Industrial 3D anomaly detection improves with logic and geometry checks
Beyond Geometry: Benchmarking and Consistency Reasoning for 3D Logical Anomaly Detection
Computer Vision and Pattern Recognition
Summary
Many industrial defects in 3D objects are caused not just by shape problems but by errors in design or how parts fit together. The authors created a new dataset called ILGAD that shows these logical errors in 3D scans and helps train and test detection methods. They also developed a new framework that looks at how object parts relate and fit, spotting mistakes that simple shape checks miss. Their tests show this approach finds both logical and shape problems better than before and works well on different datasets.
What this means in practice
- •For industrial quality control teams: Detect design and assembly rule mistakes in manufactured parts using 3D scans with logic-based anomaly detection.
- •For robotics engineers: Improve robot quality inspection by integrating 3D logical anomaly detection to identify errors beyond simple shape defects.
Authors
Zhiqiang Qin, He Xie, Junfei Yi, Yang Yang, Hao Wang, Yunkang Cao, Hui Zhang, Yaonan Wang
Abstract
Existing 3D industrial anomaly detection mainly targets local geometric deviations. In contrast, many industrial anomalies violate object-level design or assembly rules, which we define as 3D logical anomalies. To address these challenges, we introduce the Industrial Logical Anomaly Detection Dataset (ILGAD), the first scalable benchmark dedicated to logical anomalies in industrial point clouds. ILGAD contains 2,774 samples from 15 categories with point-level annotations and covers existence, specification, pose, and assembly-state errors. To detect such 3D logical anomalies, we propose a consistency reasoning framework that assesses whether local geometry, structure coverage, and spatial relations conform to the normal design. The framework detects geometric changes, unsupported expected structures, and abnormal local arrangements. Experiments on ILGAD, Anomaly-ShapeNet, and IEC3D demonstrate superior object-level detection and point-level localization, showing that the framework effectively detects logical anomalies and generalizes to conventional geometric defects.