Studying Relationships and Differences Between Different Heart Beat Signals and Data

This study's objective is to explore the feasibility of translating heart rate information between two different modalities, namely video and audio, using deep learning techniques. Additionally, it also aims to study the ability to predict heart rate from both video (using both deep learning and direct analysis) and heart rate associated features.

Important Links

Timesheet	Slack channel	Project report

Video/demo/GIF

Table of Contents

1. Demo

2. Installation

3. Reproducing this project

1. Example demo

What to find where

Here is where you can find important files

project_07
├── src                                                 ## Source code of the package itself 
    ├── amazing                                         ## Utility functions and code to train models
        ├── functions.py                                ## Function to calculate HR from video, Utility functions for GUI +processing data
        ├── train_predict_heart_rate_from_feature.py    ## Code to train model to predict heart rate from features
        ├── train_predict_heart_rate_from_video.py      ## Code to train model to predict heart rate from video
    ├── pretrained_models                               ## Pre-trained models for use in run.y
    ├── sample_training_data                            ## Sample dataset for training the models
    ├── run.py                                          ## Main run file to start up the GUI
├── sample_videos_to_predict                            ## Sample videos that can be used for prediction in the GUI
├── README.md                                           ## You are here
├── requirements.txt                                    ## Python modules required

2. Installation

Python 3.11

git clone https://github.com/sfu-cmpt340/project_07
cd project_07
pip install -r .\requirements.txt
python src/run.py

3. Reproduction

To recreate/train the models:

1. Download TrainingData.zip
2. Extract TrainingData.zip into project_07/src
3. In the project_07 directory run the following command:

python src/amazing/train_predict_heart_rate_from_video.py

4. In the project_07 directory run the following command:

python src/amazing/train_predict_heart_rate_from_feature.py

5. Close out any produced figures to continue the training

To use the GUI:

1. In the project_07 directory run the following command to start up the GUI

python src/run.py

2. Choose a desired prediction method
3. Upload a sample video file from the "sample_videos_to_predict" folder or fill out the feature form
4. Click "Predict My Current Heart Rate Range!" to get a prediction
5. The video prediction result is displayed

(Note: When using option "Predict Heart rate from a video (without DL), make sure to close the figure to see the result)