Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)

IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

Analysis of Traffic Management System using

YOLOv8
2023 2nd International Conference on Automation, Computing and Renewable Systems (ICACRS) | 979-8-3503-4023-5/23/$31.00 ©2023 IEEE | DOI: 10.1109/ICACRS58579.2023.10405118

Karnatakam Anvitha, [Link] Kandregula Vanaja,

UG Student, Associate Professor, UG Student,
Department of Computer Science and Department of Computer Science and Department of Computer Science and
Engineering, Kalasalingam Academy of Engineering, Kalasalingam Academy of Engineering, Kalasalingam Academy of
Research and Education, Tamil Nadu, Research and Education, Tamil Nadu, Research and Education, Tamil Nadu,
India India India
anvithakarnatakam@[Link] suchandika@[Link] Vanaja.kandregula22@[Link]

Kapa Bhavana,
Karanam Sai sharan
K V S N Sai Kishan, UG Student, UG Student,
UG Student, Department of Computer Science and Department of Computer Science and
Dept of Computer Science and Engineering, Kalasalingam Academy Engineering, Kalasalingam Academy of
Engineering of Research and Education, Tamil Research and Education, Tamil Nadu,
Kalasalingam Academy of
Nadu India
Research and Education, Tamil Nadu,
kapabhavana263@[Link] karanamsaisharan@[Link]
India.
kvsnsaikishan2004@[Link]

Abstract- The identification of traffic violations plays a pivotal

role in contemporary efforts to manage traffic effectively and
enhance safety on the roads. Traffic violation detection holds II. RELATED WORK
immense significance due to its profound influence on road Wang [Link][1] proposed a video-based traffic violation
safety, traffic control, and the overall welfare of communities. detection system based on OpenCV and C++ techniques, that
It plays a pivotal role in molding cities that are both sustainable utilizes video streams from various cameras as input data.
and adaptable, ultimately guaranteeing an enhanced quality of With the aid of open CV background detection is done
life for both their inhabitants and guests. The proposed model
smoothly, but its detecting rate is lowered is its limitation.
elevates traffic enforcement and surveillance through its
Guoyu [Link][2] presented a real-time traffic violation detection
capability to ascertain the volume of vehicles traversing a road
simultaneously by affixing a label denoting the vehicle's count system, known as "Real-time vehicular traffic violation
and its category. Furthermore, it discerns instances of vehicles detection in traffic monitoring stream." which utilizes parallel
breaching red traffic signals, meticulously logging the vehicle's computing techniques and processes video feeds from various
count number, thereby streamlining the process of accessing cameras. Notably, it excels in low-light conditions, but it
comprehensive vehicle information. Our model detects the suffers from a drawback of relatively slow computational
multiple vehicles at a time which violates the rules and yields speed. Sipan [Link][3] introduced a Vehicle Detection and
97% classification accuracy. tracking system using Machine Learning techniques,
specifically Support Vector Machines (SVM) and
Keywords - Traffic Management System, YOLOv8, Violation Convolutional Neural Networks (CNN) with various features
monitoring, Automated Recognition like flexibility, specificity, resilience, transparency, and user-
friendliness. Based on the integration it predicts fewer objects.
[Link] [Link][4] proposed a vehicle detection system with
Haar-like features for extracting vehicle shape characteristics.
I. INTRODUCTION With the artificial neural network, it classifies the vvehicle in
The significance of traffic management and surveillance is various angles. Govindsamy [Link] [5] tried to improve traffic
growing rapidly, with implications for urban development and sign violation detection based on YOLOv3 to automate and
public safety alike. To avoid accidents and ensure a smooth expedite the process of detection and assists the police in
flow of traffic, traffic regulations must be efficiently promptly addressing violations and closely monitoring
monitored and enforced paramount. Amidst the backdrop of vehicle activity to prevent accidents. Salna joy [Link] [6]
escalating urbanization and a growing number of vehicles on utilized YOLO for continuous traffic monitoring to detect
the road, the effective and precise identification of traffic helmetless and triple-riding two-wheeler violations, aiding
violations stands as a crucial element in contemporary traffic
accident prevention. Rajanand [Link][7] used YOLOv3 with
management and law enforcement. Its central role in fostering
road safety and guaranteeing compliance with traffic computer vision to detect the new vehicles in the congested
regulations cannot be overstated. In crowded metropolitan roads and a spike in traffic violations leading to accidents. It
areas, this is difficult to monitor so an aautomating traffic also detects violations such as signal jumping, speeding and
monitoring and traffic infraction detection requires the use of vehicle count. Cheng-Jian Lin [Link] [8] used a intelligent
cutting-edge computer vision and deep learning technologies traffic-monitoring system integrates YOLO for vehicle
such as SORT(Simple Online and Realtime Tracking), detection and a convolutional fuzzy neural network for
YOLOv8 (You Only Look Once) to achieve efficient real- classification and counting the vehicles. This system is well
time multiple vehicle recognition and tracking through suited for real-time vehicle classification and counting in real
surveillance cameras. time settings, all while maintaining a rapid detection speed.
Yeresime Suresh [Link][9] proposed a solution to utilize image

