BONAI

BONAI (Buildings in Off-Nadir Aerial Images) is a dataset for building footprint extraction (BFE) in off-nadir aerial images.

[Paper] [Dataset]

Description

BONAI contains 268,958 building instances across 3,300 aerial images with fully annotated instance-level roof and footprint for each building as well as the corresponding offset vector. Compared to BONAI, existing BFE datasets only annotate building footprints.

The images of BONAI are taken from six representative cities of China, i.e., Shanghai, Beijing, Harbin, Jinan, Chengdu, and Xi'an, the detailed number of images and object instances per image set and city are reported in the below table.

Download

You can download the dataset on Google Driver.

Evaluation

Training, Validation and Testing sets are publicly available. The evaluation codes will be upload soon.

LOFT & FOA

The codes of LOFT and FOA is now publicly available. You can refer to MMDetection to install and run this project.

Contact

This repo is currently maintained by Jinwang Wang ([email protected]).

Citing

If you use BONAI dataset, codebase or models in your research, please consider cite.

@article{wang2022bonai,
  author={Wang, Jinwang and Meng, Lingxuan and Li, Weijia and Yang, Wen and Yu, Lei and Xia, Gui-Song},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, 
  title={Learning to Extract Building Footprints from Off-Nadir Aerial Images}, 
  year={2022},
  volume={},
  number={},
  pages={1-1},
  doi={10.1109/TPAMI.2022.3162583}}