CARPOOL: Continual leArning foR PeriscOpe videO anaLysis

In Phase II, the machine learning (ML) model update strategies that were developed as part of Phase I efforts will be refined for different use-case scenarios. This includes developing and validating mathematical models on both the standard image/video datasets and the provided representative periscope imagery data. A system prototype will then be developed to enable timely vessel detection, identification, and re-acquisition in periscope imagery. The prototype will then be validated through testing to demonstrate improvements relative to individual performance metrics as well as mission performance metrics. Finally, a Phase III plan will be developed in transitioning the developed technology into relevant platform hardware.