search

hszhao / SAN

Exploring Self-attention for Image Recognition, CVPR2020.
MIT License
748 stars 133 forks source link

ResNet-26, ResNet-38, ResNet-50 in comparison #16

Open xwjabc opened 3 years ago

xwjabc commented 3 years ago

Thank you for the repo! I can reproduce the SAN-10 pairwise / patchwise results. However, for the CNN baselines mentioned in this paper (i.e. ResNet-26, ResNet-38, ResNet-50), I wonder how they are trained. Do you simply replace the SAN network by the ResNet in this code base, keep the optimization settings and train? Or use some other repos such as https://github.com/pytorch/examples/tree/master/imagenet? Thanks!

Asherchi commented 3 years ago

how can I solve this problem? from lib.sa.modules import Subtraction, Subtraction2, Aggregation ModuleNotFoundError: No module named 'lib.sa',thanks for your reply.

negishubham commented 3 years ago

Hi Authors,

I have a similar question as Weijian Xu above. Can you please let us know about the baseline simulations for ResNet-26, ResNet-38 and ResNet-50, as in if you can provide the pre-trained models for same?

Thanks!