Revisiting Birds Eye View Perception Models with Frozen Foundation Models: DINOv2 and Metric3Dv2
This work addresses data efficiency in autonomous driving perception, offering incremental improvements by leveraging pre-trained models to enhance performance with less training.
The paper tackled the problem of reducing training data requirements for Birds Eye View perception models by integrating frozen foundation models DINOv2 and Metric3Dv2 into existing architectures, achieving a 7.4 IoU improvement with half the data and iterations for Lift-Splat-Shoot and a +3 IoU gain for Simple-BEV.
Birds Eye View perception models require extensive data to perform and generalize effectively. While traditional datasets often provide abundant driving scenes from diverse locations, this is not always the case. It is crucial to maximize the utility of the available training data. With the advent of large foundation models such as DINOv2 and Metric3Dv2, a pertinent question arises: can these models be integrated into existing model architectures to not only reduce the required training data but surpass the performance of current models? We choose two model architectures in the vehicle segmentation domain to alter: Lift-Splat-Shoot, and Simple-BEV. For Lift-Splat-Shoot, we explore the implementation of frozen DINOv2 for feature extraction and Metric3Dv2 for depth estimation, where we greatly exceed the baseline results by 7.4 IoU while utilizing only half the training data and iterations. Furthermore, we introduce an innovative application of Metric3Dv2's depth information as a PseudoLiDAR point cloud incorporated into the Simple-BEV architecture, replacing traditional LiDAR. This integration results in a +3 IoU improvement compared to the Camera-only model.