The usage of Unmanned Aerial Vehicle (UAV) is preferred in certain situations, such as when sending in human can be dangerous or if we want to deliver important payloads to a target person in a crowded area. In this paper, we demonstrate and present a development of an Unmanned Aerial Vehicle (UAV) system that is capable of an autonomous flight and identifying a target (person) from a group of people. The autonomous flight system consists of an offline path planning module and a real time object detection module. Results show that the path planning allows the UAV to plan its path with latitude and longitude accuracy of 0.072 m and 0.087 m respectively using the A* algorithm. Experiments also show that the object detection module allows depth estimation with an 83.5% accuracy using a single camera. For identifying target, the UAV employs facial recognition algorithm based on FaceNet which extract embedding features as representative of the face. The system can recognize a person up to a horizontal distance of 3.8 m and vertical distance of 3.0 m with a minimum FPS of 6.4 when the number of faces in the frame is 10. Results show that the develop UAV platform is potential to be used for identifying a person in a crowd.
Person Identification System for UAV
Advs in Intelligent Syst., Computing
International Conference on Innovative Technology, Engineering and Science ; 2020 December 22, 2020 - December 22, 2020
2021-03-11
11 pages
Article/Chapter (Book)
Electronic Resource
English
Person identification in TV programs
British Library Online Contents | 2001
|European Patent Office | 2018
|