-Development of an optical sensor that directly learned the taste and texture felt by human via AI-
A Group of Researchers from Institute of Food Research, NARO (NFRI) have succeeded in measuring the "tastiness" of tomatoes by letting an optical sensor learn the "taste" and "texture" felt by human while eating tomatoes. The developed optical sensor instantly displays the characteristics of tastiness such as "sweetness", "umami", "juiciness", and "firmness" together with the sugar content and lycopene content by just placing the fruit on the optical sensor. It is expected that a new "food business model" can be created which utilizes tastiness data, providing distribution services that meet the detailed needs of consumers with tools visualizing the "tastiness" of agricultural products and the ingredient contents.
Technology that digitizes various qualities of food is required to build a smart food supply chain that optimizes food production, distribution, and consumption. "Tastiness" is one of the most important quality considered by consumers and it was being investigated so far by sensory evaluation where assessors actually eat and evaluate. However, in sensory evaluation, it is difficult to examine a large number of samples simultaneously, and there are issues like variation in the results depending on the assessors. On the other hand, an optical sensor (near-infrared sugar content meter) that can measure the fruit sugar content without cutting it (non-destructively) has been developed and is widely used in fruit sorting facilities. This optical sensor technology is utilized to develop a technology that can estimate "tastiness" without actually eating food.
Specifically, the estimation of "tastiness" was realized by letting the optical sensor directly AI-learn the "taste" and "texture" scored by the trained assessors. Also, a prototype based on a commercial machine (Fruit Selector, manufactured by KUBOTA Corporation) was developed. By irradiating tomato fruits with light, sensory evaluation results such as sweetness, umami, juiciness, and hardness are estimated. Moreover, the functional ingredient lycopene, the sugar content and acidity of fruit are simultaneously estimated and displayed along with the "tastiness" using the technology developed by NARO.
This research study has shown the possibility of easily and objectively evaluating the tastiness of food, and completion of a prototype that anyone can easily measure the tastiness. In the future, this device will be used to verify the effect of the actual evaluation axis of "tastiness" on the commercial value of food products.
Xinyue Li, Mizuki Tsuta, Fumiyo Hayakawa, Yuko Nakano, Yukari Kazami, Akifumi Ikehata (2021). Estimating the sensory qualities of tomatoes using visible and near-infrared spectroscopy and interpretation based on gas chromatography-mass spectrometry metabolomics. Food Chemistry, 343, 128470 (2021). https://doi.org/10.1016/j.foodchem.2020.128470