Unleashing HyDRa: Hybrid Fusion, Depth Consistency and Radar for Unified 3D Perception
Philipp Wolters, Johannes Gilg, Torben Teepe, Fabian Herzog, Anouar Laouichi, Martin Hofmann, Gerhard Rigoll
This repository is an official implementation of HyDRa, our novel camera-radar fusion architecture for 3D perception.
[2024/3/19]Official PyTorch implementation is coming soon.[2024/3/12]Our preprint is available on arXiv.
3D Object Detection on nuScenes
| Method | Backbone | NDS | mAP | Config | Checkpoint |
|---|---|---|---|---|---|
| HyDRa | R50 | 58.5 | 49.4 | config | model |
| HyDRa | R101 | 61.7 | 53.6 | config | model |
| HyDRa | V2-99 | 64.2 | 57.4 | config | model |
3D Semantic Occupancy Prediction on Occ3D
| Method | Backbone | mIoU | Config | Checkpoint |
|---|---|---|---|---|
| FB-OCC | R50 | 39.1 | config | model |
| HyDRa | R50 | 44.4 | config | model |
If this work is helpful for your research or applications, please consider giving us a star 🌟 and citing it by the following BibTeX entry.
@article{wolters2024unleashing,
title={Unleashing HyDRa: Hybrid Fusion, Depth Consistency and Radar for Unified 3D Perception},
author={Wolters, Philipp and Gilg, Johannes and Teepe, Torben and Herzog, Fabian and Laouichi, Anouar and Hofmann, Martin and Rigoll, Gerhard},
journal={arXiv preprint arXiv:2403.07746},
year={2024}
}We would like to thank other great open source projects:
- BEVDet, SoloFusion, FB-BEV , CRN, Occ3D