Exploring Large Vision-Language Models for Robust and Efficient Industrial Anomaly Detection

TL;DR

The paper addresses industrial anomaly detection (IAD) by leveraging large vision-language models to jointly reason about visuals and textual descriptions. It introduces CLAD, which learns a shared embedding space for visual and textual representations via a contrastive cross-modal loss, aligning $z_{\text{v}}$ and $z_{\text{t}}$ for normality while pushing anomalies apart, and adds a contextualized reasoning module for explanations. The approach achieves state-of-the-art results on the MVTec-AD and VisA benchmarks for both image-level anomaly detection and pixel-level localization, with comprehensive ablations and a human-evaluation study validating key components and interpretability. This work advances practical IAD by improving generalization across domains and providing interpretable localization, which is crucial for industrial deployment and maintenance planning.

Abstract

Industrial anomaly detection (IAD) plays a crucial role in the maintenance and quality control of manufacturing processes. In this paper, we propose a novel approach, Vision-Language Anomaly Detection via Contrastive Cross-Modal Training (CLAD), which leverages large vision-language models (LVLMs) to improve both anomaly detection and localization in industrial settings. CLAD aligns visual and textual features into a shared embedding space using contrastive learning, ensuring that normal instances are grouped together while anomalies are pushed apart. Through extensive experiments on two benchmark industrial datasets, MVTec-AD and VisA, we demonstrate that CLAD outperforms state-of-the-art methods in both image-level anomaly detection and pixel-level anomaly localization. Additionally, we provide ablation studies and human evaluation to validate the importance of key components in our method. Our approach not only achieves superior performance but also enhances interpretability by accurately localizing anomalies, making it a promising solution for real-world industrial applications.