As the world is modernizing, we are spending more time sitting at desks. It can lead to various other health-related problems, which can be prevented by exercising. These exercises include walking, jogging, weight lifting etc. Exercises like weight lifting can lead to multiple sprains, strains, or fractures if they are not done correctly. Learning how to do them correctly requires supervision from gym experts. But in our busy lives, we barely have time to go to the gym. So in our study, we are exploring whether we can predict if a person is exercising correctly or not by analyzing data collected from various sensors attached to the user as they exercise. Previous authors haven’t explored the deep learning approach, so in our study, we set out to find if using deep learning can be a feasible option or not for predicting if a user is correctly exercising or not. The dataset was collected from an online website and is also avaiable in Kaggle.
-
- Fully Connected Network : 72.85%
- Convolutional Neural Network : 61.22%
- Convolutional Neural Network + Long Short Term Memory : 58.3%
-
- Language : Python
- IDE : Google Colab
-
- Tensorflow
- Scikit-learn
- Numpy
- Pandas