ML-Enabled Internet-of-Things Edge Device for Vehicle and Weapon Detection from Live Video Surveillance

The U.S. Army has identified a need to enhance force protection and intelligence operations by automating the detection and classification of vehicles and weapons from live video sources and correlating these objects with specific individuals.  To address these needs, Tygart Technology, Inc. will develop, enhance and commercialize an Internet-of-things (IoT) edge-device capable of monitoring live video surveillance feeds using machine learning (ML) algorithms which detect people, vehicles, weapons and faces.  The edge-device will be coupled with and populate an enterprise cloud-based repository of observed objects-of-interest which will also include a watch list management and face recognition matching capability.  In Phase I, Tygart will demonstrate the feasibility of the system by specifying the overall system design; identifying and evaluating existing ML models and training data sets; and conducting initial prototyping of the IoT edge-device and cloud object repository and face matching capability.  In Phase II, Tygart will enhance/retrain ML vehicle and weapon detection models utilizing Tygart’s Machine Learning Model Pipeline Platform.  The models will be fully integrated into the edge-device to “vehicle, weapon and persons-of-interest enable” a large array of live video surveillance cameras.  A successful program will transition to support PM Biometrics programs of record.