A novel approach to a hybrid security system using Operator Machine Augmentation Resource (OMAR)

Abstract

The recent increase in the threat level concerning the public has made government and security organizations look for better and more secure ways of monitoring crowds. These systems employ both vision and non-vision techniques in order to control crowds of people and avoid situations that might be difficult to predict or control. This paper looks at these methodologies and explores how they can be combined with artificial intelligence (AI) to improve surveillance. Another contribution of this work is the proposition of the Operator Machine Augmentation Resource (OMAR) framework that leverages state-of-the-art technologies such as computer vision and specific training for CCTV operators to overcome the weaknesses that characterize legacy surveillance systems. The OMAR framework is designed to provide the operators with better tools for improving productivity and, thereby, surveillance system efficiency and security. The research also looks at the outcome of the hybrid security systems based on the level of training of the personnel involved and the level of AI support. The study does this by developing a human security information model and then using it to forecast the potential performance of these systems using elements such as response time, accuracy, cognitive load, visual discrimination, trust, and confidence. Subjects were split into trained and non-trained groups and were assigned surveillance tasks with and without AI. The analysis of the results of the trained models, such as Linear Regression and Random Forest, showed that training enhances performance through error reduction and accuracy increase and that AI enhances speed and efficiency but sometimes may increase the errors, especially in the non-trained groups. AI support also helped decrease the cognitive workload in the trained participants and increased the trust and confidence of non-trained participants. The study's results also identify the relationships between the operator, AI support, and training in hybrid surveillance systems. These findings underscore the necessity of creating measures to foster the trust of operators in AI-based systems and improve the cooperation between the operators and AI. The study indicates that future studies should test the OMAR framework and possibly expand these models to examine their usefulness in contributing to public safety and security.