International Journal of Advances in Engineering and Management (IJAEM)

Volume 4, Issue 6 June 2022, pp: 1411-1414 www.ijaem.net ISSN: 2395-5252

“Crowd Size Estimation Using Raspberry

Pi”
Mehraj Pathan1, Pratiksha Patil2, Ashwini Doke3, J.S. Shinde4.
1
Student-ENTC-Mehraj Pathan
2
Student-ENTC-Pratiksha Patil
3
Student-ENTC-Ashwini Doke
--------------------------------------------------------------------------------------------------------------------------------------
Submitted: 05-06-2022 Revised: 17-06-2022 Accepted: 20-06-2022
---------------------------------------------------------------------------------------------------------------------------
ABSTRACT current situation, COVID-the 19 crowd detection
Crowd size estimation is an awkward situation. cameras can help in isolating the people.
Especially when the crowd is spread over a In this paper, we process the people
significant geographical area. In this time when counting system where the Raspberry pi along PI
COVID-19 is expanding speedily, it is imperative camera will be mounted on the ceiling at the
to maintain social distance and avoid large public entrance of the religious place, railway station,
convocation at one place to fragment the chain of gardens, etc. This will help to detect the number of
corona infection but maintaining this is not easy. people enteringa particular place and also track the
Many people knowingly and unknowingly, gather place. This model works with real-timeimages
and roam the street. Keeping an eye fixed on these captured from the area under observation using the
activities is not a simple job. The authority raspberry pi camera module. This study focuses on
neeneedsuthfull technology to continue track of capturing the image in which it is placed and then
these activities. Our project crowd size estimation detects the people and then detects the people in the
can help to detect the number of people in the image. The most direct technique to count each
crowd. This system uses a raspberry pi with an RPI person in the crowd is the method using crowd size
camera for capturing the frame. After this frame estimation. We can take the image using a PI
will be compared with OpenCV’s pre-trained camera and count the people in the crowd using
model for people detection. The people counting OPENCV’S software. With this technique, it is
will be displayed on the Thingspeak channel which possible to easily count the crowds in any place.
can be monitored from anywhere in the world. So, The obvious improvement of this technique is that
we have decided to make a crowd size estimation it solves the problem of overlapping so that a high-
using raspberry pi that will keep a watchful eye on density crowd can be estimated by this method.
all illegal activities and detect any crowd on the
road or in any area. II. LITERATURE REVIEW:
Keyword: It is mainly used in real life for automated
Hardware Component: Raspberry Pi, Pi camera Hd public monitoring such as surveillance and traffic
HDMIo VGA connector.Software Tools: Opencv, control also the feature is appliedto interior image
Thingspeak detection and detailing work videos of crowd
scenes problems in computer vision. In this
I. INTRODUCTION: COVID-19 situation if the crowd is not directed
In our country historical places, religious properly it influences the safety of people in
places, gardens, railway stations, etc are crowded crowded areas like malls, and auditoriums. These
mostly tourist places are very crowded. It is days controlling the crowd properly is very salient.
difficult to count people in this crowd. Crowd Sometimes, the human eye cannot remark
counting or crowd estimating is a methodology all the cameras at the same time. Therefore an
used to count or estimate the number of people in a automated technique must be used for long-
crowd estimating crowd numbers, evaluating crowd duration Histogram Oriented Object Descriptor. In
density, and understanding safety limits for event this proposed work we are using a HOG detector
capacity is an essential requirements for crowd and image processing for object detection. The
safety and many people knowingly and technique counts the phenomenon of gradient
unknowingly gather on any place keeping an eye orientation in the localized portion of the image.
on all these activities is not an easy job. In our The HOG descriptor focuses on the structure of an

DOI: 10.35629/5252-040614111414 Impact Factor value 7.429 | ISO 9001: 2008 Certified Journal Page 1411
 International Journal of Advances in Engineering and Management (IJAEM)
Volume 4, Issue 6 June 2022, pp: 1411-1414 www.ijaem.net ISSN: 2395-5252

object. It is better than any border description as it working on crowd size estimation using raspberry
uses magnitude as well as the angle of the gradient pi. In that, we calculate the number of people in a
to compute the features. Some other researchers crowded area.
have done their work on being present of person- In our proposed system, firstly capturing
based part analysis some researchershave the image using a g pi camera module. This
performed crowd detection using a density. capturing the frames and this frame will be a
procedure with HOG (Histogram Oriented
III. METHODOLOGY: ObjecDescriptoror ) descriptor to detect the object
we proposed the work of this paper in that captured frame. Afterward, these captured
present in this section crowd rowdyis mainly frames will be compared with OpenCV, this model
mentionedin the population count of a certain for human being detection. The people counting
place. The Over suspect work is to develop a will be shown on the Thingspeak channel which
detection of crowd counting using raspberry pi. can be observed.
People counting is s problematic task so We are

