Assisting UAV Localization Via Deep Contextual Image Matching

In this article, we aim to explore the potential of using onboard cameras and pre-stored geo-referenced imagery for Unmanned Aerial Vehicle (UAV) localization. Such a vision-based localization enhancing system is vital importance, particularly in situations where integrity global positioning (GPS) question (i.e., occurrence GPS outages, jamming, etc.). To end, propose complete trainable pipeline localize an aerial image orthomosaic map context UAV The proposed deep architecture extracts features from localizes it pre-ordained, larger, geotagged image. idea train learning model find neighborhood consensus patterns that encapsulate local established dense feature correspondences by introducing semi-local constraints. We qualitatively quantitatively evaluate performance our approach on real imagery. training testing data acquired via multiple flights over different regions. source code along with entire dataset, including annotations collected images has been made public. 11 https://github.com/m-hamza-mughal/Aerial-Template-Matching. Up-to knowledge, such dataset novel first its kind which consists 2052 high-resolution at times three areas Pakistan spanning total area around 2 km xmlns:xlink="http://www.w3.org/1999/xlink">2 .