/auairdataset

Python API and tools for the AU-AIR dataset

Primary LanguagePythonMIT LicenseMIT

AU-AIR Dataset API

News

16.09.2020

  • Researchers from Fraunhofer IOSB in Ettlingen have released a subset of the AU-AIR dataset, namely AU-AIR-Track. AU-AIR-Track contains:
    • Two images sequences ranging from 887 frames and 512 frames, containing respectively 63 and 27 annotated objects.
    • Annotations for visual object tracking, with occlusion (also annotated).
    • Two 3D reconstructions for each image sequence, and pseudo depth maps associated with the camera poses. Paper, Dataset and Tools

15.09.2020

  • We have updated the annotation file to improve annotations. The new link can be found in "Download" section below.
  • We have also released a json file for VGG Image Annotator to edit annotations. Feel free to customize annotations for your research. You can download AU-AIR VIA annotations here.

Dataset

The AU-AIR is a multi-modal aerial dataset captured by a UAV. Having visual data, object annotations, and flight data (time, GPS, altitude, IMU sensor data, velocities), AU-AIR meets vision and robotics for UAVs.

https://bozcani.github.io/auairdataset

alt text

Specifications

  • 8 raw RGB videos (more than 2 hours in total)
  • 32,283 extracted and labelled frames
  • Bounding box annotations for eight objects related to traffic:
    • human, car, van, truck, bike, motorbike, bus, trailar
  • Time, GPS, altitude, IMU sensor data and linear velocities of the drone are avaliable for each extracted frame.

Download

Please download both the AU-AIR images and annotations to run the demo and use the API:

Images: https://drive.google.com/open?id=1pJ3xfKtHiTdysX5G3dxqKTdGESOBYCxJ (2.2 GB)

Annotations (V.1.1): https://drive.google.com/file/d/1GyoBK-NalDFfAtRt9LO6FBujbObyaZLv/view?usp=sharing (55 MB)

Dependencies

You will need common dependencies like numpy and opencv.

Installation

To install the package from source, simply clone or download the repository to your machine git clone https://github.com/bozcani/auairdataset

cd auairdataset

python setup.py install

References

I. Bozcan and E. Kayacan, "AU-AIR: A Multi-modal Unmanned Aerial Vehicle Dataset for Low Altitude Traffic Surveillance", IEEE International Conference on Robotics and Automation (ICRA) 2020.