Utilization for work checks and accident prevention at factories and other worksites

Detect abnormalities from the flow and movement of people to prevent human-caused accidents and shoplifting in stores

For example, in factory where many machines are operated, there is always a risk, and it is not easy to completely eliminate human errors in places with repetitive tasks. In such cases, checking the flow and movements of people with VisionPose and learning from them can be useful for improving quality and risk management.

Furthermore, it is also possible to incorporate VisionPose into surveillance cameras. Normally, static data cannot distinguish between ‘sleeping’ and ‘fallen’ postures, but using time-series data, one can infer the condition of a person experiencing sudden illness based on causality, such as ‘standing posture → lying posture = fallen’. Additionally, it can be utilized for detecting suspicious movements and checking individuals at unmanned cash registers.

VP area anomaly detection application

This is a feature enhancing application exclusively for i-PRO AI network cameras, capable of alerting when specific parts of the body enter designated areas. It enables the detection of abnormalities in detailed body parts, which was challenging with conventional surveillance and security cameras. It can be utilized for preventing unforeseen accidents and enhancing security measures.
Application Details(Japanese)

VP single person work detection application

This is an application capable of detecting when a person is working alone, despite the requirement for multiple individuals in areas such as manufacturing sites, server rooms, pharmaceutical factories, vaults, and warehouses to prevent incidents and accidents. It can be utilized for unforeseen accidents and security measures. It is sold as one of the feature enhancing software for i-PRO AI network cameras developed by i-PRO Corporation.
Application Details(Japanese)

Example of customization specialized for fingertip detection

This is a video showcasing VisionPose’s fingertip specialized model. We are developing this model to analyze the detailed movements of the hands in various scenes.
The number of inferred joints and accuracy are still under development.

Analysis of screw tightening operation

In this demo video, we utilized the ‘Image-Based Behavior Analysis’ feature of ‘VP-Motion,’ is based on VisionPose, to classify tasks based on screw tightening. We analyzed both normal and abnormal tasks, allowing viewers to observe how judgments are made on the analysis screen.
More details about VP-Motion

VisionPose’s Usage in facial recognition software Bio-IDiom KAOATO by NEC Solution Innovators

Case Study Details

Back to Use case

Download PDF