"It's incredible the change in technology just in the last twelve months," says SnapCore CEO Chris Rodley in reference to the capabilities of cameras for vessels.
Not only can they provide better protection for the privacy of crew members, he says, but they can also identify bycatch, bird presence or even decipher between a net being on the boat or in the water.
The latest AI camera unit to be launched by the Nelson-based company has capacity to store enough data for an entire fishing season and gives capability for real-time analysis of what's in view anywhere in the world thanks to solar powered GPS and satellite communication systems.
This system, which is not much bigger than a pack of butter, results in the ability for active monitoring without the need for an observer on board. Its capabilities include species differentiation to the point it can identify a bluefin tuna from yellowfin, and facial blurring at the time of data capture so a crew member's identity is protected even before footage is saved to the hard drive or transmitted.
The new system, called Snap AI, not only protects privacy and reduces storage and review costs, says Rodley, but also gives scope for the industry to achieve greater value for products.
"Consumer trust - trying to bring an incontestable record of the products caught - is significant but there are moments in the lifecycle of the product that are really interesting, that demonstrate quality, and differentiates us from other markets around the world," he says. "I think telling that story is key and using this technology to do that is powerful while protecting the identity of those on board."
SnapCore's AI camera technology is already being used widely on fishing vessels in the USA.
"One of the benefits of the new system is it can be programmed to identify 'moments of interest' which for some markets in the USA - as far as the regulator is concerned - involves identifying bycatch species," says Rodley. "To have a human on board makes compliance costs high, but using our technology means they've been heavy reductions in some markets.
Instead, AI identifies any moments bycatch is landed and then a reviewer on land can be notified to watch just that portion of the footage. We also have the facility to allow for weighing and measuring those fish automatically."
In some cases on trawlers, around 0% less footage has to be reviewed, he says.
"What we talk about is making the video reviewers superhuman," says Rodley. "The goal is for surveillance and monitoring to be as seamless and cost-effective as possible and provide the best data to the regulator while protecting the privacy of the crew members."
It's hard to believe SnapCore started out as two brothers - Chris and Andrew - working from a garage. Having recently moved to new premises near Nelson Airport, the organisation has invested in, and made in-house, machinery to manufacture the hardware on-site. Since its inception in 2007
the firm has now grown into five business units coming under the umbrella of SnapCore, with Teem Fish monitoring responsible for vessel camera technology.
"From day one our desire has been improving transparency and reducing compliance cost," says Rodley. "Now our cameras are really robust and capable of going 2000 metres under water. We'vedesigned it, we manufacture it here in this factory - it couldn't be more 'made in New Zealand!'"
He notes that although a human reviewer on a boat takes an hour to watch one hour of footage, portions could be sped up with a traditional electronic monitoring programme. "If you add AI with an audit process though you can even more significantly reduce the time it takes to review just by focussing on those key moments."
Another advantage to the firm's AI breakthrough is that the technology requires no additional training for each application. "In the past it was very challenging to build a global model so each single vessel's technology would require individual training," says Rodley.
The firm now has cameras on vessels in Canada, Australia, and Pacific and Indian oceans, as well as the USA, with a contract for the latest AI-enabled version already in place with New Zealand firm MacLab.
"In the future, the obvious next step for this technology is to assist in the creation of an automatic logbook submission, which gives further reduction in effort by the crew."
If Rodley and his team have their way this won't be far off, he says.
