Maritime Activity Recognition through Video Analytics and Human Adjudicated Learning

To address the Department of Homeland Security (DHS) need for automated video analysis capabilities that monitor shore-based video cameras and alert operators on maritime activities of interest and anomalous activities, Intellisense Systems, Inc. (Intellisense) proposes to develop a new Maritime Activity Recognition through Video Analytics and Human Adjudicated Learning (MARVAL) system. The innovations in using a new microservice-based video analytics framework and natural language prompting will enable the MARVAL to automatically analyze visible and infrared videos sent from shore-based cameras, detect and track vessels and people, perform activity classification based on prompted activity description, and automatically alert the operator with flags and reports for adjudication. As a result, this system offers automated video analytics and human-machine teaming capability to help the operator maintain awareness. In Phase I, Intellisense will demonstrate the feasibility of MARVAL by developing its system design, identifying software and hardware requirements, developing appropriate benchmark data, and developing performance measurement metrics for the system. An initial Phase I prototype system will be integrated to evaluate and demonstrate MARVAL’s video analytics algorithms. In Phase II, Intellisense will mature the system design and develop a prototype that incorporates human-machine teaming and automates video analysis for operational testing. The successful completion of MARVAL’s automated video analytics and human-machine teaming capability at the end of Phase III will benefit the nation in both government and commercial sectors. Commercial applications of this technology include video intelligence, security surveillance, video reconnaissance, and perimeter security.