Deep Learning Analytics, Llc — Department of Defense SBIR Phase II: SB171-002

Description

US DoD personnel make direct contact with unfamiliar and potentially harmful plants, insects, arachnids, and reptiles (i.e. biothreats), impairing personnel readiness and mission effectiveness. These threats endanger personnel, disrupt operations and destabilize ecosystems around the world, often requiring expensive remediation. Currently the DoD employs specialized field and medical staff and implements comprehensive training programs to manage these threats. This expensive approach relies heavily on availability of personnel with scarce expertise. In Phase I, Deep Learning Analytics demonstrated a BioThreatID application capable of identifying thousands of species of plants, insects, arachnids, and reptiles from photographs on a handheld mobile device without a network connection. Deep Learning Analytics also separately demonstrated world class species ID performance on the largest international benchmark of its kind. In Phase II, Deep Learning Analytics will differentiate this BioThreatID application with (1) new deep learning-based technologies not available in any other competing mobile application (2) robustness to degraded image quality and (3) ruggedization. This application will enable a handheld mobile device to visually identify plants, insects, arachnids, and reptiles in the field in diverse global ecosystems, giving DoD personnel operating in harsh environments a low cost, practical, accessible means of identifying these threats.