{"ID":2888812,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2507.22692","arxiv_id":"2507.22692","title":"Zero-Shot Image Anomaly Detection Using Generative Foundation Models","abstract":"Detecting out-of-distribution (OOD) inputs is pivotal for deploying safe vision systems in open-world environments. We revisit diffusion models, not as generators, but as universal perceptual templates for OOD detection. This research explores the use of score-based generative models as foundational tools for semantic anomaly detection across unseen datasets. Specifically, we leverage the denoising trajectories of Denoising Diffusion Models (DDMs) as a rich source of texture and semantic information. By analyzing Stein score errors, amplified through the Structural Similarity Index Metric (SSIM), we introduce a novel method for identifying anomalous samples without requiring re-training on each target dataset. Our approach improves over state-of-the-art and relies on training a single model on one dataset -- CelebA -- which we find to be an effective base distribution, even outperforming more commonly used datasets like ImageNet in several settings. Experimental results show near-perfect performance on some benchmarks, with notable headroom on others, highlighting both the strength and future potential of generative foundation models in anomaly detection.","short_abstract":"Detecting out-of-distribution (OOD) inputs is pivotal for deploying safe vision systems in open-world environments. We revisit diffusion models, not as generators, but as universal perceptual templates for OOD detection. This research explores the use of score-based generative models as foundational tools for semantic...","url_abs":"https://arxiv.org/abs/2507.22692","url_pdf":"https://arxiv.org/pdf/2507.22692v1","authors":"[\"Lemar Abdi\",\"Amaan Valiuddin\",\"Francisco Caetano\",\"Christiaan Viviers\",\"Fons van der Sommen\"]","published":"2025-07-30T13:56:36Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Diffusion Model\"]","has_code":false}