We already said it a few weeks ago. It was not the first, nor was it going to be the last time in this Blog we talk about artificial intelligence (AI) and nature. If in the last article I wrote about Wild Me, a non-profit organization that has been able to develop a data management platform and a framework for research and conservation of fauna using AI, in this article I will explain how thanks to artificial intelligence can electronically observe and control fishing activity.
And no, in this case it has not been Microsoft, Google, Facebook or Amazon, or any of the large Chinese multinationals such as Baidu, Alibaba or Tencent who have developed this solution. It has been led by a Spanish company, Satlink, specialized in satellite telecommunications, which has developed together with Digital Observer Services (DOS) the first tests to apply artificial intelligence in the electronic monitoring of fishing activity.
The advance is based on a machine learning algorithm that DOS integrates into the Satlink View Manager system used by the team of biologists and researchers. This team, located in Erandio (Vizcaya) and made up of 14 professionals, is responsible for analyzing information and images of activity on board fishing vessels and generating analyzes and reports for companies, governments and institutions.
The algorithm, developed by Satlink’s R&D department, allows, from different GPS-based parameters such as time, coordinates, speed and heading, the automatic detection of the different events that comprise a fishing set, and can also detect and identify fish catches, making use of image recognition technology. All this translates into a notable reduction in analysis time and costs associated with observing the activity of fishing vessels.
With the application of this automatic detection algorithm, the time required to identify 30 sets has been reduced from five hours to one, a very significant reduction compared to the analysis by a human observer. Regarding the automatic recognition of the captures, the system registers a detection rate of 75% with a certainty ratio higher than 95% and a false-positive ratio per image lower than 3%.
According to Satlink’s forecasts, this technology could be used in more than 53,000 vessels of more than 12 meters in length around the world, between now and 2028, compared to the 1,000 currently shipping this type of system. And of those more than 50,000 vessels, 38% correspond to Asian fleets, 19% to those of the European Union, around 10% to the United States, the same percentage as Latin America, 5% to Oceania (New Zealand ) and 18% to other organizations.
Asia is undoubtedly, they say in Satlink, the market that will experience a more pronounced growth due to the need for these fleets to progressively adapt their activity to international regulations, compared to scenarios such as Europe, in which fleets, including the Spanish, are transparent in their activity and will only have to update their systems. “Electronic observation makes it possible to replace the work of monitoring compliance with fisheries regulations by observers on board, with total reliability and objectivity through image recording equipment that later allows analysis with observers on land.”
So you know, the fishing vessels and the fishermen. If someone happens to fish too much, for example, artificial intelligence will be able to guess, thanks to satellites and electronic observation, that it has done so. And they can catch you.
