How Sampling Strategy Affects Imbalance Mitigation in LiDAR Segmentation: A Study of Structured vs. Random Point-Based Architectures

2026-08-17

AI Summary

This study addresses the unclear efficacy of class imbalance mitigation strategies in LiDAR point cloud semantic segmentation by systematically evaluating reweighting schemes and loss functions across diverse sampling architectures. Through multi-dataset benchmarking and loss landscape analysis, this work reveals for the first time the distinct behaviors of mitigation mechanisms under structured versus random sampling, alongside the coupling between loss landscapes and data geometry. Notably, inverse frequency weighting exhibits performance degradation up to 12%, whereas uniform weighting achieves results comparable to complex losses within structured sampling frameworks. By elucidating the intricate interactions among sampling strategies, imbalance severity, and intrinsic data characteristics, this research provides targeted optimization guidelines for advancing 3D scene segmentation tasks.

Abstract

Class imbalance in LiDAR point clouds poses challenges for semantic segmentation in autonomous navigation and urban mapping. While 2D vision has numerous mitigation techniques, their effectiveness in 3D remains unclear. We benchmark six reweighting schemes and five imbalance-aware losses across three datasets (DALES, S3DIS, STPLS3D) using two architectures (KPConv, RandLA-Net). Inverse-frequency weighting degrades performance by up to 12% compared to uniform weighting, with catastrophic failures in minority classes. Uniform weighting performs within 2% of complex losses for structured sampling (KPConv) but benefits less for random sampling (RandLA-Net, up to 4.6% gap). Loss landscape analysis reveals a complex interplay: for structured sampling, imbalance ratio determines landscape geometry on real LiDAR data but decouples from it on synthetic data; for random sampling, landscapes show high sensitivity to dataset geometry regardless of imbalance ratio. For the two evaluated point-based architectures, these results suggest that the interaction between sampling strategy (structured vs. random), imbalance severity, and data acquisition characteristics shapes which mitigation approaches are effective.

Problem

• LiDAR segmentation
• class imbalance
• sampling strategy
• point-based architectures
• imbalance mitigation

Innovation

• Sampling Strategy
• Class Imbalance
• LiDAR Segmentation
• Loss Landscape Analysis
• Reweighting Schemes