Automatic detection of scallops in seafloor images

Nova Scotia based Clearwater is one of the largest North American seafood suppliers. The company hasdeveloped infrastructure enabling them to conduct large scale photographic and video surveys of the oceanfloor. The survey results are used to develop harvesting strategies based on biological and size characteristicsof populations. The image data are currently evaluated by human experts trained in the identification ofunderwater species, which is both slow and expensive. Clearwater does not have the in-house capability todevelop algorithms and techniques for processing the image data automatically.We will develop techniques for automatically identifying and measuring scallops in seafloor images, and fordistingushing them from other animate or inanimate objects. Image processing techniques will be used toidentify candidate regions in images, and machine learning algorithms will be employed to distinguish scallopsfrom other objects. Our system will enable Clearwater to develop scallop harvesting strategies in a more timelymanner and at a reduced cost, both for catch purposes and for resource nurturing and protection. Being able togenerate data at a higher resolution will improve stock assessment processes and contribute to sustainableharvesting practices.