One example of a GitHub repository for face mask detection using Convolutional Neural Networks is "chandrikadeb7/Face-Mask-Detection." This repository provides a Python implementation using TensorFlow and Keras for detecting face masks in images.
- get the image with a face
- apply the cascade classifire
- give me the rigion on instrect of the face, it give face location and face height and width.
- take the face and pass it to a pre train CNN
- it will give output probability of with the mask or without mask.
The dataset consisted of 1376 images, 690 face images with masks and 686 without masks. The original dataset is prepared by Prajna Bhandary and available at Github
https://github.com/prajnasb/observations/tree/master/experiements/data
this dataset use for CNN training.
- Convert image color in to gray scale
- Need comon size - so resize the images to 100 x 100 px
- this has two convolutional layers.
- input 100x100 image for this one.
- in here having 200 cornals 3x3 the first convolutional layer.
- 2nd layer 100 and 3x3 cornals.
- that convolutions I am going to Flatten
- next connect with dense layer with 50 neurons
- Final dense layer will be output layer. it has two neurons, with and without mask.
- input image through a cascade classifire
- Detect the face
- crop the image
- Convert in to 100 x 100 image
- And passed my trained CNN
- wich predict weather with mask or without mask.