Automated Target Recognition Computer Vision Model for Aerial ISR and GNC Sensors with Lexset Synthetic Data

Lexset proposes the development of an Automated Target Recognition (ATR) computer vision (CV) model that is capable of autonomously identifying adversary systems in electro-optical sensor data with targeted Mean Average Precision scores >90%. The Lexset ATR CV model will be trained on Lexset Seahaven-generated synthetic data, which has been proven to bolster CV model performance substantially in Lexset and third-party tests. Through this Phase I and its follow-on efforts, Lexset’s bespoke ATR model will be primed for integration on USAF aerial sensors that conduct Intelligence, Surveillance, and Reconnaissance (ISR) and Guidance, Navigation, and Control (GNC) at the edge, catalyzing a leap-forward advance in autonomous sensing and characterization of adversary systems by these sensors with the speed, precision, and accuracy demanded by USAF pilots and intelligence analysts.