Self-Supervised Pre-training with Combined Datasets for 3D Perception in Autonomous Driving
This addresses the need for better 3D perception models in autonomous driving, though it is incremental by building on existing pre-training ideas from NLP and 2D vision.
The paper tackles the problem of improving 3D perception in autonomous driving by proposing a self-supervised pre-training framework that uses massive unlabeled data from heterogeneous datasets, resulting in significant performance improvements on downstream tasks like 3D object detection and BEV segmentation, with steady gains as data volume scales.
The significant achievements of pre-trained models leveraging large volumes of data in the field of NLP and 2D vision inspire us to explore the potential of extensive data pre-training for 3D perception in autonomous driving. Toward this goal, this paper proposes to utilize massive unlabeled data from heterogeneous datasets to pre-train 3D perception models. We introduce a self-supervised pre-training framework that learns effective 3D representations from scratch on unlabeled data, combined with a prompt adapter based domain adaptation strategy to reduce dataset bias. The approach significantly improves model performance on downstream tasks such as 3D object detection, BEV segmentation, 3D object tracking, and occupancy prediction, and shows steady performance increase as the training data volume scales up, demonstrating the potential of continually benefit 3D perception models for autonomous driving. We will release the source code to inspire further investigations in the community.