Real-Time Wild Animal Intrusion Detection and Repellent System Using YOLOv5n and Predator Scent
Lal Raja Singh R, Sandeep Krishnaa S, V Shreenikesh · 2025
In recent days, the issue of human-wildlife conflict, especially involving wild boars, has become a increasing concern in rural and forest-bordering agricultural regions. Traditional methods like fencing, manual guarding, or noise scare devices are not very effective always, and also requires human effort or it may harm the animals in some cases. Some systems using automation are developed, but they mostly lack in accurate detection or fast response time. In this project, we proposed a non-lethal, automatic wild animal detection and repulsion system based on real-time object detection using YOLOv5n model and high-frequency audio deterrent system. The system uses a Raspberry Pi, PIR sensor and camera to detect movement and identify wild boars using trained YOLOv5n model. Once detection is confirmed, the system triggers a high-frequency sound using amplifier and also sends alert to the farmer via GSM module and cloud. The model was trained on a custom dataset and shows decent accuracy with improving performance metrics during training. This system reduce human-animal conflicts and crop damages, and also helps to avoid harming the animals, so it is both effective and ethical.