Minimum Entropy, LLC — Department of Defense SBIR Phase I: A18-127

Description

Modern force protection systems lack the ability to identify persons of interest through automated video and imagery analytics. In many force protection scenarios, especially at long distances and even with high quality sensors and optics, the number of pixels on a face are limited and thus is necessary to develop a purpose built low resolution face recognition model. Under this proposal, Minimum Entropy will partner with Progeny Systems and proposes to perform sensor profiling and CONOPS analysis to determine performance limitations and likely operational deployment goals, perform an ISA feasibility and compatibility study, and gather and label requisite data for training and evaluation. We will leverage existing state of the art face detection algorithms in order to concentrate development on a low resolution face recognition model. Lastly we will work with our partner to integrate person re-identification when distances are too far, and resolution too low to perform recognition. We believe the approach, knowledge, and partnership detailed in this proposal will lead to a successful approach that minimizes risk, while also leverages the latest research from academia in order to deliver an accurate and fast low resolution face recognition algorithm for the next generation of force protection systems.