More than 50% of the seafood consumed in the world is grown commercially, so boosting productivity and lowering the costs of fish farms is a big deal for the industry. Startup Aquabyte Inc. aims to increase efficiency in this business using machine learning and computer vision software.

The software, said Bryton Shang (pictured), founder and chief executive officer of Aquabyte, is “based on a camera that takes pictures of a fish in a fish pen, analyzes those images, and helps the farmer understand the health of the fish, the weight of the fish, how much to feed, and generally better manage their farms.”

Shang spoke with Jeff Frick, host of theCUBE, SiliconANGLE Media’s livestreaming studio, for a digital CUBE Conversation. They discussed the use of machine learning in aquaculture, Aquabyte’s experience with salmon in Norway and its plans for the future.

Single camera for wide range of applications

Built in Silicon Valley, Aquabyte is first being deployed in Norway, which produces about half of the world’s farmed salmon. In 2017, Shang went to a conference in Norway with a prototype for sizing fish with a camera and left the event with ideas for a much wider application.

“In Norway, they have recently passed regulations around counting sea lice on the fish … and farmers asked me then: ‘If you could use the camera to size fish, can you also count sea lice? And can you also detect the appetite?’” Shang said. “And then it just turned into this more platform approach, where this single camera could do a wide variety of application.”

Although fish farming is already highly mechanized in Norway, with automatic feeders, for example, there is not much machine learning and computer vision in the industry. But farmers have been quite open to experimenting with technology, especially when it helps to save labor on the farm, according to Shang.

The system developed by Aquabyte involves an installation in the cage, consisting of a camera and equipment package and a cloud-based machine-learning solution that performs image analysis.

The largest salmon farm in Norway has 2 million fish in a single pen. “We’ll eventually get to the point where we can identify every single fish in the pen and use that to track individual health and growth,” he explained. “We use the individual recognition algorithm, [and] the deal is to de-duplicate fish.”

After starting with Norwegian salmon, Aquabyte wants to take the technology to other countries and other species of fish. The company has already started to expand its applications to a second species, rainbow trout.

“Eventually, we’ll get to the point where that data allows us to run fully autonomous fish farms,” Shang said.

Here’s the complete video interview, one of many CUBE Conversations from SiliconANGLE and theCUBE:

Photo: SiliconANGLE