In the business of keeping the fishing industry sustainable and honest

A Nelson-based security camera manufacturer has developed a unique system to supply on-board monitoring of fishing boats. Now it's learning to deal with the vast amount of footage produced by its cameras.

Three years ago, Snap Information Technology’s 360° panoramic video cameras were a great solution looking for a grunty problem – which international research suggests is a scary place for a startup to be. SnapIT’s high definition cameras were better than most – they could provide 360° (or 180°) footage of tourism spots, holiday homes, or construction sites. And customers could pan around geographically, or scroll backwards and forwards in time, just by clicking and dragging a mouse.

But other companies made good security cameras too – and places like China made them considerably cheaper than SnapIT was ever going to. The company’s founders, brothers Andrew and Chris Rodley, needed a niche – preferably a global one.

Then three years ago, they found one – or it found them.

“I was down on the the Auckland waterfront doing installation work on a time lapse camera on the ASB building, and this guy approached me to ask if I could put one on a boat,” says Chris Rodley. The guy turned out to be Eric Barratt, then managing director of New Zealand’s biggest fishing company, Sanford. And Barratt definitely had a grunty problem: how to make surveillance cameras tough and reliable enough to be used instead of human observers on New Zealand’s fishing fleet.

Fisheries observers have been an crucial part of the fishing industry for the last 30 years. As the name suggests, they watch what happens on fishing boats – auditing and enforcing how much fish is being caught, and how many undersized or unwanted fish are being thrown back; recording any dolphins or seabirds being killed or injured, and ensuring health and safety standards are being followed.

But humans are expensive, and are only really practical on the 50 or so big ocean-going vessels – boats that head to sea for weeks at a time, and have plenty of space on board for a fisheries observer or two. What is more problematic is monitoring the 1000 or so smaller boats that head out for a day or two from a myriad of small ports around the coast.

So the Ministry for Primary Industries looked for a technological alternative – and the first fixed cameras were trialled on fishing boats in 2003. It didn’t really work, says David Middleton, CEO of fishing industry research company Trident Systems, which is now working with SnapIT on video-based monitoring systems. The harsh weather meant cameras were always breaking down, he says, and because fixed cameras could only watch one part of the boat, but fishing activity might be going on on either side and at the back of the vessel, a single boat might have to have up to six cameras on board – adding to the likelihood at least one wouldn’t be working.

Cameras weren’t ever completely rejected, but they weren’t widely adopted either, Middleton says.   

Andrew Rodley (GM Snap Information Technologies) Darren Guard (Trident Systems LP - Fish Eye project manager) Chris Rodley (CEO Snap Information Technologies)

A harsh environment

Barratt’s chance meeting with Chris Rodley on the Auckland waterfront marked a turning point – both for SnapIT and the MPI’s technology project. The brothers were asked to pitch (alongside local and international competitors) for a contract to put video cameras on a fleet of vessels to collect information on seabirds being caught by long-lines.

“[SnapIT] demonstrated their cameras using construction monitoring as their example,” Middleton says. “Because they were capturing a 360 or a 180 degree image, that dealt with the problem of things happening outside the sight of a fixed lens camera. Plus they could do clever things with panning and zooming, and they came up with smart ideas about using wifi to get footage ashore, rather than using hard drives, which had logistical challenges.

“We’ve spent the last three years working with them to make their technology fit for purpose on fishing boats.”

He says the trials so far have been enough to convince MPI that the cameras work, and are a cost-effective alternative to humans, particularly with the smaller boats.

“We’ve demonstrated video observation can be an adequate replacement for human observers. And it’s significantly cheaper. We haven’t done it at a large enough scale yet to be fully confident how much cheaper, but we believe it’s at least 50%.”

Still, getting to this stage hasn’t been easy, Rodley says. Cameras have to handle extremes of temperature and harsh marine environments. They can easily get damaged by fishing gear – on initial trial at sea in 2014, the bolts holding the cameras to the ship corroded after only two weeks, despite SnapIT using marine-grade aluminium.

