dastratakos / face-mask-detection

A machine learning model to identify if a person is wearing a face mask or not and if the face mask is worn properly.

Face Mask Detection

Charles Pan, Gilbert Rosal, and Dean Stratakos

November 18, 2020

A machine learning model for a CS 229 final project
Read the report » · Request a feature »

Table of Contents

• About the Project
• Getting Started
• Usage
• License
• Contact

About The Project

Across the globe, there have been 55.6M million reported coronavirus cases (up from 33.8M last month). Covid-19 has plunged countless nations into chaos and recession as they scramble to keep the virus contained. Due to the highly contagious nature of the virus, every individual must do their part in preventing the spread by taking precautions such as wearing a face mask. Yet there are still many individuals who refuse to do so - this careless behavior puts many lives at risk, so it is imperative that we hold these individuals responsible.

In light of this issue, our project aims to create a machine learning model that can accurately detect, given an image, whether a person is properly wearing a face mask or not. This project will especially be important in the global return to work effort as businesses continue to search for ways to keep their employees and customers safe. Automating the process of face mask detection will reduce human labor while creating a system of accountability.

Getting Started

To get a local copy up and running follow these simple steps.

1. Clone the repo

git clone https://github.com/dastratakos/Face-Mask-Detection.git

2. Install packages

pip install -r requirements.txt

Usage

1. Run the pipeline and pass in the model to run. Note that this pipeline will run the data preprocessing if it has not been done yet. For example:

python run_pipeline.py -m SVM

2. For ResNet models, it is recommended to use a VM through a platform such as GCP (Google Cloud Platform).

License

Distributed under the Apache 2.0 License. See LICENSE for more information.

Contact

Charles Pan, Gilbert Rosal, and Dean Stratakos - {cpan22, rosalg, dstratak}@stanford.edu

Project Link: https://github.com/dastratakos/Face-Mask-Detection

face-mask-detection's Issues

Add a clean up module to rename 'maksssksksss' and reduce XML files

Support Vector Machine

Implement vanilla SVM.

Logistic regression

Implement vanilla logistic regression.
Input is a single vector of length (height * width) of the input image.
Output is a binary classification: mask or no mask.

Command Line Arguments

Add support for various command line arguments, and update config.py or some other file accordingly.

Data augmentation

Increase the size of the input dataset using the following operations:

• horizontal flip
• vertical flip
• rotation
• color grading
• cropping individuals

Analyze different evaluation metrics

• precision
• recall
• F1-score
• make a confusion matrix

Data preprocessing and cleaning

• Extract annotation information from XML files.
• Transform PNG images into PyTorch inputs.

Update run_pipeline.py so that models can be run through that script.