Design of Bird Repellent System Based on Machine Vision Detection and Localization

Bird strike phenomenon generally refers to incidents where aircraft structural damage due to collisions with birds during low-altitude flight. Such incidents can lead to accidents, casualties, and economic losses. The civil aviation industry has attached great importance to bird strike prevention, devising regulations and standards to mitigate bird strike risks through scientific management and measures. Machine vision bird repellent systems, an environmentally friendly and non-contact method, align with current environmental trends. This paper comprehensively employed and fused traditional YOLOv5, KCF, Haarcascade, and Adaboost for target recognition and tracking. It proposes constructing a visual gimbal using a Raspberry Pi, servos, camera, etc. Python programming language is utilized to implement multi-mode bird identification within the field of view and issue voice alerts to ground control personnel promptly upon bird detection. The tracking device is activated according to predetermined conditions to efficiently and accurately handle bird situations. The designed system can accurately identify and track birds, thereby helping to reduce the probability of bird strike incidents and enhance flight safety.