Multi-View Reconstruction with Global Context for 3D Anomaly Detection

TL;DR

This paper introduces Multi-View Reconstruction (MVR), a novel approach for 3D anomaly detection that enhances global information learning by converting high-resolution point clouds into multi-view images, significantly improving detection performance.

Contribution

The paper proposes MVR, a lossless multi-view conversion method combined with a reconstruction framework, addressing the global information deficiency in existing 3D anomaly detection techniques.

Findings

Achieves 89.6% object-wise AU-ROC on Real3D-AD

Achieves 95.7% point-wise AU-ROC on Real3D-AD

Demonstrates superior performance over existing methods

Abstract

3D anomaly detection is critical in industrial quality inspection. While existing methods achieve notable progress, their performance degrades in high-precision 3D anomaly detection due to insufficient global information. To address this, we propose Multi-View Reconstruction (MVR), a method that losslessly converts high-resolution point clouds into multi-view images and employs a reconstruction-based anomaly detection framework to enhance global information learning. Extensive experiments demonstrate the effectiveness of MVR, achieving 89.6\% object-wise AU-ROC and 95.7\% point-wise AU-ROC on the Real3D-AD benchmark.