SPOT: Scalable 3D Pre-training via Occupancy Prediction for Learning Transferable 3D Representations

TL;DR

SPOT introduces a scalable pre-training method using occupancy prediction for 3D LiDAR data, enhancing transferability, robustness, and scalability across diverse perception tasks and datasets.

Contribution

The paper presents the first theoretical and empirical demonstration that occupancy prediction enables general 3D representation learning for LiDAR data.

Findings

Occupancy prediction effectively learns transferable 3D representations.

Beam re-sampling and class-balancing improve domain robustness.

More pre-training data consistently enhances downstream performance.

Abstract

Annotating 3D LiDAR point clouds for perception tasks is fundamental for many applications e.g., autonomous driving, yet it still remains notoriously labor-intensive. Pretraining-finetuning approach can alleviate the labeling burden by fine-tuning a pre-trained backbone across various downstream datasets as well as tasks. In this paper, we propose SPOT, namely Scalable Pre-training via Occupancy prediction for learning Transferable 3D representations under such a label-efficient fine-tuning paradigm. SPOT achieves effectiveness on various public datasets with different downstream tasks, showcasing its general representation power, cross-domain robustness and data scalability which are three key factors for real-world application. Specifically, we both theoretically and empirically show, for the first time, that general representations learning can be achieved through the task of occupancy prediction. Then, to address the domain gap caused by different LiDAR sensors and annotation methods, we develop a beam re-sampling technique for point cloud augmentation combined with class-balancing strategy. Furthermore, scalable pre-training is observed, that is, the downstream performance across all the experiments gets better with more pre-training data. Additionally, such pre-training strategy also remains compatible with unlabeled data. The hope is that our findings will facilitate the understanding of LiDAR points and pave the way for future advancements in LiDAR pre-training.