Correlation Filters for Unmanned Aerial Vehicle-Based Aerial Tracking: A Review and Experimental Evaluation
This code library gives our experimental results and most of the publicly available DCF-based trackers.
The trackers are in folder tracker_set and the results are in all_trk_result.
In folder demo, we offer an example on how to utilize the trackers in this code library and how to implement the tracking procedure.
The submitted version of our paper has been uploaded to arxiv which can be found at:
https://arxiv.org/abs/2010.06255.
If you want to use our experimental results or related content, please cite our paper using the format as follows:
@ARTICLE{9445732,
author={Fu, Changhong and Li, Bowen and Ding, Fangqiang and Lin, Fuling and Lu, Geng},
journal={IEEE Geoscience and Remote Sensing Magazine},
title={Correlation Filters for Unmanned Aerial Vehicle-Based Aerial Tracking: A Review and Experimental Evaluation},
year={2021},
volume={},
number={},
pages={2-387},
doi={10.1109/MGRS.2021.3072992}
}
The trackers are tested on the following platform.
- windows 10 64-bit
- Intel Core i7-8700K（3.70GHz）
- 32G RAM
- Nvidia GeForce RTX 2080
- Matlab 2019a
- CUDA10
- VS2017
All the DCF-based trackers' results are obtained using a single CPU with a single core.
Here shows some of the tracking results of 21 handcrafted DCF-based trackers.
The deep trackers used GPU acceleration, while the DCF-based tracker AutoTrack used a single CPU only.
We thank the contribution of Siyi Li, Matthias Muller, Dawei Du, and Pengfei Zhu for their previous work of the benchmarks DTB70, UAV123, UAV123@10fps, UAV20L, UAVDT, and VisDrone2019-SOT. Their papers and benchmark address are listed here.
paper: Visual object tracking for unmanned aerial vehicles: a benchmark and new motion models
paper site: https://dl.acm.org/doi/10.5555/3298023.3298169
code and benchmark site: https://github.com/flyers/drone-tracking
paper: A Benchmark and Simulator for UAV Tracking
paper site: https://link.springer.com/chapter/10.1007%2F978-3-319-46448-0_27
code and benchmark site: https://cemse.kaust.edu.sa/ivul/uav123
paper: The Unmanned Aerial Vehicle Benchmark: Object Detection and Tracking
paper site: https://link.springer.com/article/10.1007/s11263-019-01266-1
code and benchmark site: https://sites.google.com/site/daviddo0323/projects/uavdt
paper: VisDrone-SOT2019: The Vision Meets Drone Single Object Tracking Challenge Results
paper site: https://ieeexplore.ieee.org/document/9022632
code and benchmark site: http://www.aiskyeye.com/