BEVFormer.mp4
The official implementation of the paper "BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers".
Code will be released around June 2022.
[2022/3/10]: BEVFormer achieve the SOTA on nuScenes Detection Task with 56.9% NDS (camera-only)!
In this work, the authors present a new framework termed BEVFormer, which learns unified BEV representations with spatiotemporal transformers to support multiple autonomous driving perception tasks. In a nutshell, BEVFormer exploits both spatial and temporal information by interacting with spatial and temporal space through predefined grid-shaped BEV queries. To aggregate spatial information, the authors design a spatial cross-attention that each BEV query extracts the spatial features from the regions of interest across camera views. For temporal information, the authors propose a temporal self-attention to recurrently fuse the history BEV information. The proposed approach achieves the new state-of-the-art 56.9% in terms of NDS metric on the nuScenes test set, which is 9.0 points higher than previous best arts and on par with the performance of LiDAR-based baselines.
If this work is helpful for your research, please consider citing the following BibTeX entry.
@article{li2022bevformer,
title={BEVFormer: Learning Bird’s-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers},
author={Li, Zhiqi and Wang, Wenhai and Li, Hongyang and Xie, Enze and Sima, Chonghao and Lu, Tong and Qiao, Yu and Dai, Jifeng}
journal={arXiv preprint arXiv:2203.17270},
year={2022}
}