Face mask detector model

Classifies images of people as wearing face mask or not wearing face mask

Model:

Trained using fastai resnet34

Dataset:

The dataset consists of 1050 images of faces- 506 images with mask and 544 images without mask. The dataset was prepared by web-scraping 894 images and 156 images in the with mask category were created by overlaying a face mask on the image of a front-facing face, using opencv.

Future work:

Extend detection to video input

Include bounding boxes in prediction