A Spatiotemporal Correspondence Approach to Unsupervised LiDAR Segmentation with Traffic Applications
This addresses the problem of reducing annotation costs for LiDAR segmentation in autonomous vehicles and traffic infrastructure, though it is incremental as it builds on existing unsupervised approaches.
The paper tackles unsupervised semantic segmentation of outdoor LiDAR point clouds in traffic scenarios by leveraging spatiotemporal correspondences across frames and alternating between clustering and pseudo-label learning, achieving competitive performance against supervised methods on Semantic-KITTI, SemanticPOSS, and FLORIDA benchmarks.
We address the problem of unsupervised semantic segmentation of outdoor LiDAR point clouds in diverse traffic scenarios. The key idea is to leverage the spatiotemporal nature of a dynamic point cloud sequence and introduce drastically stronger augmentation by establishing spatiotemporal correspondences across multiple frames. We dovetail clustering and pseudo-label learning in this work. Essentially, we alternate between clustering points into semantic groups and optimizing models using point-wise pseudo-spatiotemporal labels with a simple learning objective. Therefore, our method can learn discriminative features in an unsupervised learning fashion. We show promising segmentation performance on Semantic-KITTI, SemanticPOSS, and FLORIDA benchmark datasets covering scenarios in autonomous vehicle and intersection infrastructure, which is competitive when compared against many existing fully supervised learning methods. This general framework can lead to a unified representation learning approach for LiDAR point clouds incorporating domain knowledge.