A group project for CSCI 1430 Introduction to Computer Vision, Spring 2021
Despite the significant vaccination process, people will still need to wear masks for a long time when they are outdoors. Yet currently many people are not wearing masks in proper ways, and most of them are unconscious of this. So in this project, a web app with CV models running on backend is developed to help people check if they are wearing masks correctly.
Visit the app at coveryournose.xyz.
Scripts for downloading the big mask / no mask dataset we created.
A collection of HAAR cascade classifiers and scripts to run them.
Three different models mask detection: Faster R-CNN, SSD, and Mobilenet.
Ensemble of a HAAR-based nose detector an a pretrained SSD model for mask detection.
A local copy of the architecture of this model can be found here.
A Flask server application with Nose+MaskDetector running on the backend, ready to deploy to cloud with Docker. Together with a tensorflow MNIST classifer server for rapid prototyping.