YOLO-Based Wild Animal Detection for Wildlife Monitoring and Conflict Mitigation

The encroachment of wildlife into human habitats like villages poses a serious threat to both human beings and animals. This paper presents a thorough analysis of the use of the YOLO (You Only Look Once) algorithm for the real-time detection of animals, with particular emphasis on its application towards wildlife conservation and mitigation of human-animal conflict. The system, based on YOLOv5, provides excellent precision of detection, striking a great balance between speed and precision. In real-time, the video feed is used to inspect and to alert local authorities on animal detection. Thus, the system acts as a deterrent mechanism. Comparison against YOLOv4 and other models shows great efficiencies of YOLOv5 in terms of real-life scenario performances. The research also highlights challenges such as small object detection and adverse environmental conditions, while some suggestions concerning future development based on these challenges are also offered. Index Terms-YOLOv5, animal detection, wildlife monitoring, human-animal conflict, deep learning.