Animal Detection in Wildlife Conservation Using Deep Learning

Animal detection is one of the wildlife conservation techniques that can help with the issue of rapid decline in global wildlife population. The main goal is to create an extremely precise and effective object detection system that can help with species monitoring, anti-poaching operations, conflict reduction between people and wildlife conservation planning. In order to eliminate the need for region proposal creation which is required in usual CNN-based techniques, a single stage object detection technique is utilized in the proposed work. The YOLO algorithm has been implemented in this paper for detecting accurately and identifying the class of the animal in a changing weather condition. The bounding boxes are clearly annotated to form the ground truth labels for object detection model training. Performance comparisons are made with the existing system in the field of wildlife conservation and the mAP of 93.8% was achieved. From the obtained results the proposed method seeks to support evidence-based decision-making and effective conservation policies by enabling precise and efficient identification and monitoring of animal population, thereby promoting the cohabitation of humans and wildlife in a sustainable manner.