Army Research Laboratory (ARL) has developed a large set of thermal face data to support facial recognition in dark environments, FedScoop reported Monday. ARL collaborated with Booz Allen Hamilton, West Virginia University, the University of Nebraska-Lincoln and Johns Hopkins University to build the ARL Visible-Thermal Face Dataset.
The team used a long wave infrared camera to capture over 500,000 thermal images of 395 faces. Thermal data on a subject's head pose, expressions and eyeglasses serve as references to reproduce physical conditions.
The Army expects the dataset to allow for facial recognition in scenarios with no sunlight, through the help of artificial intelligence. The service branch seeks to further develop thermal identification despite the lower quality of thermal images compared to traditional ones.