Inconsistency-based Active Learning for LiDAR Object Detection

2025-05-01

AI Summary

High annotation costs hinder scalable development of LiDAR point cloud object detectors. Method: This paper proposes an active learning approach leveraging model prediction inconsistency—quantifying informativeness via inter-inference variation (e.g., bounding box count) over unlabeled samples, enabling lightweight, training-free sample selection. Contribution/Results: To our knowledge, this is the first systematic integration of inconsistency-based criteria into LiDAR detection active learning. Evaluated on PointPillars and CenterPoint, the method achieves comparable mAP to full random sampling using only 50% of labeled data, substantially reducing annotation effort. It demonstrates robustness across diverse model architectures and data distributions, offering a novel paradigm for cost-effective, iterative LiDAR perception model development.

Technology Category

• Machine Learning: Active Learning
• Search and Optimization: Sampling/Simulation-based Search
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Deep learning models for object detection in autonomous driving have recently achieved impressive performance gains and are already being deployed in vehicles worldwide. However, current models require increasingly large datasets for training. Acquiring and labeling such data is costly, necessitating the development of new strategies to optimize this process. Active learning is a promising approach that has been extensively researched in the image domain. In our work, we extend this concept to the LiDAR domain by developing several inconsistency-based sample selection strategies and evaluate their effectiveness in various settings. Our results show that using a naive inconsistency approach based on the number of detected boxes, we achieve the same mAP as the random sampling strategy with 50% of the labeled data.

Problem

• Optimizing LiDAR object detection with active learning
• Reducing labeling costs via inconsistency-based sample selection
• Improving model efficiency with fewer labeled samples

Innovation

• Inconsistency-based active learning for LiDAR
• Sample selection strategies for data efficiency
• Achieves same mAP with 50% labeled data