Implementation of Deep Learning in Automatic Enemy Object Shooting to Assist in Securing Military Guard Posts

Abstract

Military guard posts play a crucial role in securing strategic areas, particularly in high-risk regions that frequently face armed threats. However, limited infrastructure and equipment often reduce security effectiveness. This study aims to develop an automated security system using deep learning technology for enemy detection and automatic shooting responses. The system employs Faster R-CNN algorithm for object recognition and classification, integrated with a Logitech C920 camera for image capture, Raspberry Pi 4 Model B for data processing, and Arduino Nano to control servo motors for weapon movement and firing. Testing results demonstrate that the system effectively distinguishes between friend and enemy objects based on clothing attributes under various lighting conditions. The servo motor achieves 87.25 percent accuracy for horizontal movement, 88.89 percent for vertical movement, and 88.87 percent shooting accuracy with an average response time of 1.011 seconds. Detection probability reaches 85.11 percent under bright conditions and decreases to 65.24 percent in dark environments. The system successfully detects enemies up to 36 meters distance. This automated system enhances guard post security and reduces risks for military personnel.