Two Steps Are All You Need: Efficient 3D Point Cloud Anomaly Detection with Consistency Models

TL;DR

This paper introduces a novel anomaly detection method for 3D point clouds that significantly reduces inference time by reformulating diffusion models with consistency learning, enabling fast, reliable detection on resource-limited systems.

Contribution

The authors propose a consistency learning approach for 3D anomaly detection that achieves up to 80x faster inference without GPU acceleration, maintaining high detection accuracy.

Findings

Achieves up to 80x faster runtime than state-of-the-art methods.

Outperforms R3D-AD with 76.20% I-AUROC on Anomaly-ShapeNet.

Remains effective on resource-constrained edge devices.

Abstract

Diffusion models are rapidly redefining 3D anomaly detection in point cloud data. As 3D sensing becomes integral to modern manufacturing, reliable anomaly detection is essential for high-throughput quality assurance and process control. Yet practical deployment on resource-constrained, latency-critical systems remains limited. Existing methods are often computationally prohibitive or unreliable in complex, unmasked regions, and diffusion pipelines are inherently bottlenecked by iterative denoising. In this work, we address this bottleneck by reformulating reconstructionbased anomaly detection through consistency learning, enabling direct prediction of anomaly-free geometry in one or two network evaluations. We further introduce a novel hybrid loss formulation that explicitly enforces reconstruction toward clean data. This design substantially reduces inference cost, achieving up to 80x faster runtime than the current state-of-the-art method, without GPU acceleration, while preserving strong detection performance. It outperforms R3D-AD on Anomaly-ShapeNet with 76.20% I-AUROC and remains competitive on Real3DAD with 72.80% I-AUROC, enabling efficient, low-latency anomaly detection on resource-constrained platforms, including drones, smart industrial cameras, and other edge devices.