We have developed a pair of metrics for the quantitative evaluation of the performance of pedestrian detection systems. The Metric of Similarity was designed to be used to assess how well the pedestrian-detection output of an infra-red Night Vision system matches its ground truth, that is, the relative level of fit or agreement between the locations in an image frame (measured in pixels) where the system indicates it has detected pedestrians and the locations in the frame where there actually are pedestrians. In contrast, the Metric of Salience was designed to be used to infer the level of acceptance of the system by a typical driver. These are complementary dimensions of system performance.