Abstract:
The proposed system is designed to classify people who is wearing face masks
or not. The model used in this system is MobilenetV2, a convolutional neural network
(CNN). The image dataset contains 7553 images. 3832 images used to train model
and 3721 images are used for testing. Firstly, the input images are needed to be
processed. Resizing, One-hot Encoding and data Augmentation are applied in
preprocessing. The porposed system is constructed with MobilenetV2 model. If a
person is wearing a mask, the system displays the face region with a green anchor
box. If a person is not wearing a mask, the face region is displayed with a red anchor
box. This system can be merged with other applications at airports, railway stations,
workplaces, schools, and other public places for safety. The accuracy is 82 % for
testing images in dataset.
