News room

ITOKI CORPORATION (Head office: Chuo-ku, Tokyo; President: Koji Minato) has begun research and development of an application that enables quantitative measurement through image analysis, primarily for the purpose of verifying whether employees are making effective use of space in offices that have been renovated or relocated.

Our Digital Transformation Promotion Headquarters is conducting research and development using cutting-edge technology to provide services related to next-generation working styles, and this initiative is part of that effort.

This time, we first conducted the survey necessary to develop an image analysis detection model in the office and classified and defined activities during work. Activities in the office are classified into various patterns, such as working alone, working with two or more people, meetings, phone calls, concentrating, and breaks. By defining each of these from scratch and building an algorithm, it becomes possible to quantitatively measure image analysis. In the future, we aim to improve the value of office construction and operation services.

This survey is being conducted with the understanding of our employees. In addition, we are using cameras in some parts of the office to record employees' work patterns and conduct image analysis, but we are not identifying individuals and are only investigating their activities. We are also considering edge processing of the images we obtain rather than storing them in the cloud, and are proceeding with this in a cautious manner.

The survey was conducted in ITOKI's newly renovated Nagoya office, "ITOKI NAGOYA DELA," in collaboration with major system integrator OGIS Research Institute Co., Ltd. (Head office: Nishi-ku, Osaka City, President: Kazuhiko Yoshimura). Part of the office was photographed with a 360-degree camera, and survey members visually recorded employee activities. Data was collected by describing each employee's activities from the perspectives of people, things, and places.

After that, we classify the activities that employees perform in the office and determine the "activity definition." Specifically, we determine the definition name and its corresponding explanation. The key point is that we are not redefining existing activities that have been classified in advance, but are determining the "activity definition" from scratch based on the facts obtained from this activity observation.

As a result of the "activity definition" workshop held after the survey, we defined informal "chat" activities. The following are their characteristics:

[Defined Activities]
chat

[Example of feature description]
- Conversations between two or more people that involve casual conversation, consultations, or reports, regardless of whether they are work-related or not
・Conversations occur when one person has the will or through a chance encounter
・The location is a space other than a specific conference room
- Less than 1 metre apart

Going forward, we will use this survey to increase the number of "activity definitions" in order to build algorithms for image recognition AI models and begin developing applications that can quantitatively measure activities.

In the field of office DX, we will continue to use data to conduct research and development on work styles and working environments, contributing to our customers' work style reforms and human capital management.

Expressing ITOKI and Nagoya through art! Nagoya office reopens as "ITOKI NAGOYA DELA"
https://www.itoki.jp/company/news/2024/2402_nagoyadela/

Our Digital Transformation Promotion Headquarters is conducting research and development using cutting-edge technology to provide services related to next-generation working styles, and this initiative is part of that effort.

This time, we first conducted the survey necessary to develop an image analysis detection model in the office and classified and defined activities during work. Activities in the office are classified into various patterns, such as working alone, working with two or more people, meetings, phone calls, concentrating, and breaks. By defining each of these from scratch and building an algorithm, it becomes possible to quantitatively measure image analysis. In the future, we aim to improve the value of office construction and operation services.