11 IV April 2023

https://doi.org/10.22214/ijraset.2023.50858
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue IV Apr 2023- Available at www.ijraset.com

Sighted-Helmet Detection and E-Challan

Application
Ritesh Mahajan1, Anish Patil2, Saheb Singh Sandhu3, Om Telang4, Niranjan Samudre5
1, 2, 3, 4
Students, 5Assistant Professor, Electronics, Atharva College of Engineering, Mumbai, India

Abstract: Helmet detection is a technology that uses computer vision algorithms to automatically detect and identify individuals
who are not wearing helmets while riding a two-wheeled vehicle. This technology is often used in conjunction with e-challan
systems, which are electronic systems for issuing traffic violations and fines. By combining helmet detection with e-challan, law
enforcement officials can more effectively enforce helmet laws and reduce the number of injuries and deaths caused by head
injuries in motorcycle accidents. The technology can be used in various forms like CCTV cameras, drones, etc. The Helmet
detection algorithm uses deep learning techniques like CNN, YOLOv8 etc to detect and classify the helmets in real-time video
feed. We have developed an application “Sighted”, which is used to track the current active challans issued in the name of the
owner and easy payment of the fines. The main objective of the system is to identify riders without helmet and issue a challan
with an application providing services to track the challan status and payment.
Keywords: Helmet Detection, E-Challan, CCTV, Deep Learning, CNN, YOLOv8.

I. INTRODUCTION
Helmet detection and e-challan are two related technologies that aim to improve road safety and enforce traffic laws. Helmet
detection uses computer vision and machine learning algorithms to automatically detect individuals who are not wearing helmets
while riding a two-wheeled vehicle. The e-challan system, on the other hand, is an electronic system for issuing and managing
traffic fines. When integrated, the helmet detection system can automatically generate and issue fines to individuals who are
detected as not wearing helmets, helping to enforce traffic laws and reduce the number of accidents caused by individuals not
wearing helmets.

II. LITERATURE SURVEY

Rising cases of accidental deaths due to not wearing the helmet are a serious concern and this concern should be solved by
tightening the laws and imposing heavy fines on the ones disobeying it. To punish these rule offenders using the latest technology
there were multiple approaches made to detect the helmets. L Allamki et al. [1] and his fellow researchers used yolov3 for detection
of five classes, helmet, no helmet, person, motorbike, license plate. All these classes are identified using a single trained weight.
And detection is run on frames of video. If no helmet classes are detected then it searches for its person classes and motorbike class
and finally detects numberplate, crops it and then finally saves it.
But all this testing is done in sparsely populated riding environment. And according to us it is accuracy will decrease if it is done in
a densely populated riding environment as there will be multiple detections overlapping each other. To tackle this issue what we did
is that we divided the detection process in 3 parts first rider will be detected from video and its cropped image will be saved then
another mode will detect the helmet if present else it will be saved also another mode will descry the helmet if present additional it'll
further process the rider image to coming model that will descry numberplate and crop it and eventually it'll be transferred forward
for E-Challan generation.
Researcher C. Vishnu et al. [2] used technologies like CNN, Gaussian mixture model. Their view was to detect motorcyclist without
helmet in videos using CNN. They approached this problem by first separating the moving objects. Next they used CNN to find
difference between motorcyclists and non-motorcyclists. Next they did head localization; this was done by cropping the
motorcyclist’s image one fourth from the top and then using CNN to separate without-helmet and with-helmet.

III. IMPLEMENTATION
While implementing our system we mainly divided into two parts, Part I and Part II. Part I majorly comprises of Detection Part i.e.
Helmet Detection and Number Plate Detection and Extraction. Whereas Part II consists of payment of E-challan after generation of
automatic E-Challan they are as follows:

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3313
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue IV Apr 2023- Available at www.ijraset.com

A. Part I.
1) Installation Of Cameras: The first step is to install cameras at strategic locations on the roads, such as at intersections, toll
booths, and other high-traffic areas.
2) Development Of Computer Vision And Machine Learning Algorithms: This involves using deep learning technique such as
CNN. (CNNs) to train the system on a dataset of images and videos of individuals wearing and not wearing helmets.

B. Part 2.
1) Integration With E-Challan System: Once the helmet detection system is developed, it can be integrated with an e-challan
system. Here we are sending SMS to the offender who is violating the RTO rules, linking the system to a database of
individuals who have been issued fines for not wearing helmets, and automating the process of issuing fines.
2) Deployment And Maintenance: After the system is developed and integrated, it can be deployed on the roads, and regular
maintenance and updates will be required to ensure that the system continues to function properly.
3) Monitoring And Reporting: Regular monitoring is required to ensure that the system is working correctly and to identify any
areas for improvement.

IV. PROPOSED FLOW OF SYSTEM

Fig1. Flow chart

V. METHODOLOGY
1) Data Collection: The first step is to collect a large dataset of images and videos of individuals wearing and not wearing
helmets. This dataset is used to train the computer vision and machine learning algorithms that will be used for detection.
2) Algorithm Development: The next step is to develop the computer vision and machine learning algorithms that will be used for
detection. This typically involves using deep learning techniques such as convolutional neural networks (CNNs) to train the
system on the dataset of images and videos. The goal is to create an algorithm that can accurately detect individuals who are not
wearing helmets in real-time.
3) Model Validation: Once the algorithm is developed, it is validated using a separate dataset, the goal is to ensure that the
algorithm can accurately detect individuals not wearing helmets with a high degree of accuracy.

