Real-Time Wildlife Tracking with Fine-Tuned YOLOv11

Wildlife monitoring is an important task in computer vision to preserve biodiversity, maintain ecosystem balance and stability, and promote sustainable development. Monitoring wild animals is important for people living near forests, as conflicts between humans and wildlife often result in the loss of precious human lives and livestock. Low-Light images, diverse animal poses, complex environments, and varied terrain structures present significant challenges in wildlife monitoring. This paper introduces a novel wildlife monitoring technique utilizing a fine-tuned YOLOv11 model to effectively monitor animals in low lighting conditions, challenging environments and diverse poses. The dataset is constructed from CCTV video footage captured at night and includes frames featuring 17 wild animal species. The mAP of our fine-tuned YOLOv11 is 97.2%.