MyoVerse - The AI toolkit for myocontrol research

What is MyoVerse?

MyoVerse is your cutting-edge research companion for unlocking the secrets hidden within biomechanical data! It's specifically designed for exploring the complex interplay between electromyography (EMG) signals, kinematics (movement), and kinetics (forces).

Leveraging the power of PyTorch and PyTorch Lightning, MyoVerse provides a comprehensive suite of tools, including:

• Data loaders and preprocessing filters tailored for biomechanical signals.
• Peer-reviewed AI models and components for analysis and prediction tasks.
• Essential utilities to streamline the research workflow.

Whether you're predicting movement from muscle activity, analyzing forces during motion, or developing novel AI approaches for biomechanical challenges, MyoVerse aims to accelerate your research journey.

Installation

MyoVerse automatically installs with the correct PyTorch version for your platform.

Basic installation:

# Install from PyPI
pip install myoverse

This will automatically:

• On Linux: Install PyTorch and TorchVision from PyPI (with CUDA support)
• On Windows: Install PyTorch and TorchVision with CUDA 12.4 support

Development

For development, install the dev dependencies:

1. Clone the Repository:

   git clone https://github.com/NsquaredLab/MyoVerse.git # Replace with your actual repo URL if different
   cd MyoVerse
   

2. Install uv: If you don't have it yet, install uv. Follow the instructions on the uv GitHub page.
3. Set up Virtual Environment & Install Dependencies: Simply run:

   uv sync --group dev
   

What is what?

This project uses the following structure:

• myoverse: This is the main package. It contains:
  • datasets: Contains data loaders, dataset creators, and a wide array of filters to preprocess your biomechanical data (e.g., EMG, kinematics).
  • models: Contains all AI models and their components, ready for training and evaluation.
  • utils: Various utilities to support data handling, model training, and analysis.
• docs: Contains the source files for the documentation.
• examples: Contains practical examples demonstrating how to use the package, including tutorials (01_tutorials) and specific use cases like applying filters (02_filters).
• tests: Contains tests to ensure package integrity and correctness.