Predicting Success of Live Facial Recognition Operational Deployments

Is the use of Live Facial Recognition a good use of police resources? Mathematically, this can be posed as a conditional probability: given a wanted watchlist subject is at the location of the deployment, what is the probability that they will pass through the ‘zone of recognition’ of the cameras? To address the problem, probability and statistics were combined with an agent-based model to simulate the crowd dynamics conditional on the placement of the police van, the zone of recognition of the cameras, and the sign publicising the deployment. The study group discovered that, with additional time-dependent data of the detections made, the ensemble of models constructed have the potential to reliably advise the optimisation of success of future Live Facial Recognition deployments.