Markerless High-Precision AI Pose Estimation Engine

About VisionPose

VisionPose is a Pose Estimation AI engine developed by NEXT-SYSTEM Co., Ltd, enabling high-precision analysis of human skeletal and pose information. This AI engine can analyze not only camera images, but also still images and videos without markers, and supports 2D/3D analysis.


  • Supports camera images, still images, and videos

    Support for camera images, still images, and videos

  • 2D and 3D coordinate analysis

    2D and 3D coordinate analysis

  • Broad platform support

    Broad platform support

  • Commercial use is possible without usage restrictions

    Commercial use OK, No usage restrictions

  • Improved accuracy by additional learning

    Accuracy improvement by additional learning

    * English Support Coming Soon

  • Detects up to 30 body keypoints

    Detects up to 30 body keypoints

  • Multi-person detection

    Multi-person detection

  • Includes ready-to-use skeleton detection tool

  • Made in Japan

    Made in Japan


VisionPose includes a set of SDK and skeletal detection tools for information analysis.
One-time purchase allows you to use the software permanently, with no usage restrictions.

VisionPose Standard

High-precision Standard SDK. 2D/ 3D analysis with 30 body keypoint detection and up to 60FPS* for the whole body.

* Varies depending on PC specs.

1 camera/ 2D analysis/ 30 keypoints

2 cameras/ 3D analysis/ 30 keypoints

* No camera required for file analysis
Windows C#,C++

Windows C#, C++

Compatible with network cameras or IP cameras.

small edge devices

Jetson AGX Xavier
Jetson Xavier NX

Support for NVIDIA small edge devices with the same precision. For IoT development that requires space-saving integration.

Select Product & License

Development License
  • You need this license to develop and evaluate the VisionPose application.
  • You can use the product free of charge during a 30-day trial period before you are being charged.
Distribution License
  • You need this license to distibute applications you have developed using VisionPose.
  • You cannot use this license for development.
  • We sell this license only to customers who have already purchased the Development license.
  • We provide a volume discount.
About the Discount Coupon
You can get a discount by entering the coupon code on the checkout page.
Click on the “
” and copy the code.
Product PlatformDevelopment LicenseDistribution License
VisionPose Standard for Windows C#
This product is for Windows 10 (64bit) only. The API provided with this product is for C# programming language. For more details on specifications and operating environment, click here.

Development License

US$3,825US$4,500 BUY

Distribution License

US$382.5US$450 BUY
VisionPose Standard for Windows C++
This product is for Windows 10 (64bit) only. The API provided with this product is for C++ programming language. For more details on specifications and operating environment, click here.

Development License

US$3,825US$4,500 BUY

Distribution License

US$382.5US$450 BUY
VisionPose Standard for
Linux, Jetson AGX Xavier, Jetson Xavier NX
Coming soon.
  • This product is available as download only. No physical product will be shipped.
  • We will electronically provide you with a license to use the VisionPose SDK, the VisionPose SDK software and related documentation, and the Product Key required to activate the VisionPose SDK. Please see the License Agreement for the terms of this license.
  • After purchase, delivery will take about 2-3 business days, due to an in-house review of your order information. If there is no problem with your information, we will send you a download URL and product key via E-Mail.
  • For any questions about our products, please visit our FAQ or contact us via this form.
  • When clicking on the “BUY” button, you will be forwarded to our shopping cart, operated by the company “Digital River”. Please continue with all payment procedures there.
  • We are using the ecommerce platform My Commerce for all purchase and payment procedures. This secure and fast platform is provided by DigitalRiver, a veteran company in ecommerce software sales.
  • For all matters related to purchase and payment, please see the MyCommerce FAQ, or contact MyCommerce directly.

VisionPose Single3D

SDK with monocular 3D analysis function. It allows you to capture motion easily without calibration.

1 camera/ 2D analysis/ 30 keypoints

1 camera/ 3D analysis/ 17 keypoints

* File analysis does not require a camera, iOS/ iPad OS cannot analyze files.
* Separate application development is required for camera control of iOS/ iPad OS.


SDK for iPhone/ iPad application development using motion capture.

Windows Unity

Windows Unity

For creating animations of CG characters and developing games in Unity.

License coming soon.

Included Skeleton Detection Tools

VisionPose includes sample programs as reference for development, and tools for skeletal detection in videos and still images.
These tools can export data in CSV format, which can be formatted and used for analysis or motion data.

VP Analyzer

Skeletal Information Detection From Videos

Pose estimation in 2D coordinates from your videos and still images, as well as visualization and output of the detected skeletal information.

Supported still image file formats
[Input] JPG, PNG, BMP
[Output] JPG, PNG
Supported video file formats
[Input] AVI, MP4, WMV, MOV
[Output] AVI


Visualizing Skeletal Information From Camera Images

The tool visualizes skeletal information in real time and displays it on the video.


Image of BodyAndColor in operation

Release Note

VisionPose Standard

  • ver1.6.3

    Windows C#, C++Linux2020.12.22Changed the library version of the execution environment (CUDA, cuDNN, TensorRT). Added GeForce RTX 30 series explanation to the manual. Added version acquisition API.

  • ver1.6.2

    Windows C#, C++Linux2020.11.25Added time stamp acquisition IF to skeleton frame. acquisition IF for C++. Added the API for reloading configuration for calibration. Bug fixes.

  • ver1.6.1

    Windows C#, C++Linux2020.11.06Enabled specification of the configuration file. Bug fixes.

  • ver1.6.0

    Windows C#, C++Linux2020.09.09Started support for TensorRT7.1/ CUDA11.0. Started support for Video streams. Changed the configuration items. Bug fixes.

  • ver1.5.3-NX

    Jetson Xavier NX2020.06.29Changed the description of the manual “6. Camera Placement”. Bug fixes.

  • ver1.5.2

    Jetson AGX Xavier2020.6.11Added CSV format for output. Bug fixes.

  • ver1.5.1

    Windows C#, C++Linux2020.02.07Changed setting items. Renamed video analysis tool to “VPAnalyzer”. Bug fixes.

  • ver1.5.0

    Windows C#, C++Linux2019.07.31Integrated C# DLL “VisionPoseWrapper.dll” into “VisionPoseCSharp.dll”. Bug fixes.

  • ver1.4.0

    Windows C#, C++Linux2019.05.27Added Windows/ C++ interface. Added Linux/ C++ interface. Added multi-thread inference function. Added multi-GPU inference function. Added Camera Index Settings. Added Cameraless mode. Bug fixes.

  • ver1.3.0

    Windows C#2019.02.27Improved analysis speed. Adjusted position of the spine (Spine Mid). Added Shooting guide to the camera calibration function. Changes in the skeleton display specifications of the sample application “BodyAndColor”. Changes in the skeleton display specifications of the file analysis tool “FileAnalyzer”. Bug fixes.

  • ver1.2.0

    Windows C#2018.02.27Added Single camera mode. Added Analysis Mode setting that emphasizes either accuracy or speed. Bug fixes.

  • ver1.1.0

    Windows C#2018.11.19Added a tool to infer video/ still image files. Bug fixes.

  • ver1.0.0

    Windows C#2018.11.01First Version.

VisionPose Single3D

  • Coming soon.


Used by more than 250 clients