Demonstration experiment of security operation to detect abnormalities and their signs from camera images at Tokyu Corporation Shibuya Station

Tokyu Security Co., Ltd. (headquarters: Setagaya-ku, Tokyo; President: Shimogata Kazunaga; hereinafter referred to as "Tokyu Security"), with the cooperation of Tokyu Corporation (hereinafter referred to as "Tokyu Corporation"), will collaborate with Macnica (headquarters: Yokohama City, Kanagawa Prefecture; President: Kazumasa Hara; hereinafter referred to as Macnica), a total Services & Solutions provider in semiconductors, networks, cybersecurity, and AI/IoT, to conduct a demonstration experiment (hereinafter referred to as "this demonstration experiment") of security operations using icetana (hereinafter referred to as "icetana"), an anomaly detection solution that detects abnormalities and their signs from security camera footage, at Tokyu Corporation Shibuya Station (Toyoko Line/Den-en-toshi Line) from July 1, 2022.

In this demonstration experiment, Tokyu Security Railway security guards checked the station camera footage that was detected as an "unusual feeling" by icetana, an anomaly detection software that utilizes Macnica 's AI image analysis technology. This test verifies whether early operations such as checking for leftovers can be carried out. Image analysis technology is used to detect problems between customers within the station, signs of sudden illness, abnormal behavior, etc., and early response to incidents and accidents. By doing so, we aim to further strengthen safety measures and improve security quality on the Tokyu Line.

The outline of this demonstration experiment is as follows.

[Outline of this demonstration experiment]
Period: Scheduled for July 1, 2022 (Friday) to July 29, 2022 (Friday)
Location: Tokyu Corporation Shibuya Station (Shibuya Ward, Tokyo)
Number of target cameras: 12

<Demonstration test image>

<Flow of demonstration experiment>

(1) Select security cameras (12 units) installed in the Tokyu Railway Shibuya Station premises (near the ticket gate, concourse, platform), and connect icetana to acquire images.
(2) Icetana analyzes the acquired security camera footage over a period of about two weeks, and AI learns the local environment.
(3) Icetana detects abnormalities in the video and their signs in real time as "unusual 'discomfort'".
(4) In the event of an abnormality, the PC for monitoring the detection results will detect the "discomfort", and Tokyu Security Railway Security Guards will carry out the operation.

*1.icetana only acquires security camera footage, and the acquired footage does not include information that can lead to the identification of individual customers.
*2. The image data acquired and analyzed in this demonstration experiment will be handled only by Tokyu Corporation, Tokyu Security, and Macnica, and will not be used for any purpose other than this demonstration experiment and post-verification.
*3.icetana is a system that connects to existing security camera systems and supports the discovery of any abnormal phenomena.
By detecting and displaying only "unusual conditions" from images from hundreds of cameras in the facility, it will be possible to notice and respond to events that could not be fully confirmed until now. Unlike regular video analysis, it is possible to discover various events with one software, so it is a versatile solution that can be used to improve the efficiency and sophistication of security.
For detailed information about icetana, please refer to the URL below. https://www.macnica.co.jp/business/semiconductor/manufacturers/nvidia/products/136844/