README.md

   

Paper

Quaternion-Based Gesture Recognition Using Wireless Wearable Motion Capture Sensors

Dataset

Codes

Codes for my Master's Thesis

Data Extraction and Conversion

Replace Missing Values.py - Replaces missing values for the Beta angle after conversion from quaternion to Euclidean.

add_headers.py - adds appropriate header columns to each dataset

columnSorter.py - breaks quaternion datasets in four different datasets that hold individual quaternion components

columnSorter_euclid.py - same as above but for datasets with Euclidean components

columnSorter_pStudy.py - same as above but for individual participant dataset

columnJoiner.py - joins every individual quaternion dataset to create a new dataset with homogenous quaternion component

convert2euclidean.py - convert quaternion components to Euclidean components

outlierRemover.py - removes outliers by applying linear interpolation using a 10-point sliding window

Data Paritioning (not Train, Validation, Test)

sortLeftRight.py - separates the left and right-hand gestures and turns them into individual datasets

*Training, Validation and Test sets were created using Weka 3.6

Feature Extraction

featureExtraction.py - Extracts five features from every dataset: Variance, Range, Velocity, Angular Velocity, Covariance

Test Scripts

test.py, test2.py, test3.py, testFileSize.py, test_covariance.py, test_range.py, test_variance.py, test_velocity.py *Each file has its own description

Miscellaneous

countDatapoints.py - counts the total number of datapoints in a dataset

Data Preprocessing/Model Evaluation/Dimensionality Reduction/Feature Selection/Result Analysis

were done in Weka 3.6