Global
Japanese
Site Map
Global
Contact Us
Japanese
Site Map
Global
Contact Us

The Toshiba Co-creation Center “Creative Circuit” Implements Video Recognition AI to Promote Value Creation with Co-creation Partners

-Accelerating the social implementation of AI technology by sharing images using Toshiba’s latest video recognition AI-

24 February, 2022
Toshiba Corporation

Overview

At the Toshiba Co-creation Center “Creative Circuit(*1),” which opened last February as a base for co-creation activities in the Toshiba Group, Toshiba has implemented functions using the company’s latest video recognition AI to create new value with co-creation partners through experiencing the latest technology.
As a first step, Toshiba has implemented four functions at Creative Circuit: crowding detection, suspicious behavior detection, unnatural posture detection, and individual recognition. By allowing co-creation partners within and outside of the Toshiba Group to experience AI technology, Toshiba will work with its co-creation partners to create new value from AI technologies, leading to earlier social implementation of Toshiba AI.
Furthermore, to make Creative Circuit a safer, more secure, and more creative environment during the COVID-19 epidemic, the company has developed a communication-activity monitoring system that uses a crowding detection function and is operated within Creative Circuit. This system detects the state of crowding while also calculating and displaying communication activity levels to support active communication while preventing excessive crowding, thereby increasing the efficiency and effectiveness of co-creation activities.
Toshiba plans to continue introducing new AI functions at Creative Circuit, which will also be used as a base for value creation using AI technology with its co-creation partners, contributing to the creation of innovative solutions for solving social issues and realizing a sustainable society.

Background for implementing a video recognition AI at Creative Circuit

Many fields apply video recognition AI as an indispensable technology for recognizing human behavior and the surrounding environment. The AI business is expected to continue expanding, with forecasts predicting that the domestic market will nearly double from 2020 levels to 1,936 billion yen by 2025(*2). Toshiba has been researching and developing image recognition technologies for over 50 years and has contributed to solving many social issues. Most recently, the company has leveraged its extensive knowledge and experience in the infrastructure business to develop many competitive proprietary image recognition technologies that support infrastructure services(*3).
To address global-scale social issues related to energy, resources, and climate change, and to realize a sustainable society, it will be important to engage in co-creation activities that bring together diverse people across the boundaries of organizations and specialties, thinking about issues and coming up with ideas together. In co-creation activities, it is necessary to quickly share the latest technologies developed with co-creation partners and deepen discussions of value creation.
Toshiba established the Creative Circuit co-creation center as a base for practicing value co-creation with its customers and partners. By implementing its latest AI technologies in this facility, thereby allowing its co-creation partners to experience them firsthand, Toshiba will be able to work with those partners to create new value by using its AI technologies.

Features of the implemented functions

Toshiba installed cameras at Creative Circuit and implemented the four AI-based video recognition functions(*4) described below so that visitors and partners can easily experience Toshiba’s highly accurate sensing video recognition AI and verify its performance. The company also implemented “communication-activity monitoring” as an example of its utilization.

Functions:

  1. Crowding detection
    To visualize crowded places, degrees of crowding, and distances between people having conversations, the system performs sensing based on object detection technologies(*5) in order to analyze and display crowded situations based on the positions of people, the joint positions of individual bodies, distances between them, and the directions they face (Figure 1).
  2. Suspicious behavior detection (detecting people who are loitering or looking around restlessly)
    To find people needing help as quickly as possible, the system detects and notifies operators of seemingly confused people or suspicious behavior such as loitering or looking around restlessly, based on the trajectory of the person’s movement and the positions of joints of individual bodies as sensed by object detection and tracking technologies(*5) (Figure 2).
  3. Unnatural posture detection (detecting persons who have collapsed or are crouching)
    To detect people needing rescue as quickly as possible, the system detects and notifies operators of people in an unnatural posture, based on the person’s position and the positions of joints of individual bodies as sensed by object detection and tracking technologies(*5) (Figure 3).
  4. Individual recognition (distinguishing between individuals, even when they’re wearing masks)
    To provide value specific to a registered person, the system uses facial recognition technology(*6) to identify the registered person based on a single stored facial photo (Figure 4).

Utilization examples:
Communication-activity monitoring
The system uses a crowding detection function to visualize communication states. While detecting crowding states, the system also calculates and displays the degree of communication activity (Figure 5). In line with lifestyle changes arising from the COVID-19 pandemic, the system can both prevent excessive crowding and contribute to promoting communication, thereby increasing the efficiency and effectiveness of co-creation activities.

Figure 1: The crowding detection function.
Figure 2: The suspicious behavior detection function.
Figure 3: The unnatural posture detection function.
Figure 4: The individual recognition function.
Figure 5: Monitoring increases in communication.

At Creative Circuit, co-creation partners will be able to experience these functions using AI, which will lead to the creation of new ideas and accelerate the social implementation of AI technologies.

Future developments

Toshiba plans to continue introducing new AI functions at Creative Circuit to stimulate co-creation activities and create new value from its AI technologies. Aiming for early social implementation, the video recognition AI technology implemented at this time will also be deployed in Toshiba Analytics AI SATLYS™(*7), provided by Toshiba Digital Solutions Corporation.

*1: Co-creation Center Creative Circuit
The Co-creation Center is located in the Toshiba Smart Community Center in Kawasaki, Kanagawa Prefecture. This facility is used for co-creation activities by the Toshiba Group and its customers and is not available for use by the general public.
https://www.global.toshiba/jp/co-creation/corporate.html (in Japanese)

*2: Domestic market for AI business
Source: Automation News, https://www.automation-news.jp/2021/07/57280/

*3: Toshiba AI Technologies Catalog
https://www.global.toshiba/jp/technology/corporate/ai.html

*4: All implemented functions are realized with existing technologies.

*5: Kobayashi and Shibata. “Simultaneous estimations for object detection and multi-object tracking using attention in space and time series.” MIRU2020.

*6: “Toshiba has achieved world-highest levels of recognition accuracy in the US National Institute of Standards and Technology’s face recognition benchmark test, and the highest recognition accuracy of any Japanese company” (November 2021).
https://pages.nist.gov/frvt/html/frvt11.html

*7:  Toshiba Analytics AI SATLYS™
https://www.global.toshiba/jp/products-solutions/ai-iot/satlys.html (in Japanese)
https://www.youtube.com/watch?v=Xy20NvUptTY