LaRE^2: Latent Reconstruction Error Based Method for Diffusion-Generated Image Detection

arXiv:2403.17465v1 Announce Type: new
Abstract: The evolution of Diffusion Models has dramatically improved image generation quality, making it increasingly difficult to differentiate between real and generated images. This development, while impressive, also raises significant privacy and security concerns. In response to this, we propose a novel Latent REconstruction error guided feature REfinement method (LaRE^2) for detecting the diffusion-generated images. We come up with the Latent Reconstruction Error (LaRE), the first reconstruction-error based feature in the latent space for generated image …

abstract arxiv concerns cs.ai cs.cv detection development diffusion diffusion models error evolution generated image image detection image generation images making novel privacy privacy and security quality raises security security concerns type