Is it Possible to Assess the Quality of Tuna Fish Using AI?

Is it Possible to Assess the Quality of Tuna Fish Using AI?

With advancements in artificial intelligence, it is now possible to use an app to assess the quality of tuna fish using AI. It is a fascinating new development in the world of AI. But, is it really viable to check the quality of a piece of fresh fish with an app? The answer is: mostly YES. A chain of Japanese sushi restaurants has pioneered the new app powered with AI to assess certain criteria relating to the quality of each piece of fish. This is enabling swift assessments whereas previously it would have taken trained and experienced human buyers years to gain the relevant skills.

The app is able to assess certain characteristics in pieces of fish such as its color, its fat content, and flesh density. Developed by Dentsu Inc., a Japanese advertising firm, the app called Tuna Scope, makes use of machine learning algorithms to analyse the constitution of the fish. The app has been trained on many thousands of images of tuna tails (cross-sections). From just one image, now the app can grade the piece of tuna on a scale from one to five. The grading is based on visual characteristics, and the company has claimed that with its new app, it has captured the “nuances of the tuna examination craft”.

Tuna Scope has also been tested alongside human graders and has proved to be more than 80% accurate. However, the app has met with criticisms and cynicism from sushi experts and fishmongers. Not everyone in the industry believes that it is possible for an app to grade the quality of tuna just as an expert would. Some industry experts are not entirely convinced that the quality of tuna should be graded on visuals alone. Touch and feel is an important factor in confirming fish quality. Perhaps this is why the app has not proved to be 100% accurate.

Some of the visual cues to look for in a quality piece of tuna include a strong bright red hue, a soft and somewhat bouncy appearance and an element of translucency to the flesh. Experts know a quality piece of fish when they see it and know with their highly trained eyes just what they are looking for.

So, how can artificial intelligence be used to make these assumptions about a piece of fish without being able to touch it? It does however seem that it’s possible for an AI-based app which has been trained to be able to carry out basic assessments on quality. It also makes sense that Japan is continuing to develop this type of technology considering the ageing population and the fact that traditional skills such as tuna fish grading are not being passed onto the younger generations consistently.

The Tuna Scope app is currently being used only by the large sushi chain restaurants, Kura Sushi. It is highly unlikely that it will meet the assessment criteria of high end restaurants and their exacting chefs. These experts and other sushi aficionados are already expressing their disinterest in the app and say that they will continue to assess by eye and hand.