The tuna industry is already big business, pulling in billions of dollars each year and now it’s due to get another boost with thanks to deep learning techniques that aim to get computers to recognize tuna automatically. The next step up from this would be for the computers to be then able to distinguish between albacore and yellowfin tuna. This is the plan moving forward for the Nature Conservancy, who are a non-profit environmental organization that looks to conserve all lands and water on which life depends.
By using AI deep learning techniques, their hope is that they can help fisherman reduce the amount of protected species that are accidentally being caught when fishing for tuna (including sharks and turtles). It will also help to prevent overfishing while allowing endangered species a chance to recover. While companies like Facebook and Google are using AI and deep learning to recognize faces in pictures, The Nature Conservancy would like to see the same technology applied in the tuna fishing industry.
Illegal shark finning is something that is happening right now. Luen Thai Fishing was fined $120,000 in 2013 for breaking this law and had many illegal items confiscated from their boat. Since then the company has implemented various procedures to prevent catching sharks and has a strict total ban shark policy. Electronic monitoring systems have also been set up by the Nature Conservancy that can record most of what happens on the boat in case of any discrepancies.
Chief technology officer of The Nature Conservancy, Matt Merrifield, said “The large fishing companies may not trust their captains to be doing the right thing. But with the monitoring system, the companies can show that the fishermen are “operating by the rules.” They have also set up a partnership with Kaggle, who are a startup company known for hosting data science competitions on behalf of General Electric and Walmart. There will be a contest in place for researchers with a $150,000 prize for whoever can build the most accurate algorithm that could sift through hours of boating footage data and try and identify the fish that it sees.
This is a tough challenge, and not one the companies are expecting to be solved quickly. Kaggle CEO, Anthony Goldbloom says “My hunch that this should be possible,” and estimates that within a month researchers should be close to the finish. The contest will run until April, and whoever has the most accurate algorithm then will be declared the winner and be given the cash prize.
More News To Read