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 1363

Authorized licensed use limited to: American International University Bangladesh. Downloaded on November 28,2024 at 04:36:50 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)
IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

processing to efficiently detect major violations such as over list, it is added, and the line color changes to green indicate a
speeding, helmet non-compliance, and automated number violation. It also stores vehicle ID in a file which helps to find
plate recognition, significantly enhancing traffic and trace out the vehicle easily. Based on SORT algorithm
management. multiple vehicles are detected accurate in the crossing
Based on the existing work the following issues are observed. [Link] flow diagram for our proposed work is shown in
Figure 1.
• Incorrect detecting the vehicle beyond the boundary.
• Time Taken for detecting the vehicle is high
• Training dataset are less for validation
Input Video

III. PROPOSED WORK

The proposed model consists of three phases: Vehicle Tracking
[Link]-time multiple vehicles recognition Initialization
[Link]-Time Tracking- with Vehicle-Specific
Class Labels
[Link] of red-light violation
Assign Labels and Unique
ID for each vehicle
A. Real-time multiple vehicles recognition
In the initial phase the proposed model performs real-time
recognition of multiple vehicles. This task involves leveraging
advanced computer vision and deep learning techniques such
as YOLO v8(You Only Look Once) to detect and identify Yes
various types of vehicles, such as cars, trucks and more, in Red Storing the
real-time video streams or image sequences. The integration light violated vehicle
and fine-tuning of YOLOv8 algorithm to recognize and violation ID
classify a wide variety of vehicles in real-time video streams. detectio
This phase sets the foundation for the subsequent phases by
providing the initial identification of vehicles within the video No
frames.
Vehicle Tracking Initialization
B. Real-Time Tracking- with Vehicle-Specific Class
Labels Figure 1: Flow Diagram of our proposed work
Building upon the initial vehicle recognition phase, this phase D. Performace Measures
performs real-time tracking with vehicle-specific class labels.
To detect the type of vehicles based on the training dataset
The SORT (Simple Online and Realtime Tracking) algorithm
and need to ensure the correct vehicle detection ensure the
is a crucial component in this multiple-object tracking phase.
efficiency of object detection based on the following
To forecast and revise item placements in video frames,
parameters. Based on Vehicle detection accuracy is
SORT employs Kalman filtering. Apply continuous loop on
measured. In the existing methods detection of vehicles takes
video frames and YOLO object detection to get bounding box
time and inaccurate, to ensure accurate detection YOLO 8 is
detections. Update the SORT tracker with these detections.
used for vehicle detection and its accuracy is superior to other
The each tracked vehicle is associated with its class label such
YOLO versions. [10-15]. To ensure the detection the
as car, truck, etc. An individual ID is given to each object that
following metrics is computed.
is recognized for tracking.
𝑄𝑇𝑃
𝑇𝑃𝑅 = (1)
C. Detection of red light violation 𝑄𝑇𝑜𝑡𝑎𝑙
In this final phase, the system detects red light violations
committed by vehicles. The system needs to recognize the Where TP represents the true positive which detects the
traffic signal at an intersection and determine when it turns vehicle correctly and Total represents the total number of
red. This can be done using Computer vision technique to vehicles used for detection
analyze the signal's state. Once the traffic signal turns red, the
system analyzes the behavior of tracked vehicles. It looks for
vehicles that continue to cross the intersection after the signal IV. RESULT ANALYIS
has turned red, which is considered a red-light violation.
When a redlight violation is detected, the system can trigger The proposed model is about traffic management system
alerts such as Maintain a list of total Count, to track objects which follows real-time tracking of multiple vehicles with
that have triggered a violation. If an object's ID is not in the specific labels and assigns a unique Id along with red light

