OpenPose: Real-time multi-person keypoint detection library
Multi-person 135-keypoint anatomical detection in C++.
Learn more about OpenPose
OpenPose is a C++ library for detecting human pose keypoints in images and video sequences. It uses deep learning models built on Caffe to identify anatomical landmarks across the body, hands, face, and feet simultaneously. The library processes multiple people in a single image with runtime that remains constant relative to the number of detected individuals for body and foot keypoints. Common applications include motion capture, human behavior analysis, fitness tracking, and interactive systems.
Constant Runtime Detection
Body and foot keypoint detection time stays constant regardless of the number of people in the frame. Processing scales with image resolution, not person count, enabling predictable performance for multi-person scenarios.
135-Point Multi-Region Tracking
Detects body (25 points), face (70 landmarks), hands, and feet (6 points) simultaneously in a single forward pass. Eliminates the need for separate detection pipelines or model chaining across anatomical regions.
3D Reconstruction Pipeline
Extends 2D keypoint detection to full 3D pose reconstruction with multi-camera support. Includes Unity plugin for real-time integration into game engines and interactive applications without custom bridge code.
const openpose = require('openpose');
const params = {
model_folder: './models/',
net_resolution: '656x368'
};
const detector = new openpose.OpenPose(params);
const image = openpose.loadImage('person.jpg');
const keypoints = detector.forward(image);
console.log('Detected', keypoints.length, 'people');See how people are using OpenPose
Top in AI & ML
Related Repositories
Discover similar tools and frameworks used by developers
DeepSeek Coder
Code language models (1B-33B parameters) supporting completion and infilling across 80+ languages.
Crush
LLM-powered coding agent with LSP and MCP integration.
DINOv2
PyTorch vision transformers pretrained on 142M unlabeled images.
Ultralytics YOLO
PyTorch library for YOLO-based real-time computer vision.
AutoGPT
Block-based visual editor for autonomous AI agents.