Background: Optical imagery increasingly serves as the primary data source for many existing NOAA surveys (e.g. Fishery-independent surveys and electronic monitoring) which directly support Natural Resource Management, Sustainable Fisheries, Ecosystem-Based Fishery Management, the New Blue Economy, and a Climate Ready Nation. Optimized optical technology is critical for efforts to move away from traditional ship-based and visual aerial survey methods and towards uncrewed systems and platform agnostic data collections. Recent efforts such as the Advanced Sampling Technology Working Group (ASTWG) catalyzed significant advancements in optical and acoustic sampling platforms while the Automated Image Analysis (AIASI) and the Untrawlable Habitat (UHSI) strategic initiatives made significant progress addressing foundational aspects of using optics and which this Optical SI will build upon. For instance, several developments from the earlier SI’s have become valuable tools for NOAA analysts such as ML/DL based analysis and performance evaluation (e.g. VIAME and CoralNET). While useful in their current state, these tools would benefit from modernization, improved user interfaces, cloud deployment, and continued integration of the constantly evolving set of tools and algorithms developed by the computer science industry. Automation will be critical as the availability of days at sea (DAS) aboard NOAA Fishery Survey Vessels (FSV) is decreasing and recent initiatives to begin collecting data in otherwise inaccessible locations (e.g. wind lease areas) heighten the importance of a transition to platform-agnostic data collections. With increasing application of optical data collection platforms and scalable automated processing solutions, the volume and utility of optical data sets is increasing exponentially. To address this issue several science centers have created transition plans to move automated processes into operations. Realization of end-to-end automation of optical data processing is contingent upon components highlighted in this vision document to be resolved, in particular data management and processing speed. Significant investment is needed to make them efficient, accessible, and relevant for resource management (e.g. stock assessment).
0 commit comments