Jizaie Launches Risk Detection Function Using Video Understanding AI

Jizaie Inc. (Headquarters: Chiyoda-ku, Tokyo; Representative Director & CEO: Junki Nakagawa; hereinafter "Jizaie") has launched a video detection function powered by video understanding AI that can be controlled with natural language.

This function allows AI to analyze camera footage and automatically detect people and situations. Its most significant feature is "Custom Detection," which enables users to specify detection targets using natural language (text). Users can simply input the situation they want to detect in text, allowing for flexible detection settings tailored to on-site conditions.

Furthermore, it comes standard with detection models directly related to safety management, such as crowd counting and intrusion detection, allowing for immediate operational use.

This contributes to advanced safety management and operational improvement by enabling real-time remote understanding of work site conditions and risks.

This function will be made available to users of "JizaiEyes," Jizaie's fixed-installation network camera.

Figure 1: Fixed-installation network camera "JizaiEyes"

Development Background

In work sites such as construction, manufacturing, and infrastructure, camera footage is widely used for safety management and situation awareness. However, it is difficult to continuously monitor all footage manually, and overlooking dangerous situations can lead to accidents and problems.

Furthermore, on-site hazards are not limited to single events but are diverse and situation-dependent. Therefore, the conventional approach of individually developing AI algorithms for each specific condition made it difficult to address all risks.

In response to these challenges, Jizaie developed an AI video analysis function that utilizes video understanding AI to flexibly detect situations defined by natural language. This enables rapid and comprehensive responses to various on-site risks.

Main Features of AI Video Detection Function

【1】Custom Detection that Allows Specifying Detection Targets with Natural Language

By simply inputting the situation to be detected in text, the AI detects the target from the video. This allows on-site personnel to intuitively set detection targets according to their needs, without requiring specialized AI model configuration.

・ Excessive loading on heavy machinery

・ Person near a vehicle

・ Intrusion by a person at night

Figure 2: Custom Detection Result Confirmation Screen

【2】Standard Detection Models

It comes standard with detection models for safety management purposes, such as the following:

・ Crowd counting

・ Helmet non-wearing detection

・ Intrusion detection

These models automatically detect work site conditions and support safety management. In the future, Jizaie plans to expand the types of detection models according to on-site needs.

Figure 3: Detection Result Management Screen (List View)

【3】Notification of Detection Results and Data Management

When a detection occurs, real-time alerts are delivered via the notification function. Detection results can also be viewed from the management screen and downloaded in CSV format. This allows the detection data to be used for operational improvement and safety management by reviewing it. By accumulating detection history, it can also be utilized as a data foundation for analyzing on-site risk trends.

Future Developments

In the future, in addition to expanding detection targets and strengthening data analysis functions, Jizaie aims to realize integrated site management from AI-based situation awareness to response by promoting collaboration with remote operation platforms.

Through its "Site OS," Jizaie will continue to support the advancement of site management and improvement of safety by combining AI and remote operations.

About Jizaie

With the mission of "creating a world where everyone can work across time and space," Jizaie is a startup originating from the Inami Laboratory at the University of Tokyo, developing "Site OS," which integrates video, operation, and control necessary for remote work. Through its unique video compression and transmission technology, it achieves high-definition, low-latency video transmission even in environments with unstable communication. Jizaie offers "JizaiEyes," a fixed-installation network camera for outdoor use, and "JizaiHands," a wearable cloud camera that allows real-time sharing of worker-perspective video. Furthermore, through the remote operation platform "JizaiBase" and the integrated viewing platform "JIZAIPAD," it provides an industrial remote work platform that integrates everything from video acquisition to remote operation and equipment control. Jizaie aims to build a new employment infrastructure that transcends the constraints of distance and environment in sites with significant physical limitations, such as construction, manufacturing, infrastructure, and disaster response.

Company Profile

Company Name: Jizaie Inc. Established: November 2022 Representative: Junki Nakagawa Location: Otemachi Building 5F Inspired.Lab 569, 1-6-1 Otemachi, Chiyoda-ku, Tokyo Official Website: https://jizaie.co.jp/ note: https://note.com/jizaie X (formerly Twitter): https://twitter.com/jizaie Corporate Movie: https://youtu.be/cnRYRXzXdWI

Contact for this matter

Contact: Jizaie Inc. Public Relations

Email: pr@jizaie.co.jp