A university-launched venture in Japan has been striving to spread its meat quality evaluation technology using artificial intelligence, aiming to make the wagyu beef grading done by human eyes more precise, while using the data to improve cattle breeding.

MIJ labo Inc. in the northernmost main island of Hokkaido has developed cameras that use an AI system to calculate more than 10 factors such as the density and color of marbling by taking a picture of the surface of a carcass.

A graduate student at Obihiro University of Agriculture and Veterinary Medicine takes a photo of the surface of a beef carcass in Obihiro, Hokkaido, in October 2020. (Photo courtesy of Keigo Kuchida)(Kyodo)

"A fair evaluation of luxury wagyu will be an advantage in marketing them overseas. I hope our products will be widely utilized," said Keigo Kuchida, a professor at Obihiro University of Agriculture and Veterinary Medicine who developed the cameras and the analyzing system.

Since its founding in 2018, MIJ labo's cameras have been used in about 10 meat markets and research institutions in Japan and overseas including in the United States and Australia.

Currently, wagyu grading is done by Japan Meat Grading Association officials, who examine the surface of a carcass and check criteria such as the proportion that is edible, the density of the marbling as well as the firmness and texture of the meat.

This means the grading from C1 to the highest A5 is largely dependent on the judgement of the officials.

"Our technology has enabled more accurate evaluation," Kuchida said, adding he is aiming for the analyzed data to be used as a reference when official graders rank wagyu beef.

Kuchida, 56, first developed a camera in 1998 to make marbling classification more objective.

Keigo Kuchida, a professor at Obihiro University of Agriculture and Veterinary Medicine, explains his research in Obihiro, Hokkaido, on March 30, 2021. (Kyodo)

Since then, he has worked on improving the accuracy of analysis and camera miniaturization. The venture now sells three types of cameras with different shapes and prices.

The data collected during evaluations are also shared with cattle breeders via a cloud system provided by the company.

Since the quality of a fertile bull is estimated by the quality of its offspring, the data can help breeders choose appropriate stud bulls to produce better meat with more marbling, Kuchida said.

With other elements than marbling involved in determining the taste of beef, Kuchida said his next target is to develop a camera equipped with a sensor that can quantify flavor.