You are on page 1of 8

Face Detection and

Recognition
Student Attendance
System
Chapter 1

INTRODUCTION

These days, face detection and facial recognition technologies are crucial, and
we've seen that they're being used in a variety of places, including cellphones, the
military, and some high-risk information offices. In order to replace the traditional paper
attendance system and the finger print attendance system, we chose to create a gadget
that can detect and recognize faces. LabVIEW will be used for the primary function in
our project because it is a highly useful programming tool for facial applications as well
as other applications. Our idea is built on a core LabVIEW program that recognizes and
The subsystems also include an Excel sheet that is integrated with the application and a
messaging device that can be used to send messages to missing students' parents or to
recognize faces with the aid of scoring and parameter settings. Our project consists of
the LabVIEW program as the primary system and its subsystems, an Office Excel sheet
to contain the names of the students, and a computer (or laptop) to connect the
programs.

Design of an automatic class attendance system utilizing the LabVIEW facial detection
algorithm. For the system to be effectively constructed, a video capture device and an
active LabVIEW algorithm are required. It recognizes the faces and records attendance
as necessary. This approach will stop the needless loss of class time that typically
occurs during roll calls.
OBJECTIVES

Objectives of the Project

Face recognition has been around since the 1960s and 1970s, and after decades of
changes and adjustments, it has reached maturity. The structural qualities of the face
and the color characteristics of the face are mostly used in the traditional face detection
method. Some conventional face recognition algorithms extract landmarks to identify
facial traits, or features taken from the subject's face in a photograph. An algorithm
might examine, for instance, the relative placement, size, and/or form of the eyes, nose,
cheekbones, and jaw, as shown in Figure 1.1. The next step is to utilize these traits to
look for similar features in other photos. the algorithms of this type may be difficult,
demand a lot of computing resources, and thus perform slowly. Then they might be
misleading when faces display obvious emotional responses because of the size and
position in such a situation of the landmarks can be drastically changed.

Figure 1.1 Abstract humane face into features [1]


1. Minimizing time wasted while attending traditional classes.
2. Using the most recent developments in machine vision to put into practice a workable
system for tracking student attendance.
3. Creating a digital environment by automating the entire process.
4. It is only possible to prevent fraudulent roll calls by using one-to-one attendance
marking.
5. Promoting the regular usage of technology.
SIGNIFICANCE OF THE STUDY

In Chapter 1, we reviewed the application scenarios and promising future of


facial recognition technology. Many significant businesses and colleges have invested a
great deal of time and energy in this sector. The most important research in the field of
facial recognition will be discussed in the first section of this chapter.

The most commonly referenced piece in a collection of articles by Viola that actually
makes face detection practical is this one, Robust Real-time Object Detection [2]. From
this paper, we can learn about several face detection techniques and algorithms. The
article Fast rotation invariant multi-view face detection based on real Adaboost [3]
proposed a more developed and practical multi-face detection framework, and the nest
structure mentioned on the cascade structure improvements also had good results. This
was the first time real Adaboost was applied to object detection. A strong combination of
face detection and tracking, offline model and online model, was used in the article
Tracking in Low Frame Rate Video: A Cascade Particle Filter with Discriminative
Observers of Different Life Spans [4], which won the CVPR 2007 award Best Student
Paper.

The three publications mentioned above examined the issues with face tracking and
face detection. We can create real-time face detection systems in accordance with the
study findings presented in these papers. Finding the location and size of each face in
the photo or video is the primary goal, but for identifying the relationship between the
various faces in the frame is essential for tracking.
OPERATIONAL DEFINITION OF TERMS

Computer - A machine that performs tasks, such as calculations or electronic


communication, under the control of a set of instructions called programs.

Software - It is a non-tangible part of computer system. The programs and other


operating information used by a computer

System - It is any collection of components elements that work together to perform


tasks.

R.A.D (Rapid App Development) - It was designed and created to determine and ensure
implementation of a system.

Database - Any collection of data organized for storage in a computer memory and
designed for easy access by authorized users.

Debugging - It is to find and remove errors in a system, especially a computer program


or device.

Inventory - It is the act or an instance for recording the inventory related tools and
equipment’s or the process of being organizing the inventory system.
CHAPTER 3

TOOLS AND MATERIALS

This system will make use of its own database to store data. In this post, we'll
utilize the MySql relational database management system as an example. Using blobs
is an obvious method of storing the human face image in the database (binary large
object). However, if the image is saved in the database, the GPU will have to extract the
vector from the image again each time it performs the face comparison. images. There
will be a lengthy wait for large datasets. We will first extract the vectors from the human
face image and just save the vectors into the database to avoid these problems.
Additionally, we must assign the vector an attribute for identification. Customer id,
employee id, and user id are all viable options for this identifying characteristic.
therefore, based on how many vectors will the model extract, the main table structure in
the database will look like below:

ID Vector1 Vector2 Vector3 … … … Vector<n-1> Vector<n>

The database for this system shouldn't keep anything other than the ID and face vector,
in accordance with database normalization requirements. Any additional columns would
be unnecessary and require extra work to preserve data integrity because many user
details can change quickly inside an organization. With merely an ID column, the
database may communicate and retrieve data to other enterprise databases like ERP,
Clarify, and others, as shown in figure below.
One modification to the database design suggested by our experiment is to avoid
making the ID column a unique key or primary key column. The performance can be
greatly enhanced by doing this. For instance, you might gather a face with ID=9 10
times with marginally different vectors after the system has been operational for 10
days. All ten of these vectors can be saved into this primary table. When a person with
ID=10 passes in front of the camera on the eleventh day, the picture is processed, and
the top three candidates from the processing engine will display ID=10 more than once.
The outcome will give the customer additional assurance to support the claim that the
person is the one with ID=10.

You might also like