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 1364

Authorized licensed use limited to: American International University Bangladesh. Downloaded on November 28,2024 at 04:36:50 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)
IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

violation detection. The unified system analyses video data,

applies the SORT algorithm for object tracking, and provides Upon detecting a red-light violation, the system has the
object IDs and locations for each frame to support capability to initiate alerts, including displaying the count of
surveillance applications. Sample output screenshot with vehicles that have disregarded the signal, as well as storing
labelled colours aids the car detection with the limit is shown the vehicle IDs in a file. This proactive approach assists in
in Figure 2. swiftly locating and tracking the respective vehicles,
simplifying the identification process. Values which are
stored with vehicle id in database sample is shown in Figure
5

Figure 2: Vehicle detection

With information like class labels, confidence ratings, and Figure 5: storing violated Vehicle ID in database
accurate item placements inside video frames, YOLO v8 is
used to recognise objects like cars, trucks, and motorbikes. This model is adaptable and have uses in traffic flow analysis,
Figure 3 represents the unique ID for each vehicle. surveillance, and the detection of traffic violations. Together,
this model provides a thorough solution for monitoring,
evaluating, and controlling different traffic scenario elements
with the potential to increase safety and security. Our
proposed model yields 97% of accurate detection which is far
better than existing techniques like RCNN which contains
92% and YOLO yields 95% [16],17] and its comparison chart
is shown in Figure 5.

Comparison with Existing Work

98 97
Figure 3: Assigning class label and Unique ID for each 96 95
vehicle.
94
92
The proposed model additionally identifies red light 92
infractions by utilizing image processing and computer 90
vision methods to recognize the traffic signal's red phase at
intersections. This model finds its most valuable application 88
in enhancing traffic surveillance, conducting traffic flow Accuracy
analysis, and pinpointing violations, collectively
RCNN YOLO 4 YOLO 8
contributing to the advancement of traffic scenario
assessment is shown in Figure 4

V. CONCLUSION
The proposed model offers a comprehensive solution for
overseeing, assessing, and managing various components
within traffic scenarios. By seamlessly integrating real-time
multiple vehicle recognition, vehicle-specific class labelling,
and the detection of redlight violations, it establishes a
powerful framework for enhancing safety and security on the
roads. The model not only enables efficient traffic
surveillance but also facilitates in-depth traffic flow analysis
and the enforcement of critical regulations. Its potential
Figure 4: Redlight Violation Detection impact extends to promoting safer road behaviour, improving

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 1365

Authorized licensed use limited to: American International University Bangladesh. Downloaded on November 28,2024 at 04:36:50 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)
IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

congestion management, and providing valuable data for

traffic authorities. In essence, it serves as a multifaceted tool
for optimizing the safety and efficiency of our transportation
systems.

