Can you build an image classification model that can discriminate between real and AI generated images?

:hugs: HuggingFace just launched its first competition.

The dataset contains roughly 31000 images, some created by AI. Your goal is to construct a model that distinguishes AI-generated images. I’ve created a starter notebook using SuperGradients to help kick-start your participation in the competition (trying to get to 700 stars in the next two days, help us out we’re less than 50 away).

Here’s a link to the notebook, which I ironically created using a Kaggle notebook because I ran out of Colab GPU compute units.

If you have questions, let me know!