Using AI deep learning, Davidson-developed Apollo Vision automates answering essential elements of information on priority intel requests by parsing analyst request and scanning geospatial imagery with Automatic Target Recognition (ATR). Apollo Vision is a concept designed to assist geospatial analysts in automating the process of answering Essential Elements of Information (EEI’s). EEI’s are a single question asked of a geospatial analyst, many of which can be answered by scanning imagery directly. For example, an EEI may be: how many U2 are currently at Beale AFB?
Apollo Vision leverages a deep learning model (specifically YOLOv5) to perform Automatic Target Recognition (ATR) to identify aircraft type in imagery. Apollo Vision has additional features, such as a chatbot interface with the analyst, AI/ML trust built in directly, and Natural Language Processing capabilities. However, these capabilities have not been developed yet.
