ArtiFact: Real and Fake Image Dataset

Home » Dataset Download » ArtiFact: Real and Fake Image Dataset

ArtiFact: Real and Fake Image Dataset

Datasets

File

ArtiFact: Real and Fake Image Dataset

Use Case

ArtiFact: Real and Fake Image Dataset

Description

The ArtiFact dataset is a comprehensive collection of 2.5 million real and synthetic images across diverse categories. Specifically designed for evaluating synthetic image detectors, it includes images generated using 25 methods, such as GANs and diffusion models, thus providing a robust benchmark for real-world image detection tasks.

Description:

The ArtiFact dataset is a comprehensive, large-scale image collection that includes a diverse array of both real and synthetic images from multiple categories. These categories encompass Human/Human Faces, Animal/Animal Faces, Places, Vehicles, Art, and a variety of other real-life objects. Importantly, the dataset comprises images from 8 carefully selected sources to ensure diversity. Moreover, it features images synthesized using 25 distinct methods, including 13 GANs, 7 Diffusion models, and 5 other miscellaneous generators. In total, the dataset contains 2,496,738 images, consisting of 964,989 real images and 1,531,749 fake images.

Download Dataset

To maintain diversity across different sources, the dataset randomly samples real images from source datasets with numerous categories. Meanwhile, it generates synthetic images to match these categories. Furthermore, it uses captions and image masks from the COCO dataset to generate images via text-to-image and inpainting generators. Conversely, noise-to-image generators use normally distributed noise with various random seeds. To enhance realism, the dataset further processes the images to reflect real-world scenarios by applying random cropping, downscaling, and JPEG compression, all in accordance with IEEE VIP Cup 2022 standards.

The primary goal of the ArtiFact dataset is to serve as a benchmark for evaluating the performance of synthetic image detectors under real-world conditions. With its broad spectrum of diversity in terms of the generators used and levels of syntheticity, it provides a challenging environment for image detection tasks. Consequently, it pushes the boundaries of current detection capabilities.