Pretrained model doesn't give the predictions shown in the samples

[Hisan-007]

I used the pre-trained model to get the predictions on the test set without any changes. Even though i used everything as it is, i'm getting very bad results with the pre-trained model. I used the images provided in the pf-afn-test/dataset folder.

Here are the results:

I have no idea why this is happening, the results on their own provided test set should be similar to the samples images they have shown us.

[geyuying]

#3

[devkaranjoshi]

Can you give more details about your problem?

[albertigno]

I'm getting the exact same performance... I'm using a more up-to-date environment, but that shouldn't be a problem, isn't it? I'm running it on a RTX3060 with cuda 11.6, cupy 12.1.0 pytorch 1.12.0