Weapon Detection System For Security And Surveillance Using YOLOv8

The study focuses on developing and refining an advanced object detection model for integration into security systems. It begins with an exploration of challenges during early training epochs, aiming to enhance precision, recall, and overall performance. The training process involves meticulous data preprocessing, augmentation, and iterative optimization of hyperparameters to foster model convergence. Significantly, this project holds implications for security applications, offering an innovative approach to threat detection through advanced computer vision technologies. Notably, the study incorporates the usage of YOLOv3 and YOLOv8 models. The YOLOv8 model showed a mAP50 of 0.936 compared to YOLOv3, which had an mAP50 of 0.719, demonstrating superior performance with YOLOv8. Considerations for real-time processing, integration of temporal information, and collaboration with security professionals are vital for further advancements.