SuP-SLiP: Subsampled Processing of Large-scale Static LIDAR Point Clouds

Vijay Jaisankar, Jaya Sreevalsan-Nair

Abstract: Annotation is a crucial component of point cloud analysis. However, due to the sheer number of points in large-scale static point clouds, it is an expensive and time-consuming process. We address this issue using a novel lightweight approach to reduce the overall annotation time for unlabelled static point clouds in the binary road segmentation task. By leveraging models trained on other point clouds in the same distribution, and radius sampling, our approach determines a small fraction of points for annotation. It automatically labels the remaining points using nearest-neighbor aggregation. We implement this approach in an end-to-end system for mobile laser scanning (MLS) or mobile LiDAR point clouds, SuP-SLiP, i.e., Subsampled Processing of Large-scale Static LIDAR Point Clouds. We validate the robustness of this method through the bit-flipping adversarial attack and account for varying budgets by providing a feature that suggests a custom number of points to annotate for a given point cloud.

External IDs:doi:10.1145/3681769.3698585