you very much.

something about run this code

thanks for your code, there is something wrong when i run you code,in this line:
loss = torch.mean(comask_pos * self.criterion(feat, feat_recon_pos, feat_recon_neg))
the wrong is:RuntimeError: The size of tensor a (9) must match the size of tensor b (18) at non-singleton dimension 3
could you give me some help?

How to visualize the heat map？

Excuse me! I am interested interested in generating the heat map in this paper, there exists the code in this repository?