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3314
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue IV Apr 2023- Available at www.ijraset.com

VI. RESULT
In this project firstly the system must classify between motorcycles and other vehicles. Then the system must check whether the
rider is wearing helmet or not. Depending on the result, if the rider is found without helmet then the number plate will be extracted.
The contact number of the rider will be extracted from the database and an SMS of E-Challan will be sent. The SMS will include
app link through which the vehicle owner can easily track his E-Challans and can pay allotted fines.
Now we have provided the Sighted App link to the violator and they can login and pay their e-challan by using our app.

Fig2. Object Detection

Fig4. Cropped License Plate

Fig5. Extracted License Plate Character

Fig3. License Plate Detection

Here Violator can see their previous e-challans and vehicle information by clicking on my e-challans and payment status. By
clicking on Pay e-challan here violator will be directed on payment page. Here they can see challan no, date, time, fine amount,
evidence image and pay their E-challans. By clicking on ‘Pay Now’ payment status on database will get changed to paid from
pending and database will get stored. After successful payment a popup message will say Thank you!

Fig6. High Fidelity Prototype of Application

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3315
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue IV Apr 2023- Available at www.ijraset.com

Fig7. Autogenerated SMS

VII. CONCLUSION
The following proposed system helps the traffic department to detect multiple bike riders and pillion riders who are not wearing
helmet from traffic surveillance videos. This is done by passing the surveillance video through multiple Machine Learning Models
where the final output will the automated detection and generation of e-challan. The system also helps to automate the process of E-
challan generation by carrying out the detection, identification and penalization process all under one project with each model
interconnected with other, making the process swift and convenient.
Our project has also aimed to reduce corruptions and unethical practices conducted on the roads with a goal to include the everyone
under the same spectrum and to abide by the rules set up by the government.

VIII. FUTURE SCOPE

1) We are planning to add linguistic OCR detection. As in some vehicles there are custom number plates in regional language so
we would modify the model to detect regional language as well.
2) Find the best way to integrate the model with the local RTO’s traffic cameras without decreasing the accuracy. By accessing the
RTO’s database we can implement the model for every registered rider. No rider will be excluded from the fine.
3) In advancement other traffic violations such as speed monitoring on highways which will reduce accidents. In India most of the
riders do triple riding on two wheelers which is against the laws. So we can configure our model to detect it as well.
4) Riders using phones while driving which can be a reason for accidents can also be fined. Stopping your vehicle on zebra
crossing or breaking a traffic signal can also be monitored. Monitor whether your parked vehicle is obstructing the road which
may lead to traffic.
5) Riders driving on the wrong side of the road can be also fined. Repeated violation may lead to cancellation of driving license.
If any pedestrian sees any rider violating the traffic law can report the incident through our app.

IX. ACKNOWLEDGEMENT
We would like to express our sincere thanks to our guide Prof. Niranjan Samudre for taking time from his/her busy schedule to
provide us with great deal of help, support and encourage us to work diligently at every aspect of our project. His/her views have
always been equitable striking a perfect balance between encouragement and constructive criticism. His/her constructive tips and
suggestions helped us to successfully do the project. We have benefited a lot from his/her immense knowledge and experience.
We express our sincere gratitude and thankfulness to Dr. Ramesh Kulkarni (Principal ACE), Dr. Bhavin Shah (H.O.D.), Department
of Electronics Engineering (ACE) and all Staff Members of Department of Electronics Engineering (ACE) who have guided,
motivated and supported us for our Mini Project.

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3316
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue IV Apr 2023- Available at www.ijraset.com

We are thankful to the Project Coordinators, Department of Electronics Engineering (ACE), Prof. Rupali Tirale & Dr. Bhavin Shah,
for their valuable suggestions and advice. Our project at various stages has entailed us to seek help from a variety of individuals, we
would like to thank each one of them for their forbiddance and guidance.

REFERENCES
[1] L Allamki, M Panchakshari, A Sateesha,and K S Pratheek, “Helmet Detection using Machine Learning and Automatic License Plate Recognition”,Int.
Research J of Technology and Research(IRJET), Vol: 6, Issue: 12, Dec 2019.
[2] C. Vishnu, D Singh, C. K Mohan ,and S Babu,“Detection of Motorcyclists without Helmet in Videos using Convolutional Neural Network”, Institute of
Electrical and Electronics Engineers(IEEE), July 2017.
[3] Madhuchhanda Dasgupta, Oishila Bandyopadhyay, Sanjay Chatterji, “Automated Helmet Detection for Multiple Motorcycle Riders using CNN”,, Institute of
Electrical and Electronics Engineers(IEEE) Dec 2019.
[4] Google Vision API, Access date: Oct 2021, Online [Available]: https://cloud.google.com/vision
[5] Darknet framework, Access date: March 2021, Online [Available]: http://pjreddie.com/darknet/
[6] "DETECTION OF HELMET USING YOLOV4 AND GENERATION OF AN E-CHALLAN", International Journal of Emerging Technologies and Innovative
Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.9, Issue 4, page no. ppc477-c482, April-2022

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3317