Paper published in a journal (Scientific congresses, symposiums and conference proceedings)
VLMDiff: Leveraging Vision-Language Models for Multi-Class Anomaly Detection with Diffusion
HICSONMEZ, Samet; SHABAYEK, Abd El Rahman; AOUADA, Djamila
2026In 2026 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
Peer reviewed
 

Files


Full Text
_WACV_2026__VLMDIFF_paper_supp.pdf
Author postprint (27.09 MB)
Download

All documents in ORBilu are protected by a user license.

Send to



Details



Keywords :
Image Anomaly Detection, VLMs, Diffusion Models
Abstract :
[en] Detecting visual anomalies in diverse, multi-class real-world images is a significant challenge. We introduce \ours, a novel unsupervised multi-class visual anomaly detection framework. It integrates a Latent Diffusion Model (LDM) with a Vision-Language Model (VLM) for enhanced anomaly localization and detection. Specifically, a pre-trained VLM with a simple prompt extracts detailed image descriptions, serving as additional conditioning for LDM training. Current diffusion-based methods rely on synthetic noise generation, limiting their generalization and requiring per-class model training, which hinders scalability. \ours, however, leverages VLMs to obtain normal captions without manual annotations or additional training. These descriptions condition the diffusion model, learning a robust normal image feature representation for multi-class anomaly detection. Our method achieves competitive performance, improving the pixel-level Per-Region-Overlap (PRO) metric by up to 25 points on the Real-IAD dataset and 8 points on the COCO-AD dataset, outperforming state-of-the-art diffusion-based approaches. Code is available at https://github.com/giddyyupp/VLMDiff.
Disciplines :
Computer science
Author, co-author :
HICSONMEZ, Samet  ;  University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > CVI2
SHABAYEK, Abd El Rahman  ;  University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > CVI2
AOUADA, Djamila  ;  University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > CVI2
External co-authors :
no
Language :
English
Title :
VLMDiff: Leveraging Vision-Language Models for Multi-Class Anomaly Detection with Diffusion
Publication date :
2026
Event name :
IEEE/CVF Winter Conference on Applications of Computer Vision 2026
Event organizer :
IEEE/CVF
Event place :
Tucson, United States
Event date :
06 - 10 March 2026
Audience :
International
Journal title :
2026 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
Peer reviewed :
Peer reviewed
FnR Project :
DEFENCE22/17813724/AUREA
Available on ORBilu :
since 21 February 2026

Statistics


Number of views
91 (1 by Unilu)
Number of downloads
42 (1 by Unilu)

Bibliography


Similar publications



Contact ORBilu