A Hybrid Quantum-Adversarial AI Framework for Defence Vision Security and Content Integrity

Abstract

Modern defense and high-security environments require surveillance systems that provide real-time monitoring, high detection accuracy, and strong data integrity. Traditional surveillance approaches rely heavily on manual observation and basic motion detection techniques, which are often inefficient and vulnerable to operational errors and security threats. This paper presents a Hybrid QuantumAdversarial AI Framework for Defense Vision Security and Content Integrity, designed to provide an intelligent, secure, and integrated surveillance solution. The proposed framework utilizes advanced deep learning techniques, specifically a YOLO-based object detection model, to perform real-time classification of individuals as Defence Personnel and NonDefence persons. To improve reliability and robustness, the system incorporates adversarial validation mechanisms including Structural Similarity Index (SSIM) and pixel-level processing for detecting manipulated inputs such as replay attacks and frame freezing. To ensure secure data handling, alert information is protected using AES-based encryption and SHA3 hashing techniques, while a quantum-inspired key generation approach enhances overall security strength. The framework also integrates real-time alert delivery through Telegram and Web Push notifications along with a centralized monitoring dashboard for managing alerts, logs, and system status. The modular architecture enables seamless communication between detection, validation, storage, and notification components and supports deployment on cloud platforms such as AWS EC2. Experimental evaluation demonstrates that the proposed framework provides an efficient, scalable, and secure surveillance solution by combining real-time AI detection, adversarial robustness, and protected data management into a unified system.