Fishing is a way of life for coastal communities around the world. An estimated four million fishing vessels sail the world's oceans, providing fish for a global seafood market valued at over $120 billion.
'It's hard to overstate the importance of fish,' says Nick Wise, CEO of the nonprofit organization OceanMind. 'There are three billion people in the world who rely on seafood as their primary source of protein, mostly in developing nations. Twelve percent of the world's population relies on the wild-capture seafood industry directly or indirectly for their livelihoods.'
Overfishing - when fish is caught faster than stocks can replenish - is a significant factor in the decline of ocean wildlife populations, not least because of the bycatch of other marine life such as turtles and cetaceans. Each of these creatures is an important part of ecosystems and the biodiversity of the ocean.
The United Nations Food and Agriculture Organization estimates one-third of all fish stocks are now overfished and are no longer biologically sustainable.
'A collapse in fish stocks and a failure to manage fishing sustainably,' says Wise, 'would lead to a food security crisis and result in significant poverty around the world.'
To fight back against this overfishing, OceanMind is using the power of AI to map data and then feeding that information to government authorities to help catch perpetrators.
[Subscribe to Microsoft On the Issues for more on the topics that matter most.]
Regional, national and international regulations are used to manage fishing efforts and can include restrictions on fishing out of season, using banned gear or techniques, or catching more than a set quota.
There are many ways of trying to catch those flouting the law, such as patrol boats, on-board cameras, and the remote electronic monitoring of discards.
However, the vastness of the ocean makes the job difficult.
OceanMind's system currently tracks thousands of boats, with the capability of tracking millions, across the globe by gathering data from a wide range of sources, including collision-avoidance transponders aboard boats; radar images; satellite imagery; and cellphone signals. Analyzing these enormous datasets is beyond the capability of any one person. OceanMind has developed machine-learning algorithms that predict the type of fishing behavior based on vessel location, and flags suspicious and potentially illegal activity such as fishing too close to the shore.
But the system can't tell on its own whether anyone is breaking the law.
'The difference between legal and illegal fishing is simply whether or not the vessel had a license to do what it did in that place, at that time, and in that way,' says Wise. 'That's what makes combating illegal fishing challenging: One vessel making a particular maneuver might be legal, another vessel doing the same thing next to it might be illegal.'
OceanMind's fisheries experts verify the alerts flagged by AI and coordinate closely with the relevant authorities, who can then decide whether to investigate further. The organization already has partnerships with governments, including Thailand's, which can then target resources to catch offenders.
Until now, OceanMind has used onsite servers to process the data that comes in every day. 'We were basically running a day behind,' explains Wise. 'We reviewed things that were happening yesterday.'
Through a Microsoft AI for Earth grant, OceanMind is moving its data analytics to the Microsoft Azure cloud. 'The collaboration with Microsoft is going to bring all of that data through our system much more quickly and apply the AI in near real time.'
That transformation will make a big difference to enforcement. Real-time monitoring will help authorities plan patrols that can catch illegal fishing as it happens.