Some Problems on Reproducing the Results on cub200 about antialiased-cnns HOT 2 OPEN

Thanks for experimenting on this! Not really a "reproduction" problem, as we didn't have these results in the paper.

Accuracy may go up or down for different datasets. The shift-invariance should be improved more reliably!

from antialiased-cnns.

Hi, I want to train antialias-cnn on cub200 dataset. I use ResNet-50 backbone.

However, I do not see increase of accuracy with BlurPool after several experiments. Do you have any suggestions for antialias-cnn on fine-grained datasets (e.g. cub200)?

Thank you.

Expected Behaviour

Actual Behaviour

Reproduce Scenario (including but not limited to)

Steps to Reproduce

Platform and Version

Sample Code that illustrates the problem

Logs taken while reproducing problem

Can you post the performance on the CUB200?
Thanks.

from antialiased-cnns.