S

Synthpose Vitpose Huge Hf

Developed by stanfordmimi
SynthPose is a keypoint detection model based on the VitPose huge backbone network, fine-tuned with synthetic data to predict 52 human keypoints, suitable for kinematic analysis.
Downloads 1,320
Release Time : 1/10/2025

Model Overview

This model employs the VitPose huge backbone network, fine-tuned with synthetic data, capable of predicting 52 anatomical landmarks including COCO keypoints, particularly suited for motion capture and biomechanical analysis scenarios.

Model Features

Dense Keypoint Prediction
Capable of predicting 52 anatomical landmarks, including 17 standard COCO keypoints and 35 additional keypoints for biomechanical analysis.
Synthetic Data Fine-tuning
Fine-tuned with synthetic data on a pre-trained model, improving prediction accuracy for specific keypoint sets.
Two-stage Detection Pipeline
First detects human bounding boxes, then predicts keypoints, enhancing detection accuracy.

Model Capabilities

Human Keypoint Detection
Kinematic Analysis
Biomechanical Landmark Prediction
Multi-person Pose Estimation

Use Cases

Motion Capture
Sports Biomechanics Analysis
Used to analyze athletes' movement postures, providing precise joint angles and motion trajectory data.
Outputs accurate positions of 52 anatomical landmarks.
Medical Rehabilitation
Rehabilitation Training Monitoring
Monitors changes in patients' movement postures during rehabilitation training.
Provides detailed joint motion data for therapeutic evaluation.
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase