VisionPose | NEXT-SYSTEM Co., Ltd.

AI Human Pose Estimation
Trusted by Over 450 Japanese Companies like Toyota and Honda!

About VisionPose

VisionPose* is a high-precision AI pose estimation engine which detects multi-person skeletons
from camera images without the use of markers or depth sensors.
The skeletal data obtained from the analysis can be used for commercial purposes or for research and development
regardless of the application or field.
*VISIONPOSE is the U.S. registered trademark of NEXT-SYSTEM Co., Ltd.

Feature

Detecting 30 keypoints of the human body at up to 60 FPS in real time
3D inference with two cameras (Standard);
3D inference with a single camera (Single3D)
Two ready-to-use apps:
1) Real-time pose estimation displayed on the video (BodyAndColor) source code included
2) Pose estimation using video files or still images (VP Analyzer)
No usage restrictions, including commercial use
Delivered as SDK, and supports development through C#/C++/Swift!
Trusted by over 450 Japanese companies like Toyota and Honda!

Product Lineup

VisionPose Standard

Standard VisionPose with accuracy and performance
3D Inference Using Dual Cameras

Standard Lineup

VisionPose Single3D

Easy Pose Estimation for Windows Unity or iOS/iPadOS
3D Inference Using a Single Camera

Single3D Lineup

3D Inference

30 Keypoints
Dual Cameras

17 Keypoints
Single Camera

2D Inference

30 Keypoints
Single Camera

Platform

Windows C#
Windows C++

Linux (Ubuntu)

Jetson AGX Xavier
Jetson Xavier NX

Unity

iOS / iPadOS