July 9, 2024

 

AI revolutionises gender sorting in salmon aquaculture

Some aquaculture companies have turned to artificial intelligence (AI) to automate gender identification in salmon aquaculture operations, a difficult task especially when dealing with immature fish, with tiny smolts being the hardest to classify.

 

Traditionally, trained personnel have used ultrasound images to identify sex characteristics, such as gonads, and sort fish accordingly.

 

"You want to make decisions regarding fish as early as possible," explained Stian Rognlid, CEO of AI-focused Aquaticode. The company is developing a gender-sorting platform that can handle up to 10,000 fish an hour. Aquaticode's sorter is currently in full production in Chile.

 

Econexus has provided ultrasound-based aquaculture solutions since 1998, offering a service that performs gender identification of smolts using teams of people, each capable of sorting 50,000 to 80,000 fish a day. Now, Econexus is set to deploy its AI-based gender-sorting solution. The first version will be semi-automated and require fewer trained personnel, with field testing starting in June. A fully automated version is expected to be in the field by the end of the year.

 

"We have our own artificial intelligence system, and it's working really well," said Econexus CEO Jaime Stange. "But it's just the first step. For us, the sex-sorting is just one of 10 or 20 more applications that are useful in aquaculture."

 

Similarly, Aquaticode's Rognlid believes gender-sorting is only the beginning. The imaging systems can also identify fish that have diseases and other issues that will hinder growth. Stange noted that these improvements in fish quality occur early on, before the fish enter the grow-out phase.

 

Gender sorting itself has practical benefits. Rognlid noted that male salmon mature faster than females, allowing more frequent harvesting. Additionally, females produce more fillets than males, and their flesh colour differs, allowing producers to sell males in the whole fish market while females go into fillet processing.

 

Einar Wathne, an aquaculture industry expert and advisor to Aquaticode, highlighted additional benefits of separating the sexes: "Reduced stress can improve feed intake and growth. Reduced stress may also enhance animal welfare. Since males and females differ slightly in shape and meat colour, these traits can be used in production and sales planning."

 

Sorting relies on detecting sex-based differences, typically immature gonads in males, in smolts weighing 65 grams or less. This involves ultrasound technology, which generates high-frequency sound that penetrates the fish. The sound reflects off internal organs and structures, creating an image of the fish's insides.

 

Traditionally, spotting gonads in these images has required trained technicians to study and assign gender to the smolts. AI substitutes a machine learning model for these experts. Using a comprehensive training set of images of fish with and without gonads, the technology categorises fish as male or female. Rognlid stated that imaging occurs while the fish are sedated, typically right after vaccination, and achieves better than 99% accuracy compared to the 90–95% accuracy of a human-based approach.

 

Stange, however, is not yet convinced that AI systems outperform human ones. He does believe that automated systems will eventually perform as well as or better than people.

 

Wathne pointed out a logistical challenge with gender-sorting: extra tanks are required to hold the separated male and female fish, which take up space and add costs. Whether these additional tanks are needed throughout production depends on the facility.

 

Rognlid emphasised that ultrasound images provide detailed information about the internal organs and structure of each fish. Combining this data with external images captured in the visible spectrum and possibly adding an infrared camera can further increase the information available. An AI system can then use these images to cull fish with disease markers, indications of mineral accumulations, or other signs that suggest they are unlikely to grow quickly or be healthy, thus optimising production.

 

-      Global Seafood Alliance