You are on page 1of 9

PROJECT REPORT

INDIAN INSTITUTE OF INFORMATION TECHNOLOGY


ALLAHABAD

Social Distancing and Face Mask Detection

Faculty Supervisor: Dr.Rajat Singh

Team Members : Saumya Gupta (IEC2017047)


Shivansh Srivastava (IEC2017054)
Rohan Shekhawat (IEC2017503)
ABSTRACT
____________________________________________________________

With The Advent on Unlock of Lockdown in Country rate of spread of Covid-19


is likely to increase. People need to follow preventive measures like using mask
In public places and maintaining Social distancing with each other in order to
decrease the rate of mass spread. But sadly people are too careless to follow
these measures.
So our project aims at solving this particular problem.
INTRODUCTION
______________________________________________________________

As the daily life is coming back to normal and public places are being crowded
again there is an increasing need of strict measures to maintain preventive
solutions .
COVID-19 spreads mainly among people who are in close contact (within about 6
feet) for a prolonged period. Spread happens when an infected person coughs,
sneezes, or talks, and droplets from their mouth or nose are launched into the air
and land in the mouths or noses of people nearby.
So it is strongly recommended to wear a mask and maintain social distancing at
all times.
The Objective of the Project aims at developing a system that will help authorities
To detect weather preventive measures for Covid-19 are being followed properly
or not . We built a machine learning application that can be easily installed with
monitoring systems(CCTV’s , Surveillance Cameras etc) at various locations
which will not only detect the people not wearing mask but also if in a particular
Area social distancing is being followed or not.
Uses YOLOv3 along with DBSCAN clustering for recognizing potential intruders. A
Face Mask Classifier model (ResNet50) is trained and deployed for identifying
people not wearing a face mask. For aiding the training process, augmented
masked faces are generated (using facial landmarks) and blurring effects
(frequently found in video frames) are also imitated.
SYSTEM MODEL
________________________________________________________________

TOOLS & LIBRARIES USED

● NumPy
● Tensorflow
● Keras
● Matplotlib
● Open CV
● Face Recognition Dataset
● Google Collab
MODELS AND DATASETS USED
● YOLOv3
● ResNetv2
● Face Recognition Dataset
METHODOLOGY
________________________________________________________________

1. Face Mask Detection


2. Social Distancing Tracker
RESULT
________________________________________________________________
CONCLUSION
________________________________________________________________
The face detection technique is implemented using the ResNetv2 model in
a deep convolutional neural network, having 100 convolutional layers
followed by global average pooling and a 1000-class fc layer. We remove
the average pooling layer and the fc layer and only use convolutional layers
to compute feature maps.
Social Distancing is implemented with the help of YOLOv3(You Only Look
Once) model and DBSCAN Clustering.

Below is the Training Loss and Accuracy graph, we obtained while training
the model.

You might also like