IV. BLOCK DIAGRAM OF PROPOSED SYSTEM:

Working Principle: Afterward,theseframes will be compared with

The working of this project is, first of all, OpenCV's model for people detection.
the components required for the projectwe are used The people counting will be shown on the
to raspberry pi and pi camera. In that pi camera Thingspeak channel which can be monitored from
module is used for continuously capturing the anywhere in the world. We can show the output of
frames and then the timeframes are processed with this project on the Thingspeak channel as a graph
HOG (Histogram Oriented Object Descriptor) to as well as a numeric count.
detect the objects in the image.

DOI: 10.35629/5252-040614111414 Impact Factor value 7.429 | ISO 9001: 2008 Certified Journal Page 1412
 International Journal of Advances in Engineering and Management (IJAEM)
Volume 4, Issue 6 June 2022, pp: 1411-1414 www.ijaem.net ISSN: 2395-5252

V. RESULT:

VI. FUTURE SCOPE : [3]. Crowd Pedestrian Counting Considering

This will be able to detect the crowd in public Network Flow Constraints in Videos-
places for avoiding covid19. J.Wang.Crowd safety: A Real Time System
 In further we will be able to maintain the For Counting People — Dr.Arthanariee.
Stamped across the country. Realization of Multiple Human Head
 It can be used in Army to detect enemy Detection and Direction Movement Using
movements even in dark by some modification Raspberry Pi- S.Syed Ameer Abbas, M.
in the visuals of the camera. Anitha.
 It can be further modified into robot surveying [4]. K. Terada, D. Yoshida, S. Oe, and J.
for the auction of booms and other things Yamaguchi, A method of counting the
which can be watched from far distances. passing people by using the stereo images,
International conference on image
VII. CONCLUSION: processing, 0-7803-54672,1999
[5]. Haritaoglu and M. Flickner, Detection and
 In this project we have to detect or counted the
tracking of shopping groups in stores,
number of people in the crowd.
Proceedings of th01 IEEE Computer Society
 it is easy the use this project which is crowd
Conference on Computer Vision and Pattern
size estimation using Raspberry Pi.
Recognition, 0-7695-1272-0,2001.
[6]. Gary Conrad and Richard Johnsonbaugh, A
REFERENCES: real-time people counter, Proceedings of the
[1]. Crowd analysis and Its application by Nilam 1994 ACM symposium on Applied
Nur Amir Sharif, Siti Mariyam Shamsuddin- computing, 0-89791-647-6,1994[ROS94]:
Johar. M. Rossi and A. Bozzoli,Tracking and
[2]. A Study on Crowd Detection and Density Counting Moving People, IEEE Proc. of Int.
Analysis for Safety Control by Archana Conf. Image Processing, 1994.
Ghotkar- Pune institute of computer [7]. Daniel Hernåndez-Sosa, Modesto
technology. Castri116n-Santana, Javier Lorenzo-

DOI: 10.35629/5252-040614111414 Impact Factor value 7.429 | ISO 9001: 2008 Certified Journal Page 1413
 International Journal of Advances in Engineering and Management (IJAEM)
Volume 4, Issue 6 June 2022, pp: 1411-1414 www.ijaem.net ISSN: 2395-5252

Navarro, "Multi-sensor People

Counting," Iberian Conference on Pattern
Recognition and Image Analysis, pp. 321-
328, 2011.
[8]. Guo J, Wu X, Cao T, Yu S, and Xu Y,
"Crowd Density Estimation via Markov
Random Field (MRF)," Proceedings of 8th
World Congress on Intelligent Control and
Automation, pp. 258-263,2010.

DOI: 10.35629/5252-040614111414 Impact Factor value 7.429 | ISO 9001: 2008 Certified Journal Page 1414