Automated Intruder Detection and Alert System using CNN-based Facial Recognition for High-Security Rooms

Aim: The study's objective is to create a CNN-based facial recognition system for high-security rooms that uses automated intruder detection and alerting. The technology is made to identify unwanted access and send out email and audio alerts when it detects it. Materials and Methods: Two groups participated in this study. Group 2 refers to the current Front Door Security (FDS) algorithm with 20 samples, whereas Group 1 refers to the suggested Convolutional Neural Network (CNN)-based model with 15 samples. The threshold is 0.05% with a 85% confidence range, and the G Power value is 70%. PyCharm is the development environment used for this project. Result: Accuracy of the FDS algorithm is between 50.45% and 60.00%, and the CNN-based model's range is between 70.50% and 85.30%. The CNN-based model achieves an optimum accuracy of 85.76% and a significance level of around 0.0048. Conclusion: The CNN-based model performed better compared to the FDS algorithm with respect to the accuracy. The CNN-based intrusion detection model significantly outperforms the current FDS approach in terms of accuracy. For high-security rooms, the proposed solution provides reliable and efficient monitoring.