For scenes where you want to analyze individuals' actions while considering privacy and not identifying them!
VisionPose is an SDK (Software Development Kit) for a posture estimation AI engine that can detect human skeletal information in real-time from camera footage, equipped with the ability to analyze and display only the skeletal information of individuals captured by the camera. ■ Utilization in "Care and Nursing" Considerations for personal space, where respect for privacy is essential, and scenes where one person is being monitored inevitably occur in care and nursing. For example, if an accident occurs in places like restrooms or bathrooms, a delayed discovery could lead to fatal incidents. In such cases, displaying only skeletal information allows for judgment of the situation while respecting privacy, enabling safe monitoring. Additionally, it can help detect unintended outings, such as wandering by the elderly, without infringing on personal information, thus reducing the likelihood of wandering. ■ Utilization in "Crowd Flow Analysis" Crowd flow analysis can also gather detailed information while protecting privacy. For instance, responses to outdoor advertisements can be collected from skeletal movements, allowing for the estimation of advertising effectiveness. Furthermore, in stay surveys, it is possible to gather data on people's conditions. By analyzing the number of people within the analysis range and individual or group attributes, insights can be gained. The collected attributes and behavior patterns are significantly beneficial in stay surveys.
Inquire About This Product
basic information
■System Features - Capable of real-time detection of 30 key points on the human body at up to 60 FPS (varies with PC specifications) - Supports 3D inference with two cameras (Standard) and 3D inference with one camera (Single3D) - Comes standard with two immediately usable applications - No restrictions on usage, including commercial use - Provided as an SDK, allowing development in C#/C++/Swift - Additional training is possible (additional training is an optional service and has conditions for implementation) ■PC Operating Environment Operating OS: Windows 10 (64bit) / Windows 11 Operating PC: - CPU: Core i7-6700 or equivalent or higher - Memory: 16GB or more - GPU: Recommended operating environment: NVIDIA GeForce RTX 3060 or equivalent performance or higher Required operating environment: NVIDIA GPU with a minimum of 6GB VRAM - CUDA: CUDA-compatible NVIDIA graphics driver (451.22 or later) *For the development environment for iOS/iPadOS, Android, and other details, please refer to the product's website.
Price information
■VisionPose Standard Edition 【Purchase Price】Development License: 398,000 yen (excluding tax) 【Purchase Price】Release License: 30,000 yen (excluding tax) A DVD containing the software/manual/other accompanying files will be mailed. *The product key will be issued separately and sent. ■iOS/iPadOS Version, Android Version 【Purchase Price】Development License: 498,000 yen (excluding tax) The software/manual/other accompanying files will be provided via data transfer. *Includes unlimited release license *The product key will be issued separately and sent.
Price range
P3
Delivery Time
P3
※Custom development and additional training are available upon consultation.
Applications/Examples of results
Do you have any of these concerns? ■ I want to do motion capture, but it seems too difficult. ■ Similar products are too expensive for me to afford. ■ There are usage restrictions that prevent me from using it. ■ I want an alternative to cameras with depth sensors. ■ I want to capture specific movements through motion capture. In industries such as manufacturing, healthcare, entertainment, and sports, posture estimation and motion capture through skeletal analysis have a wide range of applications. When you contact us, we will respond to your requests based on our achievements and examples. Additionally, Next System has numerous development achievements and products in AI and XR (AR/VR/MR), so we can offer various approaches beyond this product. Please feel free to reach out to us. ■ Next System Website https://www.next-system.com ■ Product List https://www.next-system.com/product/ ■ Development Cases https://www.next-system.com/solution/ ■ Contact Information https://www.next-system.com/contact/
Detailed information
-
VisionPose is a domestic AI posture estimation engine that enables skeletal detection through deep learning inference and analysis (PoseEstimation). The analyzed skeletal information can be applied in various fields and uses, realizing new services and functions. Additionally, there are no usage restrictions for this product, and it can be used permanently with a one-time purchase.
-
Pose Estimation is an AI technology that estimates the posture of individuals from videos and images without markers. VisionPose Single3D utilizes technology to estimate 3D coordinates from 2D images, allowing for the detection of up to 30 key points in 2D coordinates and up to 17 key points in 3D coordinates using a single camera to capture human skeletal information. It can be used as an alternative to depth sensor-equipped cameras for motion capture.
-
VisionPose detects a total of 30 skeletal points by default. Additionally, VisionPose provides an annotation tool, allowing for accuracy improvement through additional training on areas where detection is weak, as well as the ability to add skeletal points, detect only hand points, and apply it to animals.
-
This is our proprietary edge AI solution for posture estimation that operates on edge devices. In traditional IoT development, it was common to process AI's vast computational tasks in the cloud via the internet or to store acquired information in the cloud. However, this reliance on cloud communication can compromise real-time performance and introduce security risks associated with storing data in the cloud. In response, edge computing has emerged as a technology gaining attention alongside the evolution of hardware. Recent advancements in hardware have enabled the rapid processing of AI's extensive computations directly on devices in the field. It is compatible with NVIDIA's compact edge devices designed for AI deep learning, including "Jetson AGX Xavier," "Jetson Xavier NX," and "Jetson Nano."
-
Without attaching markers or anything to the body, it is possible to detect human movement in real-time from camera footage. Therefore, it can be used as a simple motion capture for applications such as avatar control in games and smartphones.
-
It can be used for posture assessment and movement evaluation in fields such as sports.
-
It can be used for evaluating the work of factory operators and detecting abnormal work (incorrect work).
catalog（10）
Download All Catalogs
Company information
Under the corporate vision of "Turning dreams into reality and creating the future," we are developing cutting-edge systems and providing services by integrating the latest technologies (AI, XR) with content (3DCG, Movies, Music). Our proprietary posture analysis AI engine, "VisionPose," has already been provided to over 400 companies in Japan and is utilized for sports motion analysis, rehabilitation gait analysis, work analysis of factory workers, and motion capture in the entertainment sector. (Examples of implementation) - Provided VisionPose to Toyota Motor Corporation for their walking assist robot "Welwalk WW-2000." - Provided VisionPose to Avex for their dance skill scoring smartphone app "Dance COMMUNE." Additionally, we can perform time-series behavior analysis and anomaly behavior analysis by incorporating time-series analysis to analyze a series of movements. Furthermore, it is applied to work load analysis through ergonomic evaluation. Our AR signage "Kinesys" has been implemented in over 100 companies in Japan and is used for virtual fitting and interaction services.