GaussianOcc: Fully Self-supervised and Efficient 3D Occupancy Estimation with Gaussian Splatting
This addresses the need for efficient and self-supervised 3D perception in autonomous driving and robotics, though it appears incremental as it builds on existing Gaussian splatting techniques.
The paper tackles the problem of 3D occupancy estimation by introducing GaussianOcc, which uses Gaussian splatting to enable fully self-supervised training without ground truth poses and achieves competitive performance with 2.7× faster training and 5× faster rendering.
We introduce GaussianOcc, a systematic method that investigates the two usages of Gaussian splatting for fully self-supervised and efficient 3D occupancy estimation in surround views. First, traditional methods for self-supervised 3D occupancy estimation still require ground truth 6D poses from sensors during training. To address this limitation, we propose Gaussian Splatting for Projection (GSP) module to provide accurate scale information for fully self-supervised training from adjacent view projection. Additionally, existing methods rely on volume rendering for final 3D voxel representation learning using 2D signals (depth maps, semantic maps), which is both time-consuming and less effective. We propose Gaussian Splatting from Voxel space (GSV) to leverage the fast rendering properties of Gaussian splatting. As a result, the proposed GaussianOcc method enables fully self-supervised (no ground truth pose) 3D occupancy estimation in competitive performance with low computational cost (2.7 times faster in training and 5 times faster in rendering). The relevant code is available in https://github.com/GANWANSHUI/GaussianOcc.git.