Label-Efficient Point Cloud Segmentation with Active Learning

2025-12-05

AI Summary

To address the high annotation cost of 3D point cloud semantic segmentation, this paper proposes an efficient active learning strategy: point clouds are projected onto a 2D grid for spatial partitioning, and prediction uncertainty is estimated via deep neural network ensembles; the most informative grid regions are then prioritized for annotation. Crucially, “annotation area ratio” is introduced as a novel evaluation metric—replacing conventional “number of annotated points”—to better reflect point cloud sparsity and geometric structure. Experiments on three real-world datasets—S3DIS, Toronto-3D, and Freiburg City—demonstrate that our method achieves performance comparable to or exceeding state-of-the-art active learning approaches while reducing annotation budgets by 30–50% on average. This validates its effectiveness and practicality in substantially lowering human annotation effort.

Technology Category

• Machine Learning: Active Learning
• Computer Vision: Segmentation
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Semantics and Knowledge: Representation, semantic annotation, enhancement, enrichments, access and/or integration of a variety of data on the Web
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Security and Privacy: Large-scale security measurements

Abstract

Semantic segmentation of 3D point cloud data often comes with high annotation costs. Active learning automates the process of selecting which data to annotate, reducing the total amount of annotation needed to achieve satisfactory performance. Recent approaches to active learning for 3D point clouds are often based on sophisticated heuristics for both, splitting point clouds into annotatable regions and selecting the most beneficial for further neural network training. In this work, we propose a novel and easy-to-implement strategy to separate the point cloud into annotatable regions. In our approach, we utilize a 2D grid to subdivide the point cloud into columns. To identify the next data to be annotated, we employ a network ensemble to estimate the uncertainty in the network output. We evaluate our method on the S3DIS dataset, the Toronto-3D dataset, and a large-scale urban 3D point cloud of the city of Freiburg, which we labeled in parts manually. The extensive evaluation shows that our method yields performance on par with, or even better than, complex state-of-the-art methods on all datasets. Furthermore, we provide results suggesting that in the context of point clouds the annotated area can be a more meaningful measure for active learning algorithms than the number of annotated points.

Problem

• Reduces annotation costs for 3D point cloud segmentation.
• Proposes a simple grid-based method to split point clouds.
• Uses network ensemble uncertainty to select data for annotation.

Innovation

• Using a 2D grid to subdivide point clouds into columns
• Employing a network ensemble to estimate uncertainty for annotation selection
• Proposing annotated area as a measure for active learning algorithms