Aquabyte helps salmon farmers plan a harvest from underwater images

After Mowi Ireland removed larger salmon from one of its pens, an underwater monitoring system registered a fall in the average weight of the fish remaining. For the farm team, that was useful confirmation that the partial harvest had selected the intended fish.[1]

The observation comes from Aquabyte’s account of a trial with Mowi and Ireland’s seafood-development agency, Bord Iascaigh Mhara. Two systems were installed at a south-western Irish site in autumn 2025. It is operator testimony about a working trial, without a controlled estimate of yield or welfare improvement.[1]

A picture becomes a measurement

Aquabyte’s Hammerhead camera captures images inside the pen, selects useful frames and sends them for cloud analysis. The company describes more than a million images per day, supporting repeated weight estimates and observations of lice and welfare. Its self-cleaning lenses and sensors are designed to keep that view usable while the unit stays submerged.[3]

The weight is estimated from images. Aquabyte’s WEIGHT module presents the average, its change over time and the spread across weight bands. Staff can see whether a pen is growing evenly or contains groups ready for different decisions.[2]

Why the spread matters

A pen’s average can conceal a mixture of large and small fish. Aquabyte’s weight bands let a farmer examine that mixture before choosing a harvest. In the Irish trial, staff said the distribution helped them choose panels for separating fish by size.[2][1]

That is a useful sequence of decisions: inspect the population, select a group and examine the population left behind. Repeated measurements connect planning with the result, instead of treating the harvest as the end of the information trail.

Counting without lifting every sampled fish

For lice monitoring, Aquabyte’s AI identifies adult females and other stages in camera images. The portal retains marked photographs that staff can enlarge and review. Counting through a camera avoids handling for that measurement, but Mowi’s Irish trial had not eliminated manual handling or the country’s established lice-checking procedures.[4][1]

Norway’s food-safety authority sees potential for automatic counting to improve the picture of lice in a pen and avoid some handling. Its January 2025 conditions for new dispensations require suppliers to agree to participate in authority-initiated research. The rule recognises that operating cameras can generate evidence needed to evaluate counting methods.[5]

A separate Norwegian project with OptoScale and DNV tested automatic reporting during November–December 2025. The authority concluded that its technical framework could receive daily camera-derived lice counts securely and in a standard format. That was a reporting demonstration with different participants, rather than an Aquabyte accuracy result.[6]

For the farmer, the immediate benefit is a more continuous account of the animals in the water. A change in size distribution, a rising lice trend or a photograph worth examining can become a focused decision while there is still time to respond.

Report an error or suggest a correction