motion-analysis

PyTorch implementation of our graph convolutional network (GCN) for human motion generation from music. Also with paired dance-music data for training!

The source code and pre-trained models for Motion Matters: Neural Motion Transfer for Better Camera Physiological Sensing (WACV 2024, Oral).

(CVPR 2023) MarS3D: A Plug-and-Play Motion-Aware Model for Semantic Segmentation on Multi-Scan 3D Point Clouds

Extract and visualize gait data

CoachMe: A generative instruction model based on motion analysis for sports.

[ECCV 2026] Official implementation of "Language-Guided Transformer Tokenizer for Human Motion Generation."

Official implementation of the paper "MoSa: Motion Generation with Scalable Autoregressive Modelling."

Code and dataset of human retargeting approach to transfer motion and appearance between monocular videos. Check the examples from our dataset with several paired actors and motions!

A project that create virtual cycling simulator

Official implementation of AAAI 2023 Oral Paper "Frame-Level Label Refinement for Skeleton-Based Weakly-Supervised Action Recognition"

NintAi is a professional-grade bio-mechanical analysis tool that turns your computer into a high-end bike fitting studio. Inspired by industry leaders like MyVeloFit, NintAi uses YOLOv11 for state-of-the-art tracking and Google Gemini 3 (AI) to provide expert-level written reports

Python interface of C3Dserver software for reading and editing C3D motion capture files

Natural Coordinates with python for biomechanics

Usage of direct linear transformation (camera calibration and point reconstruction) for accurate motion analysis in biomechanical applications.

🏌️ Golf swing analysis using MediaPipe Pose & OpenCV for motion tracking, posture evaluation, and performance improvement.

GapFill: gap filling functions for motion capture marker data

Computer vision system that analyzes sports footage to provide real-time performance metrics, injury prevention insights, and tactical recommendations.

In-depth motion analysis of mobile lung cancer tumors. Designed for 4D-CT scans of the thorax and provide valuable information for proton therapy treatment planning

Compute causal relationships between individual pixels in 2D videos over space and time to reveal salient dynamics using a variety of causal measures

This repository contains Python code used for the extraction and clustering of muscle and kinematic synergies from preprocessed EMG data, as described in the research paper . The data was initially imported from Vicon c3d files.