Helmet And Number Plate Detection Using

Yolov-3

Mohit Gupta, Naman Tyagi, Ritika Mittal, Princy, Mr. Shahid

Department of Computer Science and Engineering

Meerut Institute of Engineering and Technology, Meerut, UP, India
{[email protected], [email protected],
[email protected], [email protected] }
DOI: 10.47750/pnr.2022.13.S10.452

Our world is developing on a very fast scale. One of the reasons is Automobile industry in which motorcycle has always been
the predominant mode of transport. Since the rises of two wheelers (motorcycles, scooters) have increased so it led to accidents
and injuries. Many riders generally do not use helmets and that is one of the major reasons. It can be check physically that the
rider uses the helmet or not. Once the traffic police spot those driving motorcycles in junctions-without helmet, they also use
video from CCTV to take control of the drivers of those vehicles and penalize those who are riding without one. The software
uses You Only Look Once which consists of Convolutional Neural Network (CNN) to detect the Common Objects. CNN is a
Deep Learning algorithm which takes image as input, assign variables to various objects in the image so that it is easy to
differentiate one from the other. CNN algorithm is used for image classification and recognition due to its high accuracy. CNN
is based on a hierarchical model which works on the principle of building a network and gives out a fully connected layer
where all the neurons are inter-connected to each other, and the output is processed.

Keywords: YOLO, Convolutional Neural Network, Object, Computer Vision, Deep Learning, Helmet.

1. Introduction
a. General overview of the topic:
Intelligent monitoring has emerged as the most crucial application in recent years. Object detection is considered
as a hotspot of computer vision. As the bike riders in our country are increasing and the road misshapes are also
increasing day by day, due to which many accidents occur which led to death and most of them are caused due to
not wearing helmets. So, our model detects the helmet of the rider. If the two-wheeler rider does not wear the
helmet, it detects the number plate of the vehicle. So that the traffic police spot the rider and take the e-challan to
them. We are using YOLOV3 (You only look once version. To detect the objects, this deep learning algorithm
uses CNN (Convolutional Neural Network) that recognizes specific objects in videos, live images or feeds [1][2].

b. Prior research
Looking at the current scenario of automation and industrialization, accessibility becomes a important feature.
Today technology is used in various sectors to save time, reduce costs and give accurate results. Deep learning is
an emerging technology which is used in a variety of sectors to improve the features of existing models. It gives
the user a different environment for completing a task. In this fast-moving world we can see the number of
accidents in the last few years. Helmet wear is mandatory. Therefore, it has become vital that we develop a system
which can help in detecting the helmet and number plate earlier so that accidents can be cured, and lives can be
saved [3][4].
c. A rationale for the paper

Journal of Pharmaceutical Negative Results ¦ Volume 13 ¦ Special Issue 10 ¦ 2022 3735

Because there are so many people who don’t wear helmets on Indian roads, many head injuries result on fatalities;
thus, to prevent accidents, our technology automatically recognizes those who don’t wear helmets and recognizes
motorbike number plates to penalize that person’s networks. The rationale of our paper can be summarized- 1. To
use deep learning in safety measures [5][6].
2. To obtain data in real time.
3. To detect the helmet and number plate so that precautions can be taken immediately.
4. To process and classify the data gathered.
d. Methodology
In this paper, we have a system that detects helmets and number plates to prevent accidents. And we have used
deep learning for the same. The stages involved are:
Step-1 First upload the real time video of the riders.
Step-2 Then preprocess the video and extract the features from the image.
Step-3 Using various algorithms like YOLOV3 and other computer vision techniques [7][8].

