AI helps to track lobsters when 'no two shells are the same'

Like facial recognition for lobsters, AI technology tracing from pot to plate B By Bec Pridham

Scientists have developed artificial intelligence technology capable of recognising individual lobsters.

The innovation allows traceability from pot to plate and has the potential to strengthen consumer trust — and even break into new markets.

The scientists want to further develop the technology for both the processing sector and the fishery.

After nearly three weeks at sea, Joe Sayers docks a boat filled to the brim with treasure — it's red, thorny and alive.

"There is one in here, I know there's 220 millimetres just in the carapace … there's a couple of fish in here that are probably going to be three, four kilos, I'd say, somewhere there," he said.

Every year, Tasmania's commercial fishers catch just over 1,000 tonnes of southern rock lobster, valued at about $90 million.

While things are now on the up, it's been a tough few years for the industry, brought on by trade restrictions and a domestic market flooded with cheap lobsters.

"Now we're up to getting a few dollars in the bank from better prices, but our costs are going up … all the time, like fuel and baits," Mr Sayers said.

But for Mr Sayers, it's a way of life, and after more than 30 years at sea, he wants to hand over an industry in good shape to the next generation.

Scientists from the University of Tasmania's Institute for Marine and Antarctic Studies (IMAS), in partnership with Fiordland Lobster Company and the South Australian Lobster Company, are trying to ensure just that, through newly developed artificial intelligence (AI) technology.

"It's just like a fingerprint or facial recognition technology," associate professor at IMAS Quinn Fitzgibbon said.

"We've used AI machine learning to look at the features on the shell of the lobster, and amazingly, we've been able to show that this technology can pick up those features and they're individual.

If the lobster has been photographed before, the AI recognises it, and if it hasn't, into the database it goes, joining thousands of others from across Australia and New Zealand.