DESIGN AND IMPLEMENTATION OF A PROTOTYPE COMPUTER VISION-BASED OMNIDIRECTIONAL DRONE DETECTION SYSTEM

• Main Author: Ziyad Akbar, Muhammad
• Format: Final Project
• Language: Indonesia
• Subjects: Teknik (Rekayasa, enjinering dan kegiatan berkaitan)
• Online Access: https://digilib.itb.ac.id/gdl/view/82173
• Institution: Institut Teknologi Bandung

Drones (Unmanned Aerial Vehicles, UAVs) are increasingly utilized across various sectors due to their benefits and ease of use. However, they also pose safety and security risks to the public and violate airspace. Additionally, drones are difficult to detect with radar because of their small size and their typically non-metal composite materials. This research aims to develop a drone detection system using computer vision. The method involves creating a cost-effective and modular drone detector prototype for massive application. This research successfully developed a drone detection system using SSD MobileNetv2 on TensorFlow Lite, Raspberry Pi 4B, and a multi-camera. The system can be built cost-effectively and can effectively perform real-time omnidirectional detection with a precision of 0.53, a confidence score of 0.57, an FPS of 0.40, and a display delay of 3.70 seconds, detecting drones with a minimum bounding box size of 40x30 pixels. The results of this research can be used as a basis for developing affordable and efficient anti-drone technology.