Unmanned Aerial Vehicle Target Detection Integrating Computer Deep SORT Algorithm and Wireless Signal

Unmanned Aerial Vehicle Target Detection Integrating Computer Deep SORT Algorithm and Wireless Signal

Ao Li (Yangzhou Marine Electronic Instrument Institute, China)
DOI: 10.4018/IJITN.373429
Article PDF Download
Open access articles are freely available for download

Abstract

With the advancement of unmanned aerial vehicle (UAV) technology, accurately detecting UAV targets has become increasingly challenging. This study addresses this issue by proposing a novel UAV target detection method that integrates real-time target tracking algorithms with wireless signal detection technology. Experimental results demonstrate that each improved module positively contributes to the overall detection method. Compared to traditional object detection approaches, the proposed method achieves superior performance on both the VisDrone and COCO datasets, with precision, recall, F1 score, and mean squared error values of 96.07%, 95.84%, 96.33%, and 0.023%, respectively. This integrated approach effectively enhances the accuracy of UAV target detection, offering a robust solution for positioning and tracking in UAV applications.
Article Preview
Top

Introduction

With the widespread application of unmanned aerial vehicles (UAVs) in military reconnaissance, environmental monitoring, traffic monitoring, and other fields, the accurate and quick detection and tracking of targets have become key to improving UAV performance (Al-lQubaydhi et al., 2024; Deng & Wang, 2023; Wang et al., 2024). However, traditional object detection methods often have limitations in both real-time tracking and accuracy, making it difficult to meet the current needs of UAV object detection (Shen et al., 2023). With the development of technology, real-time object tracking algorithms based on deep learning (simple online and real-time tracking with deep association [Deep SORT]) metric have gradually entered the field of view of technical personnel. The Deep SORT object tracking algorithm combines the feature extraction capability of deep learning with the real-time performance of the simple online and real-time tracking (SORT) algorithm, which can effectively recognize and distinguish targets. It is very suitable for scenarios that require a fast response, and it has been explored by many researchers (Mathias et al., 2022; Tu et al., 2024). Wu et al. (2024) proposed an optimized Deep SORT metric that is based on the You Only Look Once Version 7 (YOLO v7) algorithm to address bidirectional passenger flow detection and tracking in integrated hub stations. The experimental results indicated that the algorithm was helpful in achieving real-time detection and accurate tracking of bidirectional passenger flow at stations. Once an abnormal situation occurred, station staff could quickly respond, improving the operational safety of the station. Hu & Zhang (2023) proposed a nighttime trajectory extraction framework on the basis of an optimized Single-Shot MultiBox Detector and Deep SORT algorithm to address the common interference issues in traffic monitoring and recording of nighttime videos. In the performance evaluation of the 70-M Visibility dataset, the Multiple-Object Tracking Accuracy of the framework was 43.06%. K. Zhang et al. (2024) proposed a monocular vehicle speed detection method that is based on the improved YOLOX and Deep SORT for simple vehicle scenes with fixed shooting angles, which offers lower accuracy but better cost efficiency . The average error of the effective speed value measured by this method was 2.10km/hr. Compared with different versions of YOLOX, the improved object detection model increased the average accuracy from 2% to 4%.

Complete Article List

Search this Journal:
Reset
Volume 17: 1 Issue (2025)
Volume 16: 1 Issue (2024)
Volume 15: 1 Issue (2023)
Volume 14: 1 Issue (2022)
Volume 13: 4 Issues (2021)
Volume 12: 4 Issues (2020)
Volume 11: 4 Issues (2019)
Volume 10: 4 Issues (2018)
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing