TO DETECT THE HUMAN FROM IN & OUT OF THE

CLASSROOM WITH IDENTITY BY USING IOT TECHNIQUE

PRESENTED BY

Vaibhav Srivastava,
Pradeep Shrivastav, Manish Sundvesha
Under Guidance of Prof. Kirti Randhe
CONTENT
● Problem Statement ● Algorithm
● Introduction ● Prerequisites
● Abstract ● Dynamics Classes
● Literature Survey ● Future Scope
● Gap Analysis ● Limitation
● Proposed System ● Conclusion
2
INTRODUCTIO
N
Many studies have shown that the attendance of
students in universities shows a falling trend
over the years. There are various factors involved
in the low attendance such as lack of interest or
lack of good-teaching skills, other extracurricular
activities like part-time jobs or availability of
online content. Therefore, our system aims to
solve the problem of lack of attendance or
management related problem.
3
 ABSTRACT
In Universities, maintaining classroom availability & scheduling time for students dynamically
is hard for respective subject, students are unable to attend classes because of classroom
availability & scheduling. If the management is done manually then it becomes a very tedious
task. Our system aims to overcome this problem by implementing algorithm which are
present in modern days such HOG (Histogram oriented gradients), Eigenfaces, fisherfaces,
CNN (Convolutional Neural Network) & etc. are one of the many algorithms used in this
modern days & our system use the HOG algorithm to get the strength of students & arrange
the classroom dynamically for students according to the respective subject, they want to
attend. Our educational system will be different than the traditional one & will be used to
identify the person / human who is entering in & out from a classroom & count total number
of human / persons present in a classroom, it will not consider any objects. This count is
stored into our database like MySQL Database which is used for storing person identity. We
use a portable device i.e., Raspberry Pi which will be adjusted according to classroom
infrastructure. The Raspberry Pi is connected to either Ethernet or College WIFI so that other
data like Total number of lectures taken in a month or semester or year, how many numbers
of lecture taken by a teacher in a month or semester or year & Attendance Sheets, it will also
be used to modify, delete, edit & update the datasets stored in a Raspberry Pi and all these
data will be accessed by the college localhost computer.
4
 LITERATURE
SNo.
1
REVIEW
Tittle
Counting students using OpenCV and Integration 10.1109/
DOI Remarks
The Google cloud platform is being used for maintaining and storing data of the
with Firebase for Classroom Allocation. ICESC48915.2020.91 respective college.
55825
2 A Review of Face Recognition Technology 10.1109/ACCESS. The face recognition technology made great achievements, but there is still room for
improvement. In near future, there might be a special camera for face recognition,
2020.3011028 which can give required image format that resolves the problem of image filtering,
image diagnosis, etc.
3 Real-time implementation of face recognition 10.1109/ICCMC. This implementation is performed by AT&T dataset of images. AT&T dataset of
system images consists of four hundred face images of forty individuals with ten images per
2017.8282685 person each having dimensions of one hundred and two * ninety-two pixels. Each
pixel consists of an eight-bit gray scale value ranging from zero to two hundred and
fifty-five. The images are taken at different times with varying in light, facial
expressions, pose, etc. The images are in png format.
4 Face Detection and Recognition System using 10.1109/ICIMIA The list of Eigen faces is restricted for use, PCA transformation in which the system
Digital Image Processing did not have an accuracy of more than ninety percent for both (automatic and
48430.2020.907483 manual) face recognition. A further studies is required to be done in the field for fully
8 automated frontal view, the face detection when displayed virtually shows a perfect
accuracy which is required.
5 Research about human face recognition 10.1109/ICTM. Face recognition is a very popular and lively research area, in this area, new ideas,
technology methods will continue to emerge. Given the effective applications of identification
2009.5412901 methods in face recognition, future research will benefit from the further improve and
perfect of face recognition; on the other hand, advanced study of face recognition will
promote the development of identification methods and updating.
5
 GAP

