Towards High-Resolution 3D Anomaly Detection: A Scalable Dataset and Real-Time Framework for Subtle Industrial Defects

2025-07-10

AI Summary

Industrial 3D anomaly detection in point clouds is hindered by low-resolution baselines, scarcity of high-resolution training data, and the absence of real-time frameworks. To address these challenges, this work introduces MiniShift—the first high-resolution point cloud anomaly detection dataset—and proposes Simple3D, a lightweight, efficient detection framework. Its core contributions are: (1) a scalable 3D anomaly generation pipeline yielding a large-scale, high-resolution dataset rich in fine-grained defects; and (2) two novel components—Multi-Scale Neighborhood Descriptors (MSND) and Local Feature Space Aggregation (LFSA)—enabling precise geometric detail modeling and rapid inference. Evaluated on MiniShift and established benchmarks, Simple3D achieves >20 FPS real-time performance while significantly outperforming state-of-the-art methods in detection accuracy. This demonstrates the critical importance of high-resolution input representation and efficient feature aggregation for industrial defect detection.

Technology Category

• Computer Vision: 3D Computer Vision
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Hardware-aware ML

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

In industrial point cloud analysis, detecting subtle anomalies demands high-resolution spatial data, yet prevailing benchmarks emphasize low-resolution inputs. To address this disparity, we propose a scalable pipeline for generating realistic and subtle 3D anomalies. Employing this pipeline, we developed MiniShift, the inaugural high-resolution 3D anomaly detection dataset, encompassing 2,577 point clouds, each with 500,000 points and anomalies occupying less than 1% of the total. We further introduce Simple3D, an efficient framework integrating Multi-scale Neighborhood Descriptors (MSND) and Local Feature Spatial Aggregation (LFSA) to capture intricate geometric details with minimal computational overhead, achieving real-time inference exceeding 20 fps. Extensive evaluations on MiniShift and established benchmarks demonstrate that Simple3D surpasses state-of-the-art methods in both accuracy and speed, highlighting the pivotal role of high-resolution data and effective feature aggregation in advancing practical 3D anomaly detection.

Problem

• Detecting subtle anomalies in high-resolution 3D industrial point clouds
• Lack of scalable datasets for realistic 3D anomaly generation
• Need for efficient real-time frameworks for 3D anomaly detection

Innovation

• Scalable pipeline for realistic 3D anomalies
• High-resolution dataset MiniShift with 500K points
• Simple3D framework with MSND and LFSA