Detecting Images Generated by Diffusers. (arXiv:2303.05275v3 [cs.CV] UPDATED)

This paper explores the task of detecting images generated by text-to-image
diffusion models. To evaluate this, we consider images generated from captions
in the MSCOCO and Wikimedia datasets using two state-of-the-art models: Stable
Diffusion and GLIDE. Our experiments show that it is possible to detect the
generated images using simple Multi-Layer Perceptrons (MLPs), starting from
features extracted by CLIP, or traditional Convolutional Neural Networks
(CNNs). We also observe that models trained on images generated by Stable
Diffusion can detect images …

art arxiv clip cnns convolutional neural networks datasets diffusers diffusion diffusion models features generated image image diffusion images networks neural networks observe paper show stable diffusion state text text-to-image true