Key performance metrics

Summary

This section outlines important metrics for assessing the performance of live facial recognition (LFR) systems, specifically focusing on the True Recognition Rate (TRR), False Alert Rate (FAR), and Recognition Time (RT). The TRR is a measure of how many individuals on a watchlist pass through the zone of recognition correctly identified by the LFR system. The FAR reflects the rate of false alerts generated by individuals not on the watchlist. Both TRR and FAR should be monitored and reported to the Senior Responsible Officer (SRO) to ensure efficient operational performance. Finally, the RT measures the time taken for the system to generate an alert after detecting a watchlisted subject, emphasizing the need for timely responses to alerts.

Key principles

• 1Monitor key metrics such as TRR, FAR, and RT to assess LFR system performance.
• 2The TRR helps determine the accuracy of LFR by comparing alerts to known identities.
• 3The FAR should be maintained at 0.1% or less to minimize false alerts.
• 4Recognition Time (RT) should be fast enough for effective response before the subject moves out of range.

Defined terms

True recognition rate (TRR)

The number of times individuals on a watchlist are correctly identified as a proportion of the total number of times these individuals passed through the zone of recognition.

False alert rate (FAR)

The number of individuals not on the watchlist who generate a false alert as a proportion of the total number of people who pass through the zone of recognition.

Recognition time (RT)

The average time taken between a subject on the watchlist passing before a camera and the generation of an alert.