A surveillance system to provide early detection of attacks targeting army installations

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.

CRITICAL combines computer vision, deep learning techniques, and pattern of life analysis algorithms to detect anomalous objects, people, and behaviors.

“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.”

Camille Monnier,
Director of Ground Systems at Charles River Analytics and Principal Investigator on the CRITICAL effort
Camille Monnier,
Director of Ground Systems at Charles River Analytics and Principal Investigator on the CRITICAL effort

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 Unfortunately, legacy systems that use increasingly outdated rules-based detection schemes often generate a level of false alarms that drive alarm fatigue in security personnel.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 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.

Contact us to learn more about CRITICAL and our other artificial intelligence, object detection, and pattern and gesture recognition capabilities.

This material is based upon work supported by the USA-ACC-APG under Contract No. W56KGU-20-C-0073. All material, except scientific articles or papers published in scientific journals, must, in addition to any notices or disclaimers by the Contractor, also contain the following disclaimer: Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the USA-ACC-APG.

Our passion for science and engineering drives us to find impactful, actionable solutions.