Skip to content

Training and evaluating models to correct 3D (hand) poses using PyTorch.

License

Notifications You must be signed in to change notification settings

TheFloe1995/correct-pose

Repository files navigation

Hand Pose Correction with PyTorch

This repository contains code for my Master's thesis: Correcting 3D Hand Poses by Learning a Structural Prior with Graph Neural Networks

It's a PyTorch implementation, tested with version PyTorch 1.10 and Python 3.7. Required packages:

  • PyTorch
  • Scipy
  • Matplotlib

How to get started

Training scripts are located in run. Make sure to have the datasets available in the required format in data. More details and scripts for generating data are located in data_utils. Try to run the minimal example script first and check it out in detail.

To make sure everything is working on your machine I recommend to run the unit tests in unit_tests using pytest.

After getting familiar with the data format and the training scripts, have a look at the files in training.

After executing an experiment, use the evaluation Jupyter Notebook to get some insights, plots and visualizations.

About

Training and evaluating models to correct 3D (hand) poses using PyTorch.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published