M3DeTR: Multi-representation, Multi-scale, Mutual-relation 3D Object Detection with Transformers
This work addresses 3D object detection for autonomous driving by integrating diverse data representations, offering incremental improvements over existing methods.
The paper tackles 3D object detection by proposing M3DeTR, a novel architecture that unifies multiple point cloud representations and feature scales using transformers, achieving state-of-the-art performance with a 1.48% mAP improvement on Waymo Open Dataset and ranking first on KITTI for car and cyclist classes.
We present a novel architecture for 3D object detection, M3DeTR, which combines different point cloud representations (raw, voxels, bird-eye view) with different feature scales based on multi-scale feature pyramids. M3DeTR is the first approach that unifies multiple point cloud representations, feature scales, as well as models mutual relationships between point clouds simultaneously using transformers. We perform extensive ablation experiments that highlight the benefits of fusing representation and scale, and modeling the relationships. Our method achieves state-of-the-art performance on the KITTI 3D object detection dataset and Waymo Open Dataset. Results show that M3DeTR improves the baseline significantly by 1.48% mAP for all classes on Waymo Open Dataset. In particular, our approach ranks 1st on the well-known KITTI 3D Detection Benchmark for both car and cyclist classes, and ranks 1st on Waymo Open Dataset with single frame point cloud input. Our code is available at: https://github.com/rayguan97/M3DETR.