Skip to content

tej-prash/yoga-pose-classification-system

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 

Repository files navigation

Yoga Pose Classification and Fake Pose Generation System

The repository consists of source code for a yoga pose classification system and fake pose generation system built using deep learning. The goal of the project is to build a yoga pose classification system and a fake pose generation system. In the long term, we aim to use this model to build a self-assisted yoga pose correction and training system using deep learning. Our primary motivation behind this project has been the lack of availability of a good quality dataset for the development of such a system. Consequently, a yoga pose generation system can be used to build a large, good quality dataset of fake and real poses.

Overview

We explore this task by first building a model for yoga pose classification, followed by a model for fake pose generation.

Yoga Pose Classification System

Firstly, a dataset consisting of images of four different yoga poses is built and various body key points are extracted. Thereafter, key points detected are used as features for the models. The yoga pose classification system consists of Long Short Term Memory (LSTM) cells, capable of classifying 4 different yoga poses.

Fake Pose Generation System

The fake pose generation system consists of a Variational AutoEncoder. Fake poses are generated by training a variational autoencoder and interpolating the latent space.

Software Stack

  1. Python
  2. Keras & Tensorflow
  3. sklearn
  4. Pandas,Numpy
  5. matplotlib

Dependencies

The codebase requires developers to specify the paths for the datasets. Note that the dataset hasn't been made publicly available, yet.

Team

  • Tejas Prashanth
  • Sumanth V Rao

About

A deep learning driven system for classifying yoga poses and generating fake poses

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages