Through the LiDAR's Eye: A Feature-Enriched and Uncertainty-Aware Annotation Pipeline for Terrestrial Point Cloud Semantic Segmentation

Accurate semantic segmentation of terrestrial laser scanning (TLS) point clouds is limited by costly manual annotation. We propose a semi-automated, uncertainty-aware pipeline that integrates spherical projection, feature enrichment, ensemble learning, and targeted annotation to reduce labeling effort while sustaining high accuracy. Our approach projects 3D points to a 2D spherical grid, enriches pixels with multi-source features, and trains an ensemble of segmentation networks to produce pseudo-labels and uncertainty maps, the latter guiding annotation of ambiguous regions. The 2D outputs are back-projected to 3D, yielding densely annotated point clouds supported by a three-tier visualization suite (2D feature maps, 3D colorized point clouds, and compact virtual spheres) for rapid triage and reviewer guidance. Using this pipeline, we built Mangrove3D, a semantic segmentation TLS dataset for mangrove forests. We further evaluated data efficiency and feature importance to address two key questions: (1) how much annotated data is needed, and (2) which features matter most. Results show that performance saturates after $\sim$12 annotated scans, geometric features contribute the most, and compact nine-channel stacks capture nearly all discriminative power (mean Intersection over Union, mIoU plateau around 0.76). Finally, we confirm generalization of our feature-enrichment strategy through cross-dataset tests on ForestSemantic and Semantic3D. Our contributions include: (i) a robust, uncertainty-aware TLS annotation pipeline with visualization tools; (ii) the Mangrove3D dataset; and (iii) empirical guidance on data efficiency and feature importance, enabling scalable, high-quality segmentation of TLS point clouds for ecological monitoring and beyond.

Three-stage workflow for annotating terrestrial-LiDAR scans. Stage 1: Spherical projection converts raw TLS points into two-dimensional feature maps and pseudo-RGB images. Stage 2: An iterative loop combines active learning and self-training: an emsemble DNN model is repeatedly refined using uncertainty-guided queries and high-confidence pseudo-labels. Stage 3: The resulting 2-D segmentation masks are back-projected, followed by label refinement in 3D space and then reproject back to 2D, to yield a fully annotated point cloud and refined 2D segmentation mask.