A project investigating the measurement of the amplitude of Parkinson's hand tremors using computer vision (namely hand tracking) methods. Written for my dissertation at the University of Manchester.
Requires Mediapipe (pip install mediapipe), and OpenCV (pip install opencv-python).
The configuration file, hta_config.yaml, contains various settings which should be set before running. These are each explained in the config file. The simplest way to run the program is to set AUTO_MODE and USE_CUSTOM_LANDMARKS to False in the configuration file, then run python3 hand_tremor_amplitude.py with no command line arguments. You will then be prompted to choose a video file to measure tremor amplitude from, input the depth measurement and input any any other required information. Tremor will then be measured and a plot, alongside the tremor measurement and an error analysis, will be presented.
The program can also be used without any user prompting, saving results and plots to a CSV file. Switch AUTO_MODE to True to do this, and call the program using python3 hand_tremor_amplitude.py [video filepath] [depth in cm] ['resting' or 'postural'] [tracking landmarks]. This may be useful for automated batch processing of videos.
All of the code within this repository is a part of my dissertation "Development of a Computer Vision Method to Measure Parkinson's Tremor Amplitude", supervised by Dr. David Wong of the University of Manchester.
Copyright © 2022 James Bungay. All rights reserved. A licence for any use (of any kind) of the contents of this repository, code or otherwise, cannot be implicitly granted at this time. If reading this and you wish to use this work, please contact me as the licensing situation may change.