ABSTRACT
This research presents an AI-based weapon detection system designed for security purposes in armed forces, employing a fuzzy logic approach. Through 10 experimental trials, the system's performance was thoroughly evaluated, yielding notable results. Across these trials, the system consistently demonstrated high levels of accuracy, with values ranging from 90.7% to 93.7%. Precision values ranged from 87.2% to 92.0%, indicating the system's ability to minimize false alarms. Furthermore, the system exhibited strong recall values between 94.1% and 95.2%, signifying its capability to detect the majority of actual weapon instances. The achieved F1 scores, ranging from 90.6% to 93.5%, underscore the system's balanced performance in minimizing false positives and false negatives. These results highlight the system's effectiveness in accurately identifying weapons while minimizing errors, showcasing its potential for enhancing security protocols and operational efficiencies within armed forces and security agencies. Further validation and refinement in diverse real-world scenarios are recommended to ensure the system's optimal performance and applicability in critical security environments.
