LiSu: A Dataset and Method for LiDAR Surface Normal Estimation

2025-03-11

AI Summary

Limited progress in LiDAR point cloud surface normal estimation stems from the scarcity of large-scale ground-truth annotations and robust algorithms. To address this, we introduce LiSu—the first large-scale synthetic LiDAR dataset with pixel-accurate, physically grounded normal ground truth—generated via a high-fidelity traffic simulation engine. We further propose a spatiotemporal joint regularization self-training framework that jointly enforces spatial consistency and temporal smoothness, significantly enhancing robustness against sparse, noisy point clouds and erroneous pseudo-labels, thereby facilitating effective synthetic-to-real domain adaptation. Our method is compatible with both graph neural networks and point-based backbone architectures. On LiSu, it achieves state-of-the-art performance; more importantly, it substantially improves neural surface reconstruction quality on real-world scenes. Extensive experiments validate both the utility of LiSu and the generalizability and effectiveness of our approach.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

While surface normals are widely used to analyse 3D scene geometry, surface normal estimation from LiDAR point clouds remains severely underexplored. This is caused by the lack of large-scale annotated datasets on the one hand, and lack of methods that can robustly handle the sparse and often noisy LiDAR data in a reasonable time on the other hand. We address these limitations using a traffic simulation engine and present LiSu, the first large-scale, synthetic LiDAR point cloud dataset with ground truth surface normal annotations, eliminating the need for tedious manual labeling. Additionally, we propose a novel method that exploits the spatiotemporal characteristics of autonomous driving data to enhance surface normal estimation accuracy. By incorporating two regularization terms, we enforce spatial consistency among neighboring points and temporal smoothness across consecutive LiDAR frames. These regularizers are particularly effective in self-training settings, where they mitigate the impact of noisy pseudo-labels, enabling robust real-world deployment. We demonstrate the effectiveness of our method on LiSu, achieving state-of-the-art performance in LiDAR surface normal estimation. Moreover, we showcase its full potential in addressing the challenging task of synthetic-to-real domain adaptation, leading to improved neural surface reconstruction on real-world data.

Problem

• Lack of large-scale annotated LiDAR datasets for surface normal estimation.
• Need for robust methods to handle sparse, noisy LiDAR data efficiently.
• Challenges in synthetic-to-real domain adaptation for neural surface reconstruction.

Innovation

• Large-scale synthetic LiDAR dataset with annotations
• Spatiotemporal regularization for enhanced accuracy
• Self-training mitigates noisy pseudo-labels impact