ARETE ASSOCIATES — Department of Defense SBIR Phase I: In this proposal, Arete Associates describes a novel approach to dismount detection, track

Description

In this proposal, Arete Associates describes a novel approach to dismount detection, tracking and classification in ground-based infrared image sequences. The proposed approach leverages state-of-the-art hierarchical machine learning algorithms (HMLA) for autonomous target detection and activity threat level classification and proven Bayesian peak tracking (BPT) methods for robust tracking through time. The proposal describes our understanding of the problem, proposed solution and Phase I SBIR work tasks in context of the larger problem and Arete related work. A proof-of-concept test is described in which a simple HMLA algorithm implementation demonstrated successful classification of running versus walking subjects in test sequences of multi-frame infrared data.