NTT COMWARE English Site


Mealthy and NTT COMWARE Officially Collaborate on "Meal Advice Service" with Deeptector®
-- Work efficiency of nutritionists is increased by three times for quality services --

NTT COMWARE CORPORATION (Head office: Minato-ku, Tokyo; President: Satoshi Kurishima, hereinafter NTT COMWARE) has been promoting pilot projects to enhance open innovation schemes in cooperation with partner companies to create new businesses.
Mealthy Incorporated (Head office: Minato-ku, Tokyo; Managing Director: Masayuki Suzuki, hereinafter Mealthy) collaborates with NTT COMWARE, and Deeptector®*1 image recognition AI is officially selected as the first collaboration model for the "Mealthy Meal Advice Service" effective November 1, 2018.
* "Deeptector®" image recognition AI is an AI technology service commonly used in various technologies of NTT COMWARE which constitute NTT Group's AI "corevo®"*2.

1. Outline of the Service
Many companies facing a serious shortage of work force pay attention to the Health and Productivity Management scheme. Companies enhancing Health and Productivity Management are expected to maintain and improve the health and wellbeing of employees and improve their productivity. The "Mealthy Meal Advice Service" supports the employees to learn healthy and balanced diet by using AI to deliver personalized advice. The following are the steps of the service:

Step 1. "Sending a meal photo": the users will take and send a photo of their meal through Mealthy App. The photo is assessed with Deeptector® to see how healthy it is.
Step 2. "Advice from nutritionists": a professional nutritionist gives the user advice on meal planning based on the results of evaluation from Deeptector®.
Step 3. Visualization: the users appreciate a visual image of their health condition including meals they have, weight and step counts.

[Figure 1. "Mealthy Meal Advice Service" Flow]

Figure 1.

The efficiency of work of certified nutritionists who had visually checked the content of the meal photos triples thanks to Deeptector®. Specifically, comments from certified nutritionists which used to take 10 minutes per photo consultation, can be done in 3 minutes, achieving about a 70% efficiency improvement in operation time. By improving work efficiency, it is possible to deliver the advice of quality certified nutritionists considering the lifestyle of each employee. The number of employees receiving the consultation is expected to increase to meet demands from large companies as well.

2.How to Use Deeptector®
"Mealthy Advice Service" uses both the detection and categorization functions of Deeptector® in a hierarchical manner. Therefore, the users have no need to take photos of each dish for consultation; they can take a picture of the whole meal. About 500 types of meals can also be categorized, such as noodles, bowl-rice and soup to distinguish meals by their small ingredients on the high precision images.

[Figure 2. Example: Food Image Recognition of Deeptector®]

Figure 2. Example: Food Image Recognition of Deeptector®

The benefits of Deeptector® are not only quality of detection and classification accuracy but also easiness to add categories. If a new type of cuisine emerges and becomes popular and its category needs to be categorized, such categories can be added by including its relevant type as a new category. This function is used to increase the category while maintaining accuracy. The high scalability and operability are also one of the important factors for Mealthy, which provides the service.

- Comment of Masayuki Suzuki, President of Mealthy, using Deeptector®
"When a new type of food becomes popular, we add that food as a new category; however, our in-house tool took over a month to improve the accuracy having asked engineers to add the new categories. In that respect, we realize Deeptector® is efficient in adding new categories with ease, which will be absolutely necessary to expand our services in the future."

3. Future Scenario
NTT COMWARE plans to develop the dietary recognition AI proven with "Mealthy Meal Advice Service" for consultation regarding eating habits for athletes and fitness for the general public, as well as for businesses related to health insurance and medical services. In addition, we aim to invite potential partners and develop an ecosystem to make the food recognition AI popular. We will continue to promote cooperative creation of new value through open innovation in the future showcasing this collaboration model.

NTT COMWARE will exhibit the service at the 2nd AI/Business Automation Expo in Makuhari Messe from October 24th (Wed.) to 28th (Fri.) 2018.


The information shown in this page is the latest as of the day of the release. It is subject to change without further notice. Thank you for your kind understanding and consideration. PageTop