UAV
AU-AIR
Aerial traffic detection with flight metadata
Aerial traffic images from eight source clips with eight-class detection boxes and frame-associated flight-state metadata.
Access links
Primary places to inspect, download, read about, or reproduce the dataset.
Quick facts
- Vehicle type
- UAV
- Environment
- Outdoor
- Data origin
- Real-world
- Sensor count
- 2
- Ground truth
- Not reported
- Calibration
- Not reported
- Annotations
- Available
Sensor overview
High-level sensor availability before the detailed sensor records below.
Ground truth
Reference scope, method, rate, coverage, and provenance.
Not reported
Calibration and synchronization
Reported calibration level, reproducibility signals, and supporting notes.
- Level
- Not reported
- Processed parameters
- Not reported
- Raw calibration data
- Not reported
- Calibration targets
- Not reported
Calibration data
Calibration reproducibility details are not reported.
Notes
Not reportedKnown or intentional limitations
Reported constraints and characteristics to check before using the dataset.
The release describes 32,823 annotated images and 132,034 instances from eight source video clips at one Aarhus intersection. Eight source clips do not imply eight downloadable raw videos.
The old project/API pages return 404. The author-owned website source retains the official image/annotation links; both Drive preview pages were reachable during review, but bulk archive contents were not audited.
Conflicting paper split descriptions and 5 m versus 10 m lower altitude descriptions remain unresolved. The reported maximum altitude is 30 m.
Derived flight-state metadata does not establish raw IMU/GNSS streams or native device rates. The 50 Hz logger and 5 fps extracted frames have different roles.
Class imbalance, one intersection and windless recording conditions limit generalization. Some objects may be missed or loosely boxed; calibration and achieved alignment error remain unknown.
Cameras
Camera 1
- Model
- Parrot Bebop 2 onboard camera
- Setup
- Mono
- Count
- 1
- Effective cameras
- 1
- Modality
- Rgb
- Resolution
- 1920 x 1080
- Rate
- 5 Hz
- Shutter
- Not Reported
- Lens type
- Not Reported
- HFOV
- Not reported
- VFOV
- Not reported
- Notes
- Released annotated images are sampled at 5 fps from 30 fps source video. Public raw-video distribution is not established by the author download page.
IMUs
Not reported
GNSS
Not reported
LiDAR
Not reported
Radar
Not reported
Additional sensors
Odometry
- Name
- Frame-associated flight state
- Count
- 1
- Notes
- Frame metadata gives time, location, altitude, Euler angles and body velocity. The collection logger runs at 50 Hz; raw accelerometer/gyroscope or native GNSS sample streams and the exact delivered motion-state cadence are not established.
Annotations
Reported annotation availability and task support.
- Availability
- Available
- Format
- Manual image boxes/category indices; eight traffic classes and frame-aligned flight metadata.
- Class count
- 8
Semantic segmentation
Not reported
Instance segmentation
Not reported
Object detection
Available
Optical flow
Not reported
Depth ground truth
Not reported
Citation
Citation key and BibTeX kept at the end of the page for reference.
@misc{bozcan2020auair,
author = {{Ilker Bozcan} and {Erdal Kayacan}},
title = {AU-AIR: A Multi-modal Unmanned Aerial Vehicle Dataset for Low Altitude Traffic Surveillance},
year = {2020},
eprint = {2001.11737},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2001.11737v2}
}Dataset corrections
Notice a specific issue after reviewing the page?
Report an inaccuracy