Japanese AI learned to evaluate fish quality

Engineers from Japan introduced an artificial intelligence (AI) -based model that can evaluate tuna quality from a single photograph. The tool appeared as an application.

A Japanese restaurant chain has begun using an AI-based app to evaluate tuna quality. Researchers believe that this is a key product for cooking in the country, so it requires many years of experience to determine its quality. Scientists believe that this skill can be transmitted to an automated model.

The Tuna Scope app was developed by Japanese advertising firm Dentsu Inc. It uses machine learning algorithms that have trained on thousands of images of cross-sections of tuna, pieces of meat that can tell a lot about its quality.

The application can based on one photo, evaluate the quality of tuna on a five-point scale, based on visual characteristics, such as gloss of scales and layering of fat. These are the introductory ones used by live fish appraisers.

Researchers noted that those qualities that speak about the quality of tuna are not always visible to the eye, but the trained algorithm can hardly miss them. High-quality tuna has a rich bright red color and a certain degree of transparency. These data are important for the new AI.

Scientists also noted that the tool will be useful during a pandemic. It can replace part of the live evaluators – the product can be selected from a distance.