AI-ArtBench: An AI-generated Artistic Dataset

AI-ArtBench is a dataset that contains 180,000+ art images. 60,000 of them are human-drawn art that was directly taken from ArtBench-10 dataset and the rest is generated equally using Latent Diffusion and Standard Diffusion models. The human-drawn art is in 256x256 resolution and images generated using Latent Diffusion and Standard Diffusion has 256x256 and 768x768 resolutions respectively.

Recently, AI-generated art improved significanlty that it has become increasingly harder now to differentiate between a real art and AI-generated art. This dataset can be used to train ML/DL models to identify AI-generated art. Further, since the dataset contains images from three different sources including, the dataset can also be used to train models to attribute art to its correct source verifying its authenticity.

The dataset was generated as a part of a research project in building a end-to-end system to detect and attribute AI-generated art images.

Reference

This dataset was originally created for the study "ArtBrain: An Explainable end-to-end Toolkit for Classification and Attribution of AI-Generated Art and Style" by Silva et al.

> Silva et al. (2024) 'ArtBrain: An Explainable end-to-end Toolkit for Classification and Attribution of AI-Generated Art and Style', arXiv. Available at: https://doi.org/10.48550/arXiv.2412.01512

The manuscript is under review at a journal and the reference will be updated here when the study is published.

Papers with Code

Have you published a study on this dataset? Add your results to the Papers with Code page! https://paperswithcode.com/paper/artbrain-an-explainable-end-to-end-toolkit