Aquabyte — the algorithms-for-aquaculture start-up started in San Francisco — has said applying its machine-learning and computer vision to nail down feed times could save growers “20 to 30 percent” in feed costs, Forbes has reported.
Yesterday, SalmonBusiness wrote that the now Bergen-based company had raised USD 3.5 million in seed capital with the help of venture capital funds. Aquabyte said then, that sea lice counting; real-time measurements of biomass and individual fish were possible using its computer vision algorithms, now understood to be “in the making”.
“Fish are exothermic and respond to their environment which means that a lot of data — from both camera data using computer vision, multi-sensory environmental data like temperature and oxygen, and human input data such as how much to feed — make this a very rich data problem for machine learning,” Aquabyte CEO Bryton Shang told Forbes.
The data to be crunched is gathered via underwater 3D cameras in marine pens or land-based tanks. This “appetite detection” and “feed optimization” are stated goals on the company’s Web page.
Read Sonar device set to feed fish robotically
Read EWOS result marked by feed costs, brand sale
“We’re in the middle of building those algorithms (for feed calculations),” Shang was quoted as saying, adding, “Then we’ll commercialize those and use the data to develop feeding algorithms.”
With feed from one-third to two-thirds the investment in a farmed fish, there’s money to be saved should the promise of Aquabyte’s algorithms be realized.