Multimodal Industrial Anomaly Detection by Crossmodal Feature Mapping

TL;DR

This paper presents a fast, lightweight multimodal anomaly detection framework that maps features between point clouds and RGB images, achieving state-of-the-art results with improved efficiency on industrial datasets.

Contribution

A novel crossmodal feature mapping approach for industrial anomaly detection that enhances speed and memory efficiency while maintaining high detection accuracy.

Findings

Achieves state-of-the-art detection and segmentation performance.

Faster inference and lower memory usage than previous methods.

Layer-pruning improves efficiency with minimal performance loss.

Abstract

The paper explores the industrial multimodal Anomaly Detection (AD) task, which exploits point clouds and RGB images to localize anomalies. We introduce a novel light and fast framework that learns to map features from one modality to the other on nominal samples. At test time, anomalies are detected by pinpointing inconsistencies between observed and mapped features. Extensive experiments show that our approach achieves state-of-the-art detection and segmentation performance in both the standard and few-shot settings on the MVTec 3D-AD dataset while achieving faster inference and occupying less memory than previous multimodal AD methods. Moreover, we propose a layer-pruning technique to improve memory and time efficiency with a marginal sacrifice in performance.