Enhanced Horse Stable Security Monitoring Using Deep Learning: Investigating YOLO Techniques and Architecture

The aim for this paper is to develop an enhanced horse stable security monitoring using deep learning. The YOLO techniques are developed using normalized standard initialization and trained with a dataset of sample photos from horse stables, encompassing both humans and horses. The trained weights are validated using a test video from a horse stable at Tanjung Lumpur. The implementation of the findings is carried out using the Python language. Consequently, this study investigates the YOLO techniques and their architecture, analyzing the best approaches for the proposed system. YOLOv3 and YOLOv4 specifically were used. The performance analysis identifies YOLOv4 with a threshold value of 0.7 and a larger dataset as the most effective system to implement. Overall, this research delves into the investigation of YOLO techniques and their architecture, contributing to the improvement of security monitoring in horse stables. By employing deep learning and advanced object detection methodologies, the performance and reliability of security monitoring systems in equestrian environments can be enhanced.