Generating Roadside LiDAR Datasets from Vehicle-Side Datasets via Novel View Synthesis

2026-05-07

AI Summary

Existing roadside LiDAR datasets are limited in scale, hindering the training of high-performance perception models. To address this, this work proposes the VRS framework, which for the first time enables cross-domain synthesis from vehicle-mounted to roadside LiDAR data. The method recovers occluded vehicle structures through point cloud completion and generates geometrically consistent, annotated roadside point clouds by incorporating occupancy-based visibility constraints. This approach supports flexible multi-view rendering, effectively bridging the domain gap between vehicle-mounted and roadside perspectives. Experiments demonstrate that the synthesized data significantly improves 3D object detection performance in roadside settings and enhances model generalization to unseen viewpoints.

Abstract

Intelligent Transportation Systems (ITS) require reliable environmental perception to support safe and efficient transportation. With the rapid development of Vehicle-to-everything (V2X), roadside perception has become an effective means to extend sensing coverage and improve traffic safety. However, the scarcity of large-scale annotated roadside LiDAR datasets poses a major challenge for training high-performance roadside perception models. In this paper, we introduce Vehicle-to-Roadside LiDAR Synthesis (VRS), a data synthesis framework that generates labeled roadside LiDAR datasets from vehicle-side datasets via LiDAR novel view synthesis. To mitigate the vehicle-to-roadside domain gap, VRS employs vehicle point cloud completion to compensate for missing geometry in vehicle-side observations, and introduces an occupancy-based visibility constraint to handle large viewpoint changes during cross-view rendering. The proposed framework enables flexible multi-view rendering for scalable roadside data generation. Extensive experiments on roadside 3D object detection demonstrate that the synthesized data effectively complements real roadside data, mitigates the limitations of limited real-world roadside data, and improves generalization to unseen roadside viewpoints.

Problem

• roadside LiDAR
• dataset scarcity
• environmental perception
• V2X
• 3D object detection

Innovation

• novel view synthesis
• LiDAR data synthesis
• vehicle-to-roadside
• point cloud completion
• occupancy-based visibility