Exploring Single Domain Generalization of LiDAR-Based Semantic Segmentation under Imperfect Labels
Weitong Kong, Zichao Zeng, Di Wen, Jiale Wei, Kunyu Peng, June Moh Goo, Jan Boehm, Rainer Stiefelhagen
AI summary
Problem
Existing LiDAR domain generalization methods assume clean annotations, but real-world labels are often noisy, and direct transfer of 2D noisy-label techniques to 3D point clouds fails due to sparsity and irregularity.
Approach
The authors establish a controlled benchmark and adapt three image-based noisy-label strategies to 3D data, then introduce DuNe, a dual-view framework that aligns strong and weak geometric views via feature consistency and confidence-aware filtering.
Key results
- Established the first DGLSS-NL benchmark for single-source LiDAR segmentation under symmetric label noise
- Adapted and diagnosed TCL, DISC, and NPN noisy-label strategies for large-scale 3D point clouds
- Proposed DuNe, a dual-view framework enforcing feature consistency and confidence-filtered supervision
- Achieved state-of-the-art mIoU scores of 56.86%, 42.28%, and 52.58% on SemanticKITTI, nuScenes, and SemanticPOSS under 10% noise
Why it matters
Enables reliable autonomous driving perception across diverse environments without costly re-annotation or clean training labels.
Abstract
Accurate perception is critical for vehicle safety, with LiDAR as a key enabler in autonomous driving. To ensure robust performance across environments, sensor types, and weather conditions without costly re-annotation, domain generalization in LiDAR-based 3D semantic segmentation is essential. However, LiDAR annotations are often noisy due to sensor imperfections, occlusions, and human errors. Such noise degrades segmentation accuracy and is further amplified under domain shifts, threatening system reliability. While noisy- label learning is well-studied in images, its extension to 3D LiDAR segmentation under domain generalization remains largely unexplored, as the sparse and irregular structure of point clouds limits direct use of 2D methods. To address this gap, we introduce the novel task Domain Generalization for LiDAR Semantic Segmentation under Noisy Labels (DGLSS- NL) and establish the first benchmark by adapting three representative noisy-label learning strategies from image clas- sification to 3D segmentation. However, we find that existing noisy-label learning approaches adapt poorly to LiDAR data. We therefore propose DuNe, a dual-view framework with strong and weak branches that enforce feature-level consistency and apply cross-entropy loss based on confidence-aware filtering of predictions. Our approach shows state-of-the-art performance by achieving 56.86% mIoU on SemanticKITTI, 42.28% on nuScenes, and 52.58% on SemanticPOSS under 10% symmetric label noise, with an overall Arithmetic Mean (AM) of 49.57% and Harmonic Mean (HM) of 48.50%, thereby demonstrating robust domain generalization in DGLSS-NL tasks. The code is available at https://github.com/MKong17/DGLSS-NL.git.