Expert evaluation of AIAEF framework in AI-RHCA systems for real-time and historical

Abstract

Through the analysis of CCTV data, artificial intelligence (AI) considerably improves the efficiency of surveillance and data management systems in smart cities. This is accomplished through enhanced data management. Despite this, these systems continue to face challenges regarding the accuracy of their detection, the capacity of AI models to learn, and the safety of their data. This research establishes an AI Framework for Real-Time and Historical CCTV Analytics (AI-RHCA) aimed at effectively processing both real-time and historical data, utilizing Explainable AI (XAI), Deep Learning (YOLO, Faster R-CNN), and Edge Computing technologies to improve adaptability and minimize data processing demands. The efficacy of the artificial intelligence-RHCA system was assessed using the Artificial Intelligence Assessment and Evaluation Framework (AIAEF). Nine fundamental features define this paradigm: accuracy, dependability, security, interpretability of results, and so on. The assessment outcomes from 30 experts indicated that AI-RHCA had a significant degree of appropriateness (X̅ = 4.45), with the Model Selection and Model Training modules receiving the highest ratings. The system is capable of adhering to international standards, including GDPR, ISO/IEC 27001, and AI Ethics, while also facilitating applications in the industrial sector and smart cities securely and effectively.