A sushi restaurant chain in Japan is using AI to evaluate tuna cuts

The software was trained on thousands of images of tuna tail cross-sections.


The highlight of almost any sushi platter is the fatty tuna. Finding that perfect cut of tuna that melts in your mouth is something that fish buyers spend years of their life learning how to do. But now a Japanese advertising agency named Dentsu Inc has developed an app called Tuna Scope that allows someone to do the same with little to no training (via The Verge).

The firm trained the machine learning algorithm that powers the software using thousands of images of tuna tail cross-sections. The cut can tell human buyers a lot about the quality of fish they're about to purchase. In testing against human experts, Dentsu claims it found the app gave the same grade more than four out of five times.

Tuna expert

The app is currently in use by one company, conveyor belt sushi chain Kura Sushi. The restaurant buys the majority of its tuna outside of Japan. Part of the reason the company started using the app is that it allows its employees to grade tuna without traveling. That's a significant perk during the current pandemic. Moreover, conveyor belt restaurants in Japan tend to offer the least expensive sushi, so there's a cost-saving aspect at play as well.

As you might have already guessed, traditionalists are skeptical of the app. Keiko Yamamoto, a sushi chef who teaches in London, told The Verge it's challenging to convey the exact qualities tuna buyers look for when they see a fresh catch. There's also the question of whether the app can scale to meet the demands of high-end sushi restaurants and the exacting chefs that run them. Tuna Scope examines images of frozen tuna tail cross-sections to deliver its quality verdict. However, high-end restaurants tend to purchase their fish from suppliers that deal with freshly caught tuna. As they cut the fish, they give a variety of grades to different parts of the fish. Like with most instances of new technology, we’ll probably see some businesses continue to do things the way they’ve always done them.