Abstract
This work presents a deep learning based approach for UAV geo-localization through cross-view image matching. The method employs a coarse-to-fine strategy that combines Local Pattern Network (LPN) for initial coarse localization and Deep Feature Matcher (DFM) for refinement. The system takes UAV query images and matches them against a satellite image database to determine precise geographic locations. Our approach addresses the challenging problem of matching images captured from significantly different viewpoints, which is essential for autonomous drone navigation and location awareness in service robotics applications. The method demonstrates robust performance in handling the complex transformations between aerial and satellite views.
First image description.
Second image description.
Third image description.
Fourth image description.
Video Presentation
Another Carousel
Poster
BibTeX
@INPROCEEDINGS{10137193,
author={Luo, Xubo and Tian, Yaolin and Wan, Xue and Xu, Jingzhong and Ke, Tao},
booktitle={2022 International Conference on Service Robotics (ICoSR)},
title={Deep learning based cross-view image matching for UAV geo-localization},
year={2022},
volume={},
number={},
pages={102-106},
keywords={Location awareness;Deep learning;Satellites;Service robots;Image matching;Refining;Lighting;deep learning;image matching;geo-localization;autonomous drone navigation},
doi={10.1109/ICoSR57188.2022.00028}
}