Evaluation scores for TUT and VGG datasets do not match the paper

[peterwisu]

Hi, thank you for sharing the code and checkpoints.

I tried to reproduce the results using the provided code and models. For most of the evaluation datasets, I was able to get scores that are very close to those reported in the paper. However, for the TUT and VGG datasets, the scores I obtained are different.

I used the same evaluation scripts and settings as provided. Could you please confirm if there are any additional steps or specific settings required for these two datasets?

Thanks!