NA23OAR0210570

Oceanit proposes the design, construction, and testing of a hyperspectral imaging equipped Unmanned Aerial System (UAS) for the detection of various coral species and the monitoring of coral health. This system will be the prototype for the sensor part of a robust, COTS-based UAS system to aid the NOAA National Coral Reef Monitoring Program. Different coral families and species will be identified by close inspection of spectral responses combined with advanced machine learning algorithms.[1] Once a system is built to collect hyperspectral aerial imagery of corals, Oceanit will apply computer vision and machine learning techniques to identify coral families and species based on this data. There have been significant advances in differentiation of bottom types, coral families, and coral species over the years.[2][3] Bottom type classification is well-established, and coral species identification has been performed for several coral species around the world. Oceanit’s CORAL SHIELD will give NOAA and ocean researchers new capabilities to build maps of coral reefs and track populations of endangered and threatened coral species in a more timely and cost-effective manner. Our proposed system could also be used to find rare coral types, to determine overall health states of coral, and to enable rapid response to disease outbreaks. [4] CORAL SHIELD will decrease the requirement for manual coral counting and identification using divers, enabling researchers to collect data quickly and accurately. Initial prototypes will be built around a rotary-wing UAS to provide a low cost and ease of use solution. Later systems could be built around a fixed-wing UAS to enable imagery of corals at longer range or under autonomous control.