CNN-based Off-board Computation for Real-time Object Detection and Tracking Using a Drone : Sophan Wahyudi Nawawi, Ahmed Ashraf Mohamed Ahmed Abdou, Nur Azlina Abd Aziz*
Year: 2021
Type: article
Abstract: Object detection and tracking have become one of the most important applications for UAVs, especially in surveillance applications. Single object tracking using a drone is an active field for research due to its importance in surveillance. In this paper, a real-time human and cars detection and tracking approach are proposed using a CNN-based technique by combining the state-of-the-art object detection algorithm YOLOv4 with the state-of-the-art multi objects tracking algorithm DeepSORT using video streaming from a drone for the purpose of surveillance. Furthermore, an algorithm was developed to enhance the observation ability by choosing a single object to be tracked. The algorithm was implemented using Tello drone with a developed navigation algorithm based on the vision analysis. Keywords: Object detection, Single object tracking, drone, CNN, UAV
Authors: Sophan Wahyudi Nawawi, Ahmed Ashraf Ahmed Abdou, Nor Azlina Abd Aziz
Institutions: University of Technology Malaysia, Multimedia University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W3216802972
