Charles River Analytics was awarded a contract by the US Army to design a Criminal and Terrorist Indicator and Alert System (CRITICAL). CRITICAL is an enhanced surveillance system plug-in that will provide security personnel at high threat facilities with highly reliable automated alerts on suspicious activity.
Faced with constant threats from criminals and terrorists, domestic and international US Army installations have deployed numerous multi-sensor surveillance systems to alert security personnel of suspicious activity at or around bases. Unfortunately, legacy systems that use increasingly outdated rules-based detection schemes often generate a level of false alarms that drive alarm fatigue in security personnel. They’re also typically dependent on simple heuristics designed to detect obvious or active attacks; they can’t detect sophisticated external and internal threats with enough advance warning to intervene in time to stop an attack.
For example, a human observer watching activity outside the perimeter of an installation would typically identify a person incessantly pacing and casing the grounds as suspicious, but currently deployed systems are limited in their ability to flag this kind of anomaly without generating an overwhelming number of false alarms. These legacy systems deal with the diverse array of threats by either ignoring all but the most obvious threats, or by raising an alert for almost all activities, suspicious or not.
CRITICAL leverages computer vision and deep learning techniques along with pattern of life analysis algorithms to detect suspicious and anomalous objects, people, and behaviors with a low false alarm rate. CRITICAL distills the vast amount of visual data from surveillance into targeted, actionable alerts without requiring operators to be system experts. The system extends mature video analytics that provide plug-in functionality to major video management systems (VMS), facilitating integration with existing surveillance installations and operations.
“CRITICAL shows how quickly new capabilities can be developed thanks to recent advances in computer vision. By integrating mature object detection, tracking, and behavioral analysis algorithms, we can augment existing security and surveillance systems with an early warning capability that detects developing, complex threats that would usually be missed by traditional rule-based systems.” said Camille Monnier, Director of Ground Systems at Charles River Analytics and Principal Investigator on the CRITICAL effort.
CRITICAL will provide security personnel with reliable, automated cueing to suspicious activity, ultimately preventing attacks at Army bases. The innovations developed under CRITICAL support the Army’s Force Protection and Anti-Terrorism programs by creating an effective and accurate system for identifying potential threats.
