Class-Distribution Guided Active Learning for 3D Occupancy Prediction in Autonomous Driving

TL;DR

This paper introduces a class-distribution guided active learning method for 3D occupancy prediction in autonomous driving, effectively addressing class imbalance and reducing annotation costs while maintaining high accuracy.

Contribution

The proposed framework combines diversity and class-frequency weighting criteria for sample selection, improving annotation efficiency in 3D occupancy tasks.

Findings

Achieves comparable performance to full supervision with only 42.4% labeled data.

Outperforms existing active learning baselines at the same annotation budget.

Demonstrates consistent effectiveness across different datasets and architectures.

Abstract

3D occupancy prediction provides dense spatial understanding critical for safe autonomous driving. However, this task suffers from a severe class imbalance due to its volumetric representation, where safety-critical objects (bicycles, traffic cones, pedestrians) occupy minimal voxels compared to dominant backgrounds. Additionally, voxel-level annotation is costly, yet dedicating effort to dominant classes is inefficient. To address these challenges, we propose a class-distribution guided active learning framework for selecting training samples to annotate in autonomous driving datasets. Our approach combines three complementary criteria to select the training samples. Inter-sample diversity prioritizes samples whose predicted class distributions differ from those of the labeled set, intra-set diversity prevents redundant sampling within each acquisition cycle, and frequency-weighted uncertainty emphasizes rare classes by reweighting voxel-level entropy with inverse per-sample class proportions. We ensure evaluation validity by using a geographically disjoint train/validation split of Occ3D-nuScenes, which reduces train-validation overlap and mitigates potential map memorization. With only 42.4% labeled data, our framework reaches 26.62 mIoU, comparable to full supervision and outperforming active learning baselines at the same budget. We further validate generality on SemanticKITTI using a different architecture, demonstrating consistent effectiveness across datasets.