ReelData, the Halifax maker of AI for aquaculture operations, has launched its first physical product: an underwater camera meant to integrate with aquaculture operators’ computer systems.

The camera’s bandwidth and compression abilities are tailored to work well with AI systems, and the device itself is waterproof and sturdy enough for use in a range of difficult environments, CEO Mathew Zimola said in an interview.

The ReelVision, as Zimola’s team has dubbed their new camera, will complement the company’s three software products — ReelAppetite, which calculates how much to feed fish, ReelBiomass and ReelWeight.

"(ReelVision) is an underwater, rough-environments and AI-focused camera system,” said Zimola. “It’s a standard IP (internet protocol) camera. So if you go to a gas station and see the cameras that are videotaping … those are IP cameras. They plug into a network switch, and that’s how they’re powered.”

When it was founded in 2018, ReelData’s original business plan was to sell only AI software, integrating camera hardware from a third party. But Zimola said a lack of viable options eventually forced ReelData to begin developing its own about two years ago. For example, he recalled with a chuckle, several underwater cameras ReelData tested had leaks, and at least one did not properly compress its data.

The ReelVision camera can also send data directly to a GPU, or graphics processing unit. Many of ReelData’s clients are located in remote regions without reliable web access, so the company’s software runs locally on graphics cards, rather than in the cloud.

Several of the company’s existing clients have been trialing ReelVision cameras since last summer, with the commercial release now relying on a hardware-as-a-service model.

“As we’re bringing new products to market, we try to really work with a research and development mindset,” said Zimola of his focus on thoroughly testing ReelData’s technology before it sees wide release. “In the technology landscape, some people have a ‘move fast and break things’ mindset. We don’t really have that luxury because we’re taking care of animals. So we really, really make sure that when we release a product, it works.”

ReelData’s largest market is Norway, a longstanding global leader in aquaculture. The land-based aquaculture businesses that Zimola’s 32-person team sells to include facilities that raise fish on land only when they first hatch, before placing them into pens, as well as a newer generation of land-based facilities that raise fish in tanks until they are ready to be harvested for sale.

Customer discovery and business development work have also helped ReelData identify an additional market of aquaculture operations that are not yet using AI, but want to upgrade their equipment ahead of time in preparation. Those customers will have the option to buy the cameras outright, rather than under a service model.

“We’re seeing that there is an appetite from people that want to prepare themselves for artificial intelligence and buy equipment, and down the line, adopt those AI models,” said Zimola. “If there is a (fish) farmer or somebody in the ocean industry now that is looking at needing cameras for their vessel or their facility, and they know that in a couple years they’re going to need to do some sort of AI applications, they can buy ours right now.”