Illegal fishers and wayward sharks are in the sights of new multispectral imaging

The oceans are warming. Reefs are dying. Fish are on the move.

As a result, sharks and illegal fishers are scouring Australia’s coast in search of an increasingly elusive catch, and that, says ESpy Ocean founder Ian Dewey, is having an immense impact on everything from regional tourism to ocean ecologies.

Illegal fishers, like sharks, are elusive predators. Their survival depends on being fast, silent and unexpected. They’re threatening a $1.6-billion regional Australia industry.

Sharks also aren’t behaving the way they used to. They’re turning up in unexpected places, at unexpected times, which can result in tragedy.

“Everyone says use drones or dirigibles to spot them,” Dewey says. “But everyone knows that when we’re on the beach in our string bikinis and Speedos, the last thing we want is a drone above us.”

With dark fleets of illegal fishing boats turning off their tracking systems to breach international boundaries, time is of the essence in addressing the problem, just as it is with wayward sharks.

“Both are increasing problems,” Dewey says. “I only know in terms of the illegal fishing missions that we’ve been involved in, but invariably there are people around protected areas on a daily basis”.

Traditional spotter aircraft can’t cope, and using satellites to track vessels isn’t anything new. What is new is multispectral imaging.

A regular camera captures an image on just three channels red, green and blue (RGB) –  generating a crisp image of the visible spectrum if the weather is clear.

A multispectral image has up to 110 different frequencies, ranging from ultraviolet to microwave.

Read also: How do you get a super-group of whales to come together?

This imaging technology is nothing new. What is new is applying machine learning to identifying what it “sees”.

“So it was a matter of working out what we can do through clouds in all kinds of weather, preferably right on dawn,” says Dewey. “I just started going through what frequencies can do what and – if we are looking for a boat – what the hell’s it gonna look like?”

It’s a similar story for sharks – what multispectral signatures do different species give, at what depth, under what conditions, at what time?

Dewey says the potential to extract such detail from hyperspectral imaging is enormous.

It can identify what a boat is made from, what sort of paint has been used (and how old it is), and what equipment is on the deck.

“All these things mean that your picture is different to every other boat in the ocean,” he says. “If we see you today, we can see you tomorrow, match those frequencies, and say – we got you!”

ESpy demonstrated the potential of the technology for New South Wales Fisheries over the last Easter long weekend. Suitable satellites were identified, access to their hyperspectral cameras was secured, and patrol vessels were stationed in strategic locations waiting for a call to action.

“Our system is incredibly fast, which gives us the edge,” Dewey says. “Generally, our system allows boats to be caught red-handed. That makes it so much easier where the courts are concerned.”

The shark-spotting challenge is a more recent project. ESpy is in initial discussions with NSW Fisheries and the University of South Australia’s Industrial AI Research Centre to develop techniques to spot the predators first thing in the morning and use established behavioural patterns to predict where they could move during the day.

While trespassing trawlers present a major issue, the deadliest offender is often someone much closer to home. One dragnet can strip an ecosystem of everything from algae and small crustaceans to dolphins and turtles, leaving damagethat  can take years to recover.

“Our big problem in Australia is the little guy who throws out a net once or twice,” Dewey says. “He’s generally local, or at least from within 100-or-so kilometres. But he’s got a high risk of being caught, so he just wants to get in and take as much as possible as quickly as possible.”