Pattern Recognition Approaches for Gesture Classification

Final Project for EEL5825 in Fall 2021 by Mykola Maslych and Mostafa Aldilati

Approaches

kNN, MLP, GRU, LSTM and CNN

Models

1. KNN Approach
2. Multilayer Perceptron
3. Gater Recurrent Units
4. Long Short Term Memory
5. Convolutional Neural Network

Performances Summary

Dataset

We were not sure if we can upload the original data, so follow these steps to download it.

1. Download XML files from here under Unistroke gesture logs
   1. Unzip xml.zip
   2. Delete the pilot user (s01)
2. Run convert.py to convert the XML to our format
   kNN, GRU and LSTM approaches will work now
3. Run all cells in fully_connected_data_preparation.ipynb
   MLP approach will work now
4. Run all cells in dataset_to_bitmap.ipynb
5. Run all cells in generate_dataset_images.ipynb
   CNN approach will work now