Mask_Detection

This is a DeepLearning project which will detect whether a person present in given image wore mask or not.

Dataset

Datset used in this Project consists of two folders (yes,no). Yes folder consists of 1188 jpg images of person wearning mask. No folder consists of 1200 jpg images of person not wearing mask. Dataset can be downloaded from the link given in dataset.txt file.

Code

• The images downloaded are preprocessed and converted to batches (size:32) dataset. Data Augumentation is also applied as the available data is less for the Deep Learning model.
• Best parameters(Dropout,optimizer) are selected using Tensorboard visualization.(To know more about Tensorboard visit TensorBoard.
• Project is done in python using Tensorflow ML library (To know more about TensorFlow visit TensorFlow.

Visualization

Train data visualization using Tensorboard

Example of Epoch run with the help of Tensorboard hparams

Parameters visualisation using Tensorboard

Accuracy,loss variation graph

Predictions Visualisation

State of image in various steps of model

Train images

Unseen images

Image at various filters of given convolution layer