Washington – U.S. intelligence researchers are developing software algorithms to identify people and vehicles in videos from various cameras, locate those people and vehicles, and analyze the data for anomalies and threats. We are consulting the industry.
Officials at the U.S. Intelligence Advanced Research Projects Agency (IARPA) in Washington made a wide-ranging agency announcement on Tuesday about the Video Link and Information from Non-Cooperative Sensors (Video LINCS) program.
The program uses artificial intelligence (AI) and machine learning to first identify and locate people, then to identify and locate vehicles, and finally to identify common objects across the video collection. Identify and locate. IARPA is the research arm of the U.S. Director of National Intelligence.
In today's world, video is a ubiquitous sensor, IARPA researchers explain. Vast amounts of video data are being collected, but human analysts are too few and too slow to identify threats. Most human video analysis is done in response to tragic events.
Related: Artificial Intelligence and Machine Learning for Unmanned Vehicles
Moreover, different video sensors rarely communicate with each other, even within a facility. Ultimately, human analysts are overburdened and require significant expertise, memory, and resources to extract threats from mountains of video data.
The Video LINCS program consists of two technical areas: identifying people, vehicles and common objects across many video streams; and finding these objects.
The goal of Video LINCS is to re-identify objects across any video collection without prior access to representative videos or the objects to be identified.
The problem revolves around videos coming from different sources with different characteristics that are difficult to correlate.
Related: Technologies that enable image and video processing
Video LINCS attempts to use videos as input and generated identified object locations as output. Geolocalization provides object tracking, which requires a mapping from camera coordinates to a common reference frame.
Video The LINCS program is planned as a four-year effort divided into three phases. The first 18-month phase involves identifying and locating people in the video collection. His second phase, which lasts 18 months, involves identifying and locating changes in clothing, vehicles, and common objects. His third, year-long phase focuses on common objects. Bidders are required to submit proposals in all phases.
Bidders are interested in Artificial Intelligence (AI), Object Detection, Tracking, Computer Vision with Modeling, Deep Learning, Geometric Camera Projection and Backprojection, Image and Video Geolocation, Machine Learning, Modeling and Simulation, Open Must have expertise in set classification, soft biometrics, software engineering, software integration, systems integration, vehicle fingerprinting, and video data generation.
Related: The role of technology in securing our borders
Interested companies should submit proposals online to the IARPA Distribution and Assessment System (IDEAS) by July 15, 2024 at https://iarpa-ideas.gov/Client/signin.aspx .
If you have any questions or concerns, please email IARPA. [email protected]For more information, visit https://sam.gov/opp/e951e8e81d7442d6816e14b55dbbaf29/view and https://www.iarpa.gov/research-programs/video-lincs.
