Combining-EfficientNet-and-Vision-Transformers-for-Video-Deepfake-Detection icon indicating copy to clipboard operation
Combining-EfficientNet-and-Vision-Transformers-for-Video-Deepfake-Detection copied to clipboard

Published 20 hours ago • verified publisher icon davide-coccomini

test AUC

Open zxq99799 opened this issue 1 year ago • 3 comments

Hello, I trained a total of 110000 real images and 100000 fake images using dfdc and ff++, but the final test only achieved an AUC of 0.885. Can you give me some suggestions. Thank you.

zxq99799 avatar Jun 11 '23 05:06 zxq99799

Hi, which version of the model are you using?

davide-coccomini avatar Oct 02 '23 14:10 davide-coccomini

efficient-vit This is the model i'm using.

zxq99799 avatar Oct 07 '23 06:10 zxq99799

The EfficientViT obtain an AUC of 0.919 which is not too far from yours. If I understand well, your training set is not complete so it is normal to obtain different results. Anyway, if you want to improve more, I suggest to use Cross Efficient ViT which is our main method.

davide-coccomini avatar Jan 02 '24 14:01 davide-coccomini