Can a mussel-based sensor warn farmers of water quality concerns? (via. TheFishSite)
IntegraSEE’s CEO, Lawrence Taylor, explains how monitoring the behaviour of a handful of mussels in a MarineCanary unit can alert aquaculture operators to water quality issues, including pollution, pathogens and temperature anomalies.
Can you briefly describe your aquaculture career?
Right after finishing my masters’ degree in the early 90s, I jumped into a three-year sea scallop bottom culture project at Dalhousie University, Halifax. The Ocean Production Enhancement Network (OPEN) funded program focused on basic oceanographic and biological research, targeting sea scallops and cod. With a background in photography and SCUBA diving experience, I was hired to use an underwater video camera to record the ocean bottom along transect lines – capturing our seeded scallops, wild scallops and predators. Diving four to five hours a day in the fall and winter was mind-numbing, but a lot more fun than the follow-up: five days of analysing video footage in the lab…manually. And manually gluing two bee-tags on about a third of our 10,000 juvenile seed scallops ranked right up there with the video analysis.
As video production technology was rapidly pushing towards a completely digital workflow, I remained in the field specialising in underwater filming and post-production projects that focused on seafood and aquaculture research. And my mission has been to see how far we can press automated vision technology into a multi-tasking marine research tool.
What is MarineCanary and where did the idea come from?
Like a canary in a coal mine, the MarineCanary (MC) is an early warning system designed to detect waterborne issues like hazardous algal blooms, pollutants, high water temperatures and hypoxic conditions, before they hurt marine life. Each unit is created as a stand-alone, drop-and-go system that is suspended in the water column between its float and anchor mooring system. Within the MC unit lives a set of mussels. When a significant number of them close their shells, a warning is sent to the user’s mobile device.
The original MC unit was designed almost 25 years ago by IntegraSEE co-founder, Ulrich Lobsiger, and was based on recording and analysing live mussel behaviour using timelapse film technology, but – as machine learning, deep learning and AI weren’t readily available – there wasn’t a lot of incentive to commercialise a product that required manual image processing.
That all changed last July when Karan Kharecha, our computer science team member, and I, deployed our first prototype in Halifax Harbour on COVE Ocean’s Stella Maris platform. The prototype is composed of a metal frame with nine mussels secured with popsicle sticks on one side, and an underwater enclosure containing a fully digital video recording system on the other. At this time, data and power are transferred via Stella Maris’ umbilical to shore, but in future, the MC system will be completely self-reliant to capture and process images. The system only records during daylight hours, but Ocean Tech students from the local community college, NSCC, are developing an infrared lighting array that will be triggered by a photocell to turn on. We’re also investigating the use of UV light to prevent fouling on the camera end of the enclosure.
Read the full article here