Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution
This addresses the need for efficient and accurate 3D perception in autonomous vehicles, particularly for recognizing small objects like pedestrians and cyclists, with incremental improvements over existing methods.
The paper tackles the problem of efficient 3D scene understanding for self-driving cars by proposing SPVConv, a lightweight module combining sparse convolution with point-based processing, and uses neural architecture search to find optimal networks. The resulting SPVNAS model outperforms state-of-the-art MinkowskiNet by 3.3% on SemanticKITTI while achieving 8x computation reduction and 3x speedup.
Self-driving cars need to understand 3D scenes efficiently and accurately in order to drive safely. Given the limited hardware resources, existing 3D perception models are not able to recognize small instances (e.g., pedestrians, cyclists) very well due to the low-resolution voxelization and aggressive downsampling. To this end, we propose Sparse Point-Voxel Convolution (SPVConv), a lightweight 3D module that equips the vanilla Sparse Convolution with the high-resolution point-based branch. With negligible overhead, this point-based branch is able to preserve the fine details even from large outdoor scenes. To explore the spectrum of efficient 3D models, we first define a flexible architecture design space based on SPVConv, and we then present 3D Neural Architecture Search (3D-NAS) to search the optimal network architecture over this diverse design space efficiently and effectively. Experimental results validate that the resulting SPVNAS model is fast and accurate: it outperforms the state-of-the-art MinkowskiNet by 3.3%, ranking 1st on the competitive SemanticKITTI leaderboard. It also achieves 8x computation reduction and 3x measured speedup over MinkowskiNet with higher accuracy. Finally, we transfer our method to 3D object detection, and it achieves consistent improvements over the one-stage detection baseline on KITTI.