UAVDE: UAV Distance Estimation dataset


Abstract


The UAV Distance Estimation (UAVDE) dataset is specifically curated for the purpose of UAV target distance estimation. Rather than obtaining the dataset through simulation, all the images and distance measurements in the UAVDE dataset are derived from real-world scenarios. This ensures that the dataset accurately represents the challenges and complexities encountered in actual UAV target distance estimation. The following shows further information,

  • 3,895 pairs of stereo images, each with its corresponding distance.

  • Each individual image is collected at a resolution of $1280*720$ pixels.

  • The horizontal and vertical field of view (FOV) are set at $22^\circ$ and $18^\circ$, respectively.

  • The stereo camera has a baseline distance of $406mm$.

  • To capture the UAV targets accurately, the lenses used have a focal length of $12mm$, and the sensor size is $3.75mm*3.75mm$.

Samples



Devices

We use the AMOV P600 as the recording UAV for this study. It is equipped with both a stereo camera and an ultra-wideband (UWB) sensor. The target UAV we focus on is the DJI M200. Please refer to the following figures for more details:

Downloads



  • Stereo Images


  • Annotations



If the above links are not accessable, you could download the dataset using Baidu Drive(extract code:fiaq).
For more details of the dataset, please refer to the paper "Revisting Stereo Triangulation in UAV Distance Estimation".
Notes. The dataset is a sample dataset for now. The complete dataset will be released after the paper is published.

Agreement

  • The UAVDE dataset is available for non-commercial research purposes only.
  • You agree not to reproduce, duplicate, copy, sell, trade, resell or exploit for any commercial purposes, any portion of the images and any portion of derived data.
  • You agree not to further copy, publish or distribute any portion of the UAVDE dataset. Except, for internal use at a single site within the same organization it is allowed to make copies of the dataset.
  • The Imagelab,Shoutou University reserves the right to terminate your access to the UAVDE dataset at any time.

Materials


Citation

@article{zhuang2023does,
                title={Revisiting Stereo Triangulation in UAV Distance Estimation},
                author={Jiafan Zhuang and Duan Yuan and Rihong Yan and Weixin Huang and 
Wenji Li and Zhun Fan},
                year={2023},
                eprint={2306.08939},
                archivePrefix={arXiv},
                primaryClass={cs.CV}
                }

Contact


Please contact Image Lab, Shantou University for questions about the dataset.