Aerovigilance: A Comprehensive System for Unauthorized Drone Detection”

he rapid rise of Unmanned Aerial Vehicles (UAVs) has created major security risks in restricted airspaces.
Traditional methods (radar, RF, acoustic) are costly, unreliable, and environment-sensitive.
The paper proposes a hybrid framework:
- YOLOv5 deep learning model for real-time visual drone detection.
- Wi-Fi authentication to differentiate authorized vs. unauthorized drones.

1. “ Aerovigilance : A Comprehensive System for Unauthorized Drone Detection” (Paper ID-34) Mukta Jamage Assistant Professor Pimpri Chinchwad College of Engineering Pune, India

2. Introduction Problem Statement Literature Survey Proposed Methodology Results Conclusion References Outline

3. Introduction The rapid rise of Unmanned Aerial Vehicles (UAVs) has created major security risks in restricted airspaces. Traditional methods (radar, RF, acoustic) are costly, unreliable, and environment-sensitive. The paper proposes a hybrid framework: - YOLOv5 deep learning model for real-time visual drone detection . - Wi-Fi authentication to differentiate authorized vs. unauthorized drones.

4. Introduction

5. Problem Statement Objectives Existing detection systems focus only on visual or RF tracking but cannot verify if a drone is authorized. . Develop a real-time YOLOv5-based detection system.. Implement Wi-Fi authentication for drone verification . . Enable real-time alert generation and notification for unauthorized drone detection . . Problem Statement and Objectives 03 01 02 03 Unauthorized drones may exploit this gap and bypass current systems. Need for a hybrid detection + authentication approach .

6. Literature Survey Author / Year Method / Approach Strengths Limitations Chen et al., 2024 [1] Deep learning-based real-time detection on resource-constrained devices Works efficiently on low-power devices, fast inference Limited scalability for larger drones, dataset dependence Lee et al., 2023 [2] Multimodal detection (Radar + Thermal + Optical sensors) Robust in diverse weather, combines multiple signals High cost, requires multi-sensor deployment Yilmaz & Kutbay, 2024 [3][16][18] YOLOv8-based visual detection High accuracy, efficient for small drones, transformer modules improve detection Needs high-quality dataset, not tested with authentication Abu Zitar et al., 2023 [4][15] Review of UAV visual detection & tracking (YOLO, Faster R-CNN, SSD) Covers wide range of methods, highlights strengths Lacks implementation details, survey only Kumar et al., 2024 [5][17][19] AI-based detection using flight behavior patterns Can detect stealth/spoofing attempts, behavior-based Computationally heavy, limited by dataset Unlu et al., 2023 [6][14] Deep learning strategies for drone detection Good accuracy, real-time vision-based detection Affected by lighting, occlusions

7. Author / Year Method / Approach Strengths Limitations ITU Report, 2023 [9] UAV detection & countermeasure systems Global standardization, includes countermeasures Mostly technical guidelines, not a detection model Rosales, 2022 [10][12][21] Drone surveillance technologies (vision, RF, acoustic) Broad coverage of tech, practical applications Limited AI integration PyImageSearch, 2023 [11][13] YOLO object detection tutorials & resources Practical, widely used Not specialized for drones NVIDIA, 2024 [20] Edge computing for AI Efficient for real-time UAV detection on devices Hardware dependency Literature Survey

8. Proposed Methodology Two main modules : 1. YOLOv5 visual detection – real-time drone recognition . 2. Wi-Fi authentication – verifies connectivity to a predefined authorized network. Dataset Used : 1. Kaggle UAV datasets - 25000 drone images, classes-1. - Resolutions: Mix of 720p / 1080p video frames , - preprocessed to 640×640 for YOLO. 2. Wi-Fi authentication – verifies connectivity to a predefined authorized network.

9. Proposed Methodology Dataset Used : 1. Kaggle UAV datasets - 25000 drone images, classes-1. - Resolutions: Mix of 720p / 1080p video frames , - preprocessed to 640×640 for YOLO. 2. Wi-Fi authentication – verifies connectivity to a predefined authorized network.

10. YoLov5 Architecture

11. Results

12. Conclusion The proposed YOLOv5 + Wi-Fi hybrid system offers a low-cost, scalable, and efficient solution for drone detection. Provides high detection accuracy with real-time authorization capability. Ideal for deployment in airports, defense installations, and critical infrastructure to enhance airspace security.

13. References J. R. Chen, H. Zhang, and P. Lee, “Real-Time Drone Detection via Deep Learning on Resource-Constrained Devices,” IEEE Transactions on Aerospace and Electronic Systems, vol. 55, no. 6, pp. 2787–2802, 2024. [2] H. Lee, T. Kim, and C. Park, “Multimodal Detection of UAVs Using Radar, Thermal, and Optical Sensors,” International Journal of Advanced Robotic Systems, vol. 18, no. 2, pp. 1–10, 2023. [3] B. Yilmaz and U. Kutbay , “YOLOv8-Based Drone Detection: Performance Analysis and Optimization,” Computer Vision Applications, vol. 14, no. 5, pp. 1093–1105, 2024. [4] R. Abu Zitar , M. Alhassoun , and Y. Al-Amri, “A Review of UAV Visual Detection and Tracking Methods,” IEEE Sensors Journal, vol. 21, no. 3, pp. 2834–2845, 2023. [5] M. Kumar, S. Gupta, and A. Verma, “AI-Based Detection of Drones Using Flight Behavior Patterns,” Journal of Robotics and Automation, vol. 7, no. 2, pp. 245–259, 2024. [6] E. Unlu, R. Khan, and J. Watson, “Deep Learning-Based Strategies for Drone Detection,” IEEE Access, vol. 8, pp. 76345–76360, 2023. . Jain, Computer Vision: Algorithms and Applications, Berlin: Springer, 2020. [8] R. Sutton and A. Barto, Reinforcement Learning: An Introduction, 2nd ed., MIT Press, 2018 . [ 9] “UAV Detection and Countermeasure Systems,” International Telecommunications Union (ITU), Technical Report, 2023. [10] M. Rosales, Drone Surveillance Technologies and Applications, Cambridge University Press, 2022.

14. 02 nd INTERNATIONAL CONFERENCE ON INTEGRATION OF COMPUTATIONAL INTELLIGENT SYSTEMS (ICICIS 2025) Thank You!! Mrs. Mukta Jamage Assistant Professor- PCCoE IT Mobile No: [phone] Mail ID: [email protected]