You are on page 1of 6

Introduction

Our goal in this project is to develop an “Deep learning based attendance system”. As we all are
students once in our life we are very aware of how important the time is. As we all attend the class
we know our interaction with the teacher play a vital role in it. But the teacher is having further
responsibilities in addition to teaching us, attendance is a mandatory part of a class to be done.
Even if the teacher cast the roll number in pairs it is going to take a quite effective span of time
which is somewhere reducing the period of interaction between teacher and student. So what we do
for saving the time of attendance and utilizing it by increasing the time of interaction between
teacher and student is to create an application that actually help the teacher to take the attendance
of whole class in just one click at all. Deep learning based attendance system is a kind of software
that helps the teacher to take the attendance of all student by clicking the photo. The teacher just
have to click few sample photos of class for test cases as many test cases the teacher will take it
will become easier for taking attendance for the next time. The application will use the technology
known as deep learning in it. So whenever it will capture a photo it will not only count the number
of students by identifying their face it will also identify who the student is. The software will
further interact and will provide the list of all students attending the class to the teacher in the form
of a CSV or an XLS file. Now the teacher can save this in any format. And as our attendance
records are further required to be updated on system of our college and school for the evaluation of
marks also it will save the time of data entry also as the data will already be provided to the
associates in the form of CSV and XLS for marks evaluation. The main advantage of the software
is that no one in the class is able to put the proxy of other.
So as we are using the technology of deep learning so we will be using Python as the language for
the development purpose of the software. And for writing the csv and xls we will be using any file
manipulation technique of python. The “Deep learning based attendance system” will be capable
enough to get more and more accurate for every time. As the older algorithms are also able to do
the same task but as per the data increases their performance starts falling down and in some cases
they come to a state where the graph stays constant. So our task is to increase the efficiency of
system in proportion of our data. Or to simply make our system able to recognize the student in
every sort of dress up.
Objective:-

Our main objective is to increase the interaction time between students and teacher by
eliminating all the time wasted in attendance.

Feasibility Study:-

The most feasible approach we can put here is by applying the Deep learning pattern here that will
help us to identify the face of student with the help of camera. Here we can use python to create a
UI as well as to create the model of deep learning for face identification. In order to reach our
objective what we have to do is to combine the UI and model together than we need to provide test
cases to the agent in the deep learning but in case of deep learning we use the neural networks. It
will first take input of the images of all students so we have to provide different sort of images of
students to the model. In order to train it properly we must provide the model images that are taken
in different angles. As much the images are provided to the model the model will become that
much perfect in order to take the attendance more accurate. The objective was also be able to
achieve by the help of machine learning which is a kind of super set of deep leaning but, that was
not enough to make the model able to make itself better and better in every use. As it is not
necessary that every student in the class is going to sit in the same order in the class with the same
hairstyle and dress code.
Fig 1.

Fig 1.
The use of neural network is very helpful in order to deal with such
situation where data is high and can risk performance.
Methodology:-

Deep learning is a subset of machine learning which helps the machine to mimic like a human
brain. The main feature of deep learning that helps it to do so is neural networks.
Fig 2. Neural Networks

Neural network consist of input layer, hidden layer and output layer which help it to take decision.
So we have to create a model using CNN (convolutional neural network), which is also a type of
neural network but specially used for creating projects like object detection. The library which we
will use for it will be such as TensorFlow, Matplotlib and OpenCV etc.

Fig 3. Libraries

As the name of the library indicates the project will be surely develop on python platform
Bibliography
1.Understanding Machine Learning: From Theory to Algorithms Book 2014 by Shai
Shalev-Shwartz and Shai Ben-David.
2. www.openai.com : “connecting text and images , 15 January 2021.
3.Deep learning with python 2017 by François Chollet .
4.www.mygreatlearning.com :”Face Reconition with python” 18 January 2021.
5 www.viso.ai :”Deep Face Recognition: An Easy-To-Understand Overview” 27 March 2021
6. www.applexus.com :”Designing an AI-enabled Facial Recognition System: 27 August 2020

You might also like