Google develops a visual fish recognition system for fish farms

Google develops a visual fish recognition system for fish farms

Alphabet, a conglomerate of services such as Google Maps or Gmail among many others belonging to the parent company Google, has evolved to Alphabet X, or simply X, to develop particularly ambitious projects, such as Project Loon (which provides internet service through hot air balloons) or Project Wing (a delivery drone project).

His latest project is Tidal , a sophisticated system that will recognize each time individually of the thousands that are in the fish farms with the purpose of promoting the use of them and thus reduce the world consumption of meat and wild fish. According to Astro Teller , the director of X, Tidal aims to reduce the world’s dependence on terrestrial proteins, such as beef, and rid the oceans of harmful fishing practices.

At the moment, Tidal is being tested in fish farms in Europe and Asia . Combine underwater cameras with artificial intelligence techniques like computer vision to track species like salmon. There are many technologies that use machine vision, including object recognition, event detection, reconstruction of a scene (mapping), and image restoration.

Training algorithms

To train Tidal’s algorithms, data collected from a paddling pool at the company’s Silicon Valley headquarters was used. In this way, the particular shapes and movements of each fish were filmed with various cameras .

Some of the patterns collected by the cameras can have very short time laps, of just milliseconds, and are imperceptible to the human eye. However, all this data is an excellent source of information for training an algorithm. As explained by those responsible for Tidal in a statement:

Our software can track and monitor thousands of individual fish over time, observe and record fish behaviors, such as eating, and collect environmental information such as temperature and oxygen levels. This type of information gives fish farm managers the ability to track the health of their fish and make smarter decisions about how to manage the enclosure.

As a result, Tidal can help fish farm owners optimize their fish feed, reduce waste and keep them healthier, which would also reduce the use of antibiotics .

While Tidal is still in an early stage of development, its possibilities may extend beyond the realm of fish farms. Their algorithms could be used, for example, for more general monitoring of the oceans , where researchers often have trouble effectively tracking endangered species like whales and penguins.

In this way, Tidal is part of one of the many projects in which X is involved and which has its sights set on protecting the environment. Some other of his ongoing projects in this regard are: the development of domestic geothermal energy, energy storage or the use of drones to monitor agriculture.