Physical Optics Corporation — Department of Defense SBIR Phase I: SOCOM163-003

Description

To address the USSOCOM need for man-portable advanced tactical facial recognition technologies at ranges of 650 m to 1 km, Physical Optics Corporation (POC) proposes to develop a new Facial Recognition at Long Distance (FAREL) system based on an innovative combination of technologies that include lucky imaging and a customized deep convolutional neural network (CNN) method. Specifically, the innovation in deep CNN and lucky imaging will enable the FAREL system to reliably recognize the face of a target at a distance of 1 km. This technology offers high probability of detection and real-time operation in a low size, weight, and power (SWaP) processing hardware, which directly address the USSOCOM requirements. During Phase I, we will conduct a thorough technology feasibility study, including an analysis of the critical technical factors, constraints, and relationships affecting FARELs long-range facial recognition performance. We will analyze the requirements to achieve a man-portable SWaP form factor. During Phase II, we will develop a detailed FAREL design and conduct a scalable prototype demonstration that will include a deep CNN method and lucky imaging processing. We will also develop a path to ruggedized man-portable FAREL operation.