Multimodal Industrial Anomaly Detection via Geometric Prior

2026-03-23

AI Summary

This work addresses the limited effectiveness of existing methods in leveraging geometric cues—such as surface normals and 3D shape topology—for multimodal industrial anomaly detection, which often results in insufficient sensitivity to subtle deformations and irregular contours. To overcome this, we propose GPAD (Geometry-Prior-guided Anomaly Detection), a novel framework that employs a point cloud expert model to extract fine-grained geometric features and incorporates differential normals to enhance geometric detail. GPAD further introduces a two-stage multimodal fusion strategy coupled with a geometry-prior-based attention mechanism to enable effective collaboration between multimodal data and 3D geometric information, along with a geometry-aware anomaly segmentation module. Experiments demonstrate that GPAD significantly outperforms state-of-the-art methods on both the MVTec-3D AD and Eyecandies benchmarks.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data

Abstract

The purpose of multimodal industrial anomaly detection is to detect complex geometric shape defects such as subtle surface deformations and irregular contours that are difficult to detect in 2D-based methods. However, current multimodal industrial anomaly detection lacks the effective use of crucial geometric information like surface normal vectors and 3D shape topology, resulting in low detection accuracy. In this paper, we propose a novel Geometric Prior-based Anomaly Detection network (GPAD). Firstly, we propose a point cloud expert model to perform fine-grained geometric feature extraction, employing differential normal vector computation to enhance the geometric details of the extracted features and generate geometric prior. Secondly, we propose a two-stage fusion strategy to efficiently leverage the complementarity of multimodal data as well as the geometric prior inherent in 3D points. We further propose attention fusion and anomaly regions segmentation based on geometric prior, which enhance the model's ability to perceive geometric defects. Extensive experiments show that our multimodal industrial anomaly detection model outperforms the State-of-the-art (SOTA) methods in detection accuracy on both MVTec-3D AD and Eyecandies datasets.

Problem

• multimodal industrial anomaly detection
• geometric information
• surface normal vectors
• 3D shape topology
• complex geometric defects

Innovation

• Geometric Prior
• Multimodal Fusion
• Point Cloud
• Anomaly Detection
• Surface Normal