metadata
language:
- en
task_categories:
- image-text-to-text
tags:
- remote-sensing
- spatial-understanding
- computer-vision
- drone-imagery
AirSpatial Dataset
AirSpatial is a spatially-aware remote sensing dataset introduced in the paper "AirSpatialBot: A Spatially-Aware Aerial Agent for Fine-Grained Vehicle Attribute Recognization and Retrieval". It specifically addresses vehicle imagery captured by drones and includes over 206K instructions.
Key Features
- Novel Tasks: Introduces Spatial Grounding (SG) and Spatial Question Answering (SQA).
- 3D Annotations: It is the first remote sensing grounding dataset to provide 3D Bounding Boxes (3DBB).
- Fine-grained Attributes: Focuses on vehicle attribute recognition, including brand, model, and pricing information from aerial imagery.
- Large Scale: Comprises over 206,000 instruction-following samples designed to enhance spatial understanding in Vision-Language Models (VLMs).
The dataset was used to train AirSpatialBot, an aerial agent capable of identifying specific vehicle details like brands (e.g., BYD, Tesla) and models from an aerial perspective.
Citation
If you find this dataset or the associated work useful, please cite:
@ARTICLE{zhou2025airspatialbot,
author={Zhou, Yue and Ding, Ran and Yang, Xue and Jiang, Xue and Liu, Xingzhao},
journal={IEEE Transactions on Geoscience and Remote Sensing},
title={AirSpatialBot: A Spatially-Aware Aerial Agent for Fine-Grained Vehicle Attribute Recognization and Retrieval},
year={2025},
volume={},
number={},
pages={1-1},
doi={10.1109/TGRS.2025.3570895}
}