# ParkRecon3D：a surround-view dataset for parking-aware reconstruction. 

## News
- **[2026/2/21]** ParkGaussian is accepted by CVPR2026!(https://arxiv.org/abs/2601.01386)

## Data detail
It contains synchronized image streams from four fisheye cameras covering full 360° surroundings and four representative underground parking scenes. In total, the dataset includes 40K multi-camera sensor frames and 60K human-annotated parking-slot labels, supporting parking-slot perceptionand geometry-aware 3D reconstruction.

#### Four fisheye image sequences (10Hz@1280x960)
#### One Bird's Eye View (BEV) image sequence (10Hz@1354x1632, 1.05cm/pixel) generated by Around View Monitor (AVM) subsystem(https://arxiv.org/abs/2309.08180)
#### Four sets of wheel encoder data
#### One IMU dataset
#### Parking Space calibration

## Citation

If you find this project useful for your research, please consider citing our paper:
```
@inproceedings{wei2026parkgaussian,
  title={ParkGaussian: Surround-view 3D Gaussian Splatting for Autonomous Parking},
  author={Wei, Xiaobao and Ye, Zhangjie and Gu, Yuxiang and Zhu, Zunjie and Guo, Yunfei and Shen, Yingying and Zhao, Shan and Lu, Ming and Sun, Haiyang and Wang, Bing and others},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2026}
}

```

---
license: apache-2.0
---