Industrial3D: A Terrestrial LiDAR Point Cloud Dataset and CrossParadigm Benchmark for Industrial Infrastructure

2026-03-30

AI Summary

This work addresses the significant challenges in semantic understanding of industrial MEP (mechanical, electrical, and plumbing) point clouds—namely severe geometric ambiguity, heavy occlusion, and extreme class imbalance—which are inadequately represented in existing building-scale datasets. To bridge this gap, we introduce the first large-scale, high-resolution terrestrial laser scanning point cloud dataset tailored to industrial infrastructure, encompassing 13 water treatment plants with 612 million expert-annotated points at a resolution of 6 mm. We further establish a unified cross-paradigm benchmark covering fully supervised, weakly supervised, unsupervised, and foundation model approaches. Under a consistent evaluation protocol, nine representative methods are assessed: the best fully supervised model achieves an mIoU of 55.74%, whereas zero-shot Point-SAM attains only 15.79%, revealing a nearly 40-percentage-point performance gap and highlighting the substantial limitations of current methods in industrial settings.

Abstract

Automated semantic understanding of dense point clouds is a prerequisite for Scan-to-BIM pipelines, digital twin construction, and as-built verification--core tasks in the digital transformation of the construction industry. Yet for industrial mechanical, electrical, and plumbing (MEP) facilities, this challenge remains largely unsolved: TLS acquisitions of water treatment plants, chiller halls, and pumping stations exhibit extreme geometric ambiguity, severe occlusion, and extreme class imbalance that architectural benchmarks (e.g., S3DIS or ScanNet) cannot adequately represent. We present Industrial3D, a terrestrial LiDAR dataset comprising 612 million expertly labelled points at 6 mm resolution from 13 water treatment facilities. At 6.6x the scale of the closest comparable MEP dataset, Industrial3D provides the largest and most demanding testbed for industrial 3D scene understanding to date. We further establish the first industrial cross-paradigm benchmark, evaluating nine representative methods across fully supervised, weakly supervised, unsupervised, and foundation model settings under a unified benchmark protocol. The best supervised method achieves 55.74% mIoU, whereas zero-shot Point-SAM reaches only 15.79%--a 39.95 percentage-point gap that quantifies the unresolved domain-transfer challenge for industrial TLS data. Systematic analysis reveals that this gap originates from a dual crisis: statistical rarity (215:1 imbalance, 3.5x more severe than S3DIS) and geometric ambiguity (tail-class points share cylindrical primitives with head-class pipes) that frequency-based re-weighting alone cannot resolve. Industrial3D, along with benchmark code and pre-trained models, will be publicly available at https://github.com/pointcloudyc/Industrial3D.

Problem

• industrial point clouds
• semantic understanding
• class imbalance
• geometric ambiguity
• terrestrial LiDAR

Innovation

• Industrial3D
• terrestrial LiDAR
• cross-paradigm benchmark
• class imbalance
• geometric ambiguity