ANALYSIS
“Dynamic Room Allocation” has been referred to as base paper. In which they have
implemented a system for dynamic allocation of classes. In the first module, they have
compared various constraint optimization techniques like classroom assignment
problems and integer linear programming and taken various constraints like a class
constraint, course constraint, no two lectures can be scheduled in the same class at
the same time or no two teachers can be assigned the same classroom at the same
time etc. Next module is about the dynamic allocation where the model which had
been built in the first module is dynamically adapted to real-time data which is the
students present on a given day for a given lecture. They have used historical data for
training this model that is attendance from previous semesters.
 “Our system will be using the count of students attending lectures using Raspberry Pi
4B with its Camera Module and not from the traditional attendance ledger. This count
will be stored in the database and then its data will be used by college according to
their requirement.”
6
 PROPOSED
SYSTEM
Training
System Architecture (UI)

The total system is divided a) Initialize face recognizer.

into 3 modules- Database b) Get faces and Id’s from
creation, Training the database folder to train
dataset, Testing, sending the face recognizer.
alert messages as an
c) Save the trained data as
extension.
xml or yml file.

Database creation Testing

a) Initialize the camera
b) Get user id as input a) Capture the image from
camera,
c) convert the image into
gray scale, detect the b) Convert it into gray
face and scale,

d) Store it in database by c) Detect the face in it and

using given input. d) Predict the face using
 STEPS FOR COUNTING
(ALGORITHM)
• Take the input frame from the video and remove the
background.
• HOG algorithm is used to identify humans among all the
objects.
• To mark each person for keeping track of their movement in
the frame.
• By using templates provided by OpenCV for marking
attendance of students.
• Once the person cross the plane/frame (move into or exit
the classroom), make the changes to the count. Increase the
count if entering or decrease if exiting . 8
 DYNAMICS
CLASSES
The quality of technical education depends on various factors: outcome based
socially and industrially relevant curriculum, good quality motivated faculty,
teaching learning process, industry internship, evaluation of students based on
desired outcomes among others. Therefore, it was of imperative that a revised
curriculum be prepared by the institute in order to be a competitive institute
which goes to current time of change, thus some external seminar or workshop
is required for student in order to enhance their skill and meet up the required
criteria which is set or expected by the institute or university.
Our model comes in handy in such cases.
9
 PREREQUISI
TES
Hardware Requirement Software Requirement
● Raspberry Pi 4B ● Python
● Pi Module Camera ● MySQL
● Memory Card ● PHP, HTML & CSS
● OpenCV
● Scikit Learn
 FUTURE
SCOPE
Over the years, movies have fixed a futuristic fantasy in our minds that a time will come when
software would be used to recognize people by their faces. A time when our faces will be our ID
cards. With advent of facial recognition technology, that time is already here.
Today, along with drones, AI and IoT, facial recognition technology is also defining our
millennium. Facial recognition is a biometric technology used for authentication and
examination of individuals by correlating the facial features from an image with the stored facial
database. Face Recognition is one of the most popular applications of image analysis software
and no more considered as a subject of science fiction. Earlier, this technology was only used for
security and surveillance purposes, but it has safely transitioned to the real world in recent
times. Today, companies are pitching facial recognition software as the future of everything from
retail to policing.

11
 LIMITATI
ON
Despite the benefits, there are four factors that limit the effectiveness of
facial recognition technology:
1. Poor Image Quality Limits Facial Recognition & Effectiveness
2. Small Image Sizes Make Facial Recognition More Difficult
3. Different Face Angles Can Throw Off Facial Recognition & Reliability
4. Data Processing and Storage Can Limit Facial Recognition Tech

12
 CONCLUSI
ON
MySQL used for maintaining and storing data of the respective
college/University. The HOG algorithm serves as our base to identify
faces in the video (which is the count of students). This count is stored
into the MySQL database to further process and predict attendance. For
dynamic allocation of classrooms, the teacher decides which classroom
will be allotted for student according to the strength of student present
in a classroom. Also, this model can be used in identifying students who
are bunking lectures by detecting the faces of student who goes outside
before the scheduled time of lecture. Our model takes new attendance
for every new lecture.
13
THANK
YOU

14