Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Performance Evaluation of Deep Learning-based Quadrotor UAV Detection and Tracking Methods

Mohssen E. Elshaar, Zeyad M. Manaa, Mohammed R. Elbalshy, Abdul Jabbar Siddiqui, Ayman M. Abdallah

TL;DR

This study tackles the privacy and safety challenges posed by unmanned aerial vehicles by benchmarking RGB-based UAV detection and tracking using state-of-the-art detectors (YOLOv5 and YOLOv8) and trackers (BoT-SORT and ByteTrack) on the DUT Anti-UAV dataset. It establishes a comprehensive evaluation framework, releases public model weights and demos, and analyzes the strengths and limitations of each method, finding that YOLOv5 variants typically outperform YOLOv8 in detection while YOLOv8 can better handle blurred targets; BoT-SORT generally yields more accurate and stable tracking than ByteTrack. The work highlights important data limitations, threshold effects on precision/recall, and real-time deployment challenges, offering practical guidance for deploying robust anti-UAV surveillance systems. Overall, the results inform detector/tracker selection for real-time UAV monitoring and point to future research directions in data diversity and temporal tracking metrics to improve reliability in complex scenes.

Abstract

Unmanned Aerial Vehicles (UAVs) are becoming more popular in various sectors, offering many benefits, yet introducing significant challenges to privacy and safety. This paper investigates state-of-the-art solutions for detecting and tracking quadrotor UAVs to address these concerns. Cutting-edge deep learning models, specifically the YOLOv5 and YOLOv8 series, are evaluated for their performance in identifying UAVs accurately and quickly. Additionally, robust tracking systems, BoT-SORT and Byte Track, are integrated to ensure reliable monitoring even under challenging conditions. Our tests on the DUT dataset reveal that while YOLOv5 models generally outperform YOLOv8 in detection accuracy, the YOLOv8 models excel in recognizing less distinct objects, demonstrating their adaptability and advanced capabilities. Furthermore, BoT-SORT demonstrated superior performance over Byte Track, achieving higher IoU and lower center error in most cases, indicating more accurate and stable tracking. Code: https://github.com/zmanaa/UAV_detection_and_tracking Tracking demo: https://drive.google.com/file/d/1pe6HC5kQrgTbA2QrjvMN-yjaZyWeAvDT/view?usp=sharing

Performance Evaluation of Deep Learning-based Quadrotor UAV Detection and Tracking Methods

TL;DR

This study tackles the privacy and safety challenges posed by unmanned aerial vehicles by benchmarking RGB-based UAV detection and tracking using state-of-the-art detectors (YOLOv5 and YOLOv8) and trackers (BoT-SORT and ByteTrack) on the DUT Anti-UAV dataset. It establishes a comprehensive evaluation framework, releases public model weights and demos, and analyzes the strengths and limitations of each method, finding that YOLOv5 variants typically outperform YOLOv8 in detection while YOLOv8 can better handle blurred targets; BoT-SORT generally yields more accurate and stable tracking than ByteTrack. The work highlights important data limitations, threshold effects on precision/recall, and real-time deployment challenges, offering practical guidance for deploying robust anti-UAV surveillance systems. Overall, the results inform detector/tracker selection for real-time UAV monitoring and point to future research directions in data diversity and temporal tracking metrics to improve reliability in complex scenes.

Abstract

Unmanned Aerial Vehicles (UAVs) are becoming more popular in various sectors, offering many benefits, yet introducing significant challenges to privacy and safety. This paper investigates state-of-the-art solutions for detecting and tracking quadrotor UAVs to address these concerns. Cutting-edge deep learning models, specifically the YOLOv5 and YOLOv8 series, are evaluated for their performance in identifying UAVs accurately and quickly. Additionally, robust tracking systems, BoT-SORT and Byte Track, are integrated to ensure reliable monitoring even under challenging conditions. Our tests on the DUT dataset reveal that while YOLOv5 models generally outperform YOLOv8 in detection accuracy, the YOLOv8 models excel in recognizing less distinct objects, demonstrating their adaptability and advanced capabilities. Furthermore, BoT-SORT demonstrated superior performance over Byte Track, achieving higher IoU and lower center error in most cases, indicating more accurate and stable tracking. Code: https://github.com/zmanaa/UAV_detection_and_tracking Tracking demo: https://drive.google.com/file/d/1pe6HC5kQrgTbA2QrjvMN-yjaZyWeAvDT/view?usp=sharing
Paper Structure (32 sections, 4 equations, 14 figures, 5 tables)

This paper contains 32 sections, 4 equations, 14 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Contribution
  3. Related Work
  4. RGB-based Tracking
  5. RGB and Depth-based Tracking
  6. RGB and Thermal Images-based Tracking
  7. TIR based tracking
  8. UAV tracking and detection
  9. Datasets
  10. MAV-VID
  11. Drone-birds dataset coluccia2019drone
  12. AntiUAV Jiang2021AntiUAV
  13. DUT dataset
  14. Experimental Methodology and Setup
  15. Data Structure
  16. ...and 17 more sections

Figures (14)

  • Figure 1: Hawks attacking UAVs, illustrating a potential natural anti-drone defense mechanism.
  • Figure 2: Samples from the DUT Anti-UAV dataset zhao2022vision.
  • Figure 3: Aspect ratio statistics for the used images within the dataset.
  • Figure 4: Position distribution of the object(s) within the used images in the dataset.
  • Figure 5: Area ratio between the object(s) and the image size in the dataset.
  • ...and 9 more figures