The newest version of the camera is 90% smaller than the first model, and made out of stainless steel. 

The company won the Most Innovative Hi-Tech Agritech Product award at the New Zealand Hi-Tech Awards in May 2015, and the Restorative Innovation Award at the Sustainable Business Network awards in late November. It also won the People's Choice award at the 2014 Innovators' Awards.

The hard work has been worth it, Rodley says, with staff numbers now up to 10, most of them developers. As well as contracts with MPI in New Zealand, the company is also looking at rolling out cameras onto Fiji’s fishing fleet, and Rodley is just back from a trip to the US, exploring leads there.  

“We are crazy busy.”

Rolling out video monitoring

The latest New Zealand contract involves Trident Systems, using SnapIT’s camera systems, being contracted by MPI to undertake a three-year video observation programme on 20 trawlers to monitor undersized snapper being returned to the sea, both as a way of checking that boats are following the rules, and to get data about existing stocks of snapper.  This extends the position monitoring system already implemented by Trident and Snap IT for the 70 vessels in the snapper 1 fleet. 

While a few dozen boats may not seem that many from a fleet of nearly 1000, Middleton believes adoption will start building up.

“Video observation has been a challenge for the seafood industry, as many people find it hard to accept having cameras in their workplace monitoring them around the clock. But there’s an increasing acceptance in some parts of the industry that the benefits outweigh the disadvantages. It’s not compulsory, but the 14 shareholders in Trident Systems group control more than 50% of the onshore and deepwater quota. We want to see research and development supporting sustainable fishing and we have chosen to be proactive and adopt systems like this.

“I would expect to see widespread adoption [of video monitoring] within five years.”

Processing all that information

Now they have gear that works out at sea, Rodley says the next step has been developing computerised (rather than a paper-based) monitoring systems, working out the best way to get the vast amounts of data produced by 24-7 video monitoring from the boats to the shore, and then how to review it as efficiently as possible – a project where SnapIT is working with Trident Systems.

Snap’s latest system sees tracking data from the ships being beamed to shore by satellite every five minutes (rather than every four hours, as in the past), which means fisheries officials can keep tabs on boats, allowing them, for example, to link any dumping of fish with a particular fishing vessel.

Snap and Trident are also working on systems to transfer the main video data from ship to shore. Using satellite or cellular transfer when the ships are at sea is expensive with the quantities of data involved, so the companies are looking at putting in a system of port-based servers which can transfer data from boats to shore via wifi when the boat comes back to land. Then the information can be sent from the port to a central processing point via ultra-fast broadband.

“We are experimenting with wifi transfer ashore with three vessels in and out of Timaru, and we’ve also done trials in Tauranga. We will put servers into four more ports over the next month,” Middleton says.

Once the cameras are rolled out across a larger number of fishing boats, the amount of data coming in will be huge, Rodley says. “I talked to one of the larger suppliers of data storage hardware in New Zealand and they said once we get to scale, we will be one of the largest data storage facilities in New Zealand.”

Analysing all that data

The next stage is to build video analytics capabilities into the retrieval process so the central computer system can automatically view the video footage and select relevant material, says Middleton. Until this point, all the video has been reviewed manually.

Monitoring the video footage is a manual operation at the moment.

However, some of the money from a $1.2 million Callaghan Innovation grant (the latest of a number of awards) will be used to develop processes where computers can match data from the boat (speeds, positions etc) with the footage, and identify key activity (fish being brought on board, for example) and separate it out from unwanted footage (when no fishing is taking place). The holy grail, says Middleton, is eventually to have the computers being able to focus on the species and size of fish in the video footage, and process that data automatically.

Andrew Rodley (GM Snap Information Technologies) Chris Rodley (CEO Snap Information Technologies) Darren Guard (Trident Systems LP - Fish Eye project manager)