CNC: Cross-modal Normality Constraint for Unsupervised Multi-class Anomaly Detection

TL;DR

This paper introduces a novel cross-modal normality constraint method that uses class-agnostic prompts and a mixture-of-experts to improve unsupervised multi-class anomaly detection by mitigating decoder over-generalization.

Contribution

It proposes a new approach leveraging textual prompts and a mixture-of-experts to better distinguish normal from abnormal patterns in multi-class anomaly detection.

Findings

Achieves competitive results on MVTec AD dataset

Effectively reduces over-generalization of the decoder

Improves detection accuracy across multiple classes

Abstract

Existing unsupervised distillation-based methods rely on the differences between encoded and decoded features to locate abnormal regions in test images. However, the decoder trained only on normal samples still reconstructs abnormal patch features well, degrading performance. This issue is particularly pronounced in unsupervised multi-class anomaly detection tasks. We attribute this behavior to over-generalization(OG) of decoder: the significantly increasing diversity of patch patterns in multi-class training enhances the model generalization on normal patches, but also inadvertently broadens its generalization to abnormal patches. To mitigate OG, we propose a novel approach that leverages class-agnostic learnable prompts to capture common textual normality across various visual patterns, and then apply them to guide the decoded features towards a normal textual representation, suppressing over-generalization of the decoder on abnormal patterns. To further improve performance, we also introduce a gated mixture-of-experts module to specialize in handling diverse patch patterns and reduce mutual interference between them in multi-class training. Our method achieves competitive performance on the MVTec AD and VisA datasets, demonstrating its effectiveness.