In recent years advances in machine learning methods such as deep learning has led to signi cant improvements inour ability to track people and vehicles, and to recognise speci c individuals. Such technology has enormous po-tential to enhance the performance of image-based security systems. However, wide-spread use of such technologyhas important legal and ethical implications, not least for individuals right to privacy. In this paper, we describea technological approach to balance the two competing goals of system e cacy and privacy. We describe amethodology for constructing a goal-function" that reects the operators preferences for detection performanceand anonymity. This goal function is combined with an image-processing system that provides tracking andthreat assessment functionality and a decision-making framework that assesses the potential value gained byproviding the operator with de-anonymized images. The framework provides a probabilistic approach combininguser preferences, world state model, possible user actions and threat mitigation e ectiveness, and suggests theuser action with the largest estimated utility. We show results of operating the system in a perimeter-protectionscenario.
展开▼