2. Literature Review
The literature review covers the paper which is important and unique in the area of enhancement of studies
related to safety [9].
Nowadays technological development leads the development of unique machines like John Shepherd-Barron is
One of the specimens who invented ATM machine which reduces the usage of banks and is the communication
between bank and customer. We can see the large number of increases in Smart Phones. The gadget also
continuously checks the number of users and reduces the time consumed in the bank. This technology also ensures
privacy and prevents people from standing in the lines and eases deposit and withdrawal of amount [10][11].
In this fast-moving world where we have motorcycles and many different transportations, so the safety and
security of our lives are also becoming important. There are different people who propose this system. The
author made the system of two components which consist of a two-wheeler along with a helmet. Data collected
by sensors through Trigger Warning on vehicles which includes an accelerometer, a GPS module and a Hall-
effect sensor and send to the micro controller, an internet connection is there it subsequently sends the data to
the server. This system tells someone about the speed of a vehicle if anyone wanted to know the speed then they
use this technology [12][13].
P. Rojaet.al has proposed a system consisting of a remover sensor, IR sensor, Air quality sensor, Arduino Uno
microcontroller, GPRS, GSM. Helmet alert the workers regarding any inappropriate activity like harmful gases
or removing of the helmet. Here this data transmission is done using IOT technology. Takahashi et al who
introduces a system which can classify bicycles and motorcycles. In this system he uses a Gabor filter which
detects moving objects and calculates the vertical movement of the scene. If the vertical movement is pedaling
then the vehicle considers bicycles, otherwise cars, vans, buses etc. To extract the image characteristics histogram
of oriented gradients is applied. As I mentioned above in the starting lines, the objects are divided into two
wheelers and pedestrians [14][15][16].

3. Flow Chart
This system contains five phases-
a. Input video
b. Processing and Background removal
c. Segmentation of two-wheelers
d. Detection of Motorcyclist without helmet
e. Extract and display license plate number from image

Journal of Pharmaceutical Negative Results ¦ Volume 13 ¦ Special Issue 10 ¦ 2022 3736

 Figure 1.1
a. Input video- First upload the real time video of the riders.
b. Processing and Background removal – A digital image processing technique called background removal
can be used to separate an image's portions into interesting and undesirable areas. Prior to further analysis
and processing, background reduction is necessary for many applications of image processing and
computer vision [17][18].

c. Segmentation of two-wheelers- Based on vehicle type, India’s two-wheeler market has been segmented into
Motorcycle, Scooter etc.
d. Detection of Motorcyclist without Helmet – Draw a green box around bikes of those riders who are not wearing
helmets.
e. Extract and display license plate number from image – Draw a red box around number plates.

4. Data Description

The algorithm used-

Journal of Pharmaceutical Negative Results ¦ Volume 13 ¦ Special Issue 10 ¦ 2022 3737

 Fig- 1.2

Yolov3 stands for You Only Look Once Version 3. It is a deep learning algorithm which is popular due to its high
accuracy. YOLOv3 is a real time object detection algorithm which help in identification of specific objects in a
particular video, live feeds or images.

YOLO is a CNN uses in real time for object detection. CNNs are classifier-based systems which takes images as
input and recognize their patterns. It has advantage that it is faster than other algorithm and it gives high accuracy.
It scans whole image then gives the best score based on similarities. For example: During the real time traffic, it
is used to detect various kinds of vehicles mainly two wheelers like motorcycle, activa etc. YOLOv3 is a
combination of various resources which is created by other authors, and they use different algorithm for prediction.

4. Conclusion & Future Scope

In this project we built a system for automatic detection of motorcycle riders who are not wearing helmets and
their vehicles do not have number plates. Our model detect this through real time video. The algorithm and some
computer vision techniques help in attain good accuracy for detection of helmets and number plates. But, they
needed to be punished only detection is not sufficient. So, the transport officers can take better actions against the
riders. Our System detect the helmet and number plates from the real time video and gave the result in the form
of the images i.e, it marks the green square against the helmet and red square against the number plate.

6.Result

In this system it shows the people who are not wearing the helmets, so the system can detect the number plates
below figures show this

Journal of Pharmaceutical Negative Results ¦ Volume 13 ¦ Special Issue 10 ¦ 2022 3738

Figure 1.3 Green box detects the rider is wearing the helmet and detect the number plate

Fig 1.4 Red box shows the rider is not wearing the helmet and detect the number plate.