VI. REFERENCES

[1] Wang, Xiaoling, Li-Min Meng, Biaobiao Zhang, Junjie Lu, and K-L.
Du. "A video-based traffic violation detection system." In Proceedings
2013 International Conference on Mechatronic Sciences, Electric
Engineering and Computer pp. 1191-1194. 2013
[2] Ou, Guoyu, Yang Gao, and Ying Liu. "Real-time vehicular traffic
violation detection in traffic monitoring stream." In 2012 IEEE
International Conferences on Web Intelligence and Intelligent Agent
Technology, vol. 3, pp. 15-19, 2012.
[3] Dimililer, Kamil, Yoney Kirsal Ever, and Sipan Masoud Mustafa.
"Vehicle detection and tracking using machine learning techniques." In
International Conference on Theory and Application of Soft
Computing, Computing with Words and Perceptions,Springer, Cham,
pp. 373-381, 2019
[4] A. Mohamed, A. Issam, B. Mohamed, and B. Abdellatif, ‘‘Real-time
detection of vehicles using the Haar-like features and artificial neuron
networks,’’ Proc. Comput. Sci., vol. 73, pp. 24–31, Jan. 2015
[5] V, Govindasamy and V, Badmapriya and M, Poonguzhali and S,
Sanjay Kanth and V, Akila, Vehicle Detection and Tracking for Traffic
Violation Using YOLO (May 25, 2021). Proceedings of the
International Conference on Smart Data Intelligence (ICSMDI 2021
[6] B. C. R, S. Joy, U. A. Reddy, R. C. Lal, V. R and A. D. M, "Traffic
Rule Violation Recognition for Two Wheeler using YOLO
Algorithm," 2023 Second International Conference on Electronics and
Renewable Systems (ICEARS), 2023, pp. 1477-1480,2023
[7] [S. Raj Anand, Naveen Kilari, D. Udaya Suriya Raj Kumar and Sai
Nitisha Tadiboina, “Traffic Signal Violation Detection using Artificial
Intelligence and Deep Learning”, International Journal of Advanced
Research in Engineering and Technology (IJARET), Vol,12,No,2, pp.
207-217,2021
[8] C.J. Lin and J. Y. Jhang, "Intelligent Traffic-Monitoring System Based
on YOLO and Convolutional Fuzzy Neural Networks," in IEEE
Access, Vol. 10, pp. 14120-14133, 2022
[9] Yeresime, Suresh & Anusha, Chillara.. TRAFFIC RULES
VIOLATION DETECTION SYSTEM.
10.13140/RG.2.2.10612.42884.2022
[10] C. Zhao, P. Mei, S. Xu, Y. Li and Y. Feng, "Performance Evaluation of
Visual Object Detection and Tracking Algorithms Used in Remote
Photoplethysmography," 2019 IEEE/CVF International Conference on
Computer Vision Workshop (ICCVW), Seoul, Korea (South), 2019, pp.
1646-1655
[11] M. Rabah et al,, "Heterogeneous Parallelization for Object Detection
and Tracking in UAVs," in IEEE Access, vol. 8, pp. 42784-42793,
2020
[12] Mukremin Ozkul, Ilir Capuni,” Police-less multiparty traffic violation
detection and reporting system with privacy preservation”, IET
Intelligent Transport Systems, Vol. 12 No. 5, pp. 351-358,2018
[13] M. Cremer and M. Papageorgiou, “Parameter identification for a traffic
flow model,” Automatica, vol. 17, no. 6, pp. 837–843, 1981.
[14] G. Dimitrakopoulos and P. Demestichas, “Intelligent transportation
systems,” IEEE Vehicular Technology Magazine, vol. 5, no. 1, pp. 77–
84, 2010.
[15] [Link] , Z. Chen, P. Li,[Link] and R Chen , ‘Real Time Non-Motor
Vehicle Violation Detection in real time Non- Motor Vehicle Violation
Detection in traffic scenecs , Proceedings - IEEE International
Conference on Industrial Cyber Physical Systems, ICPS, 2019
[16] A. Dodia and S. Kumar, "A Comparison of YOLO Based Vehicle
Detection Algorithms," 2023 International Conference on Artificial
Intelligence and Applications (ICAIA) Alliance Technology
Conference (ATCON-1), Bangalore, India, 2023, pp. 1-6,
[17] Arora, N., Kumar, Y., Karkra, R. et al. Automatic vehicle detection
system in different environment conditions using fast R-
CNN. Multimed Tools Appl 81, 18715–18735 (2022).

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 1366

Authorized licensed use limited to: American International University Bangladesh. Downloaded on November 28,2024 at 04:36:50 UTC from IEEE Xplore. Restrictions apply.