Multi-Scale Distillation for RGB-D Anomaly Detection on the PD-REAL Dataset

TL;DR

This paper introduces PD-REAL, a large-scale RGB-D dataset for unsupervised anomaly detection in 3D objects, and proposes a multi-scale distillation framework that improves detection accuracy by leveraging hierarchical multimodal features.

Contribution

The paper provides a new scalable 3D anomaly detection dataset and develops a multi-scale teacher-student framework that effectively integrates global and local features for improved detection.

Findings

Our method outperforms state-of-the-art algorithms on PD-REAL.

Multi-scale hierarchical distillation enhances anomaly detection accuracy.

The dataset enables cost-effective and controlled 3D anomaly detection research.

Abstract

We present PD-REAL, a novel large-scale dataset for unsupervised anomaly detection (AD) in the 3D domain. It is motivated by the fact that 2D-only representations in the AD task may fail to capture the geometric structures of anomalies due to uncertainty in lighting conditions or shooting angles. PD-REAL consists entirely of Play-Doh models for 15 object categories and focuses on the analysis of potential benefits from 3D information in a controlled environment. Specifically, objects are first created with six types of anomalies, such as \textit{dent}, \textit{crack}, or \textit{perforation}, and then photographed under different lighting conditions to mimic real-world inspection scenarios. To demonstrate the usefulness of 3D information, we use a commercially available RealSense camera to capture RGB and depth images. Compared to the existing 3D dataset for AD tasks, the data acquisition of PD-REAL is significantly cheaper, easily scalable, and easier to control variables. Furthermore, we introduce a multi-scale teacher--student framework with hierarchical distillation for multimodal anomaly detection. This architecture overcomes the inherent limitation of single-scale distillation approaches, which often struggle to reconcile global context with local features. Leveraging multi-level guidance from the teacher network, the student network can effectively capture richer features for anomaly detection. Extensive evaluations with our method and state-of-the-art AD algorithms on our dataset qualitatively and quantitatively demonstrate the higher detection accuracy of our method. Our dataset can be downloaded from https://github.com/Andy-cs008/PD-REAL