7. Reference

[1]. L. Xie, T. Ahmad, L. Jin, Y. Liu, and S. Zhang, “A new CNN-based procedure for multi-directional car license plate detection,” IEEE
Trans. Intell. Transp.Syst., vol.19,no.2,pp.507–517,Feb.2018.
[2]. Dasgupta, M., Bandyopadhyay, O., & Chatterji, S. (2019). Automated Helmet Detection for Multiple Motorcycle Riders
using CNN, 2019 IEEE Conference on Communication Technology and Information.
[3]. Image processing and segmentation techniques for vehicle's plate recognition. , 2020 IEEE 4th International Conference
on Image Processing, Applications and Systems (IPAS).
[4]. Shrivastava, S., Singh, S. K., Shrivastava, K., & Sharma, V. (2020). CNN based on Automated Vehicle Registration Number
Plate Recognition System. 2020 2nd International Conference on progress in Computing, Communication Control and Networking
(ICACCCN)
[5]. Luvizon, Diogo Carbonera; Nassu, Bogdan Tomoyuki; Minetto, Rodrigo (2016). This is a Video-Based System for Vehicle
Speed Measurement in Urban Roadways. IEEE Transactions on Intelligent Transportation Systems.

Journal of Pharmaceutical Negative Results ¦ Volume 13 ¦ Special Issue 10 ¦ 2022 3739

 [6]. SaquibNadeemHashmi, Kaushtubh Kumar, Siddhant Khandelwal, Dravit Lochan, Sangeeta Mittal,(2019) “Real Time License
Plate Recognition from the Video Streams using Deep Learning”, International Journal of Information Retrieval Research, Volume
9 Issue 1 January-March 2019.
[7]. Hendry and Rung-Ching Chen, (2019) “Automatic License Plate Recognition via sliding-window darknetYOLO deep learning”,
Image and Vision Computing.
[8]. Dariusz Laskowski,Piotr Lubkowski,(2017) “Assessment of Quality of Identification of Data in Systems of Automatic Licence
Plate Recognition” In: Mikulski J. (eds) Smart Solutions in Today’s Transport. TST 2017.
[9] New Ni Kyawf, G R Sinhaf, Khin Lay Mon, (2018) “License Plate Recognition of Myanmar on Vehicle Number Plates A Critical
Review," IEEE 7th Conference on Consumer Electronics.
[10] Suresh, Anukul Patil, Abhishek Kashyap, Ankit Jaiswal, Saksham Sharma(2018)
[11] Narayan, Vipul, et al. "To Implement a Web Page using Thread in Java." (2017)
[12] Narayan, Vipul, and A. K. Daniel. "FBCHS: Fuzzy Based Cluster Head Selection
[13] Protocol to Enhance Network Lifetime of WSN." ADCAIJ: Advances in Distributed Computing and Artificial Intelligence
Journal 11.3 (2022): 285-307.
[14] Kavitha S, Varuna S, & Ramya R. (2016). A comparative analysis on linear regression and support vector regression. 2016
Online International Conference on Green Engineering and Technologies (IC-GET). doi:10.1109/get.2016.7916627
[15] Narayan, Vipul, and A. K. Daniel. "Novel protocol for detection and optimization of overlapping coverage in wireless sensor
networks." Int. J. Eng. Adv. Technol 8 (2019).
[16] Srivastava, Swapnita, et al. "An Ensemble Learning Approach For Chronic Kidney Disease Classification." Journal of
Pharmaceutical Negative Results (2022): 2401-2409.
[17] Awasthi, Shashank, et al. "A Comparative Study of Various CAPTCHA Methods for Securing Web Pages." 2019 International
Conference on Automation, Computational and Technology Management (ICACTM). IEEE, 2019.
[18] Irfan, Daniyal, et al. "Prediction of Quality Food Sale in Mart Using the AI-Based TOR Method." Journal of Food Quality 2022
(2022).

Journal of Pharmaceutical Negative Results ¦ Volume 13 ¦ Special Issue 10 ¦ 2022 3740