AUTOMATED DETECTION OF SCALLOPS AND FLOUNDER
Improving automated detection of scallops and flounder in optical surveys with stereo detection methods.
Principal investigators: Liese Siemann and Tasha O'Hara
Funding provided by: NOAA/NMFS Atlantic Sea Scallop Research Set-Aside (RSA) Grant Program
This goal of this project is to improve automated detection of scallops and flounder using the stereo image pairs taken during optical surveys. It contributes to a broader NOAA effort to support development of the Video and Image Analytics for a Marine Environment (VIAME), an open-source system for analysis of underwater imagery that is being developed by Kitware. Annotating images is the most time consuming part of generating reliable biomass estimates from optical surveys. To count live scallops or flounder, human or automated annotators have to differentiate between live vs dead scallops and identify camouflaged or partially buried flounder. To accurately measure animals that could be resting or swimming off bottom, determining their height off the seafloor is critical. Use of stereo imagery, commonly collected during many optical surveys, can address all of these issues and significantly improve automated detection