This repo contains the deep hyperspectral kernelized correlation filter tracker MATLAB implementation. More information on the HKCF can be found in our following paper :
@article{uzkent2017tracking,
title={Tracking in Aerial Hyperspectral Videos using Deep Kernelized Correlation Filters},
author={Uzkent, Burak and Rangnekar, Aneesh and Hoffman, Matthew J},
journal={arXiv preprint arXiv:1711.07235},
year={2017}
}
Our HKCF tracker utilizes two features to learn to minimize the ridge regression function. They are
- Fast Histogram of Oriented Gradients
- VGGNet 5th Layer Convolutional Features.
The HKCF tracker is designed to track objects from aerial platforms that can record RGB images. We test it
on our synthetic hyperspectral scenario generated by the Digital Imaging and Remote Sensing
software.
Our scenario comes with 61
channels captured over the part of Rochester, NY. The duration of the
scenario is 110 sec. and comes with 157 frames and 1.42 frame rate per second.
To download the hyperspectral video to test HKCF tracker
wget https://drive.google.com/file/d/0B3lpS7qMFUmwTUQwaUpiOVN2SDA/view
To download the ground truth for the 89 vehicles
wget https://buzkent86.github.io/Datasets/GroundTruth.zip
If you use our synthetic scenario in your research, please cite our following paper :
@inproceedings{uzkent2016real,
title={Real-time vehicle tracking in aerial video using hyperspectral features},
author={Uzkent, Burak and Hoffman, Matthew J and Vodacek, Anthony},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops},
pages={36--44},
year={2016}
}