Zero-Shot Image Anomaly Detection Using Generative Foundation Models
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866912511138201600 |
|---|---|
| author | Abdi, Lemar Valiuddin, Amaan Caetano, Francisco Viviers, Christiaan van der Sommen, Fons |
| author_facet | Abdi, Lemar Valiuddin, Amaan Caetano, Francisco Viviers, Christiaan van der Sommen, Fons |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_22692 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Zero-Shot Image Anomaly Detection Using Generative Foundation Models Abdi, Lemar Valiuddin, Amaan Caetano, Francisco Viviers, Christiaan van der Sommen, Fons Computer Vision and Pattern Recognition 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. |
| title | Zero-Shot Image Anomaly Detection Using Generative Foundation Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2507.22692 |