Detection of masked/unmasked faces using CNN
realized as a mini project for the IA 02: Neural Network and Deep Learning track at GoMyCode https://gomycode.tn/
I used RFMD and LFW datasets for training my CNN model
download instructions are located in https://github.com/X-zhangyang/Real-World-Masked-Face-Dataset
The model did 99,4% accuracy on test dataset
The trained model is stored in the 'my_model_128.h5' file
-
Tensorflow (tested using version 2.3.0)
$ pip install tensorflow
-
MTCNN (multitask cascaded convolutional networks)
$ pip install mtcnn
-
opencv2
MTCNN is used to detect faces in the images which are sended to my model in order to classify them masked or unmasked
to use streaming from the webcam
$ python streamv2.py