Skip to content

Repository files navigation

Garment Inertial Denoiser: Endowing Accurate Motion Capture via Loose IMU Denoiser

This repository contains the implementation of our paper, Garment Inertial Denoiser: Endowing Accurate Motion Capture via Loose IMU Denoiser, including pretrained weights, training scripts, and evaluation code.

Paper | Project Page

  • train_per_imu.py: trains Location-Specific Denoiser expert modules, each specialized for one IMU placement.
  • train_fuse.py: trains the Cross-wear Fusion model, initialized from the retained Location-Specific Denoiser checkpoints.
  • denoise.py: generates denoised IMU sequences with the trained model.
  • eval.py: evaluates saved denoised IMU data with MAE.

Dataset

The GID dataset is available at:

Google Drive: https://drive.google.com/drive/folders/1tp6yjy3AbLiOAsud96pyHmdJqPSqFo6R?usp=sharing

Baidu Cloud: https://pan.baidu.com/s/1LDw5Z8BhQLjhqbE4RepKvg?pwd=7hap

About

Garment Inertial Denoiser: Endowing Accurate Motion Capture via Loose IMU Denoiser

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages