Smart Traffic Signal System with

Violation Detection and Emergency

Response
Submitted to

University of Mumbai

Submitted in partial fulfillment of the requirements of the

degree
B.E.
(Computer Engineering)
By

Khan Mohammad Moin Abdul Moiz (23CO37)

Antuley Aman Siraj (23CO25)
Khalife Abdul Sami (23CO33)
Khan Mohd Irfan (23CO40)

Supervisor
Dr. Tabassum A. Maktum

Department of Computer Engineering

Anjuman-I-Islam’s Kalsekar Technical Campus
School of Engineering and Technology
Plot No. 2 & 3, Sector-16, Near Thana Naka, Khanda Gaon,
New Panvel, Navi Mumbai. 410206

October 2024
 Anjuman-I-Islam’s Kalsekar Technical Campus
School of Engineering and Technology
Plot No. 2 & 3, Sector-16, Near Thana Naka, Khanda Gaon,
New Panvel, Navi Mumbai. 410206

Certificate
This is to certify that, the Mini Project – 1A entitled

“Smart Traffic Signal System with Violation Detection

and Emergency Response”
is a bonafide work done by
Khan Mohammad Moin Abdul Moiz (23CO37)
Antuley Aman Siraj (23CO25)
Khalife Abdul Sami (23CO33)
Khan Mohd Irfan (23CO40)
and is submitted in the partial fulfillment of the requirement for the
degree of
Bachelor of Engineering in
Computer Engineering
Submitted to
University of Mumbai

___________________ ___________________
Supervisor Project Coordinator
(Dr. Tabassum A. Maktum) (Dr. Tabassum A. Maktum)

___________________ ___________________
Head of Department Dean (SoET)
(Prof. Tabrez Khan) (Dr. Rajendra Magar)
 Mini Project – 1A Approval

This Mini Project – 1A entitled “Smart Traffic Signal System with Violation

Detection and Emergency Response” by Khan Mohammad Moin Abdul Moiz

(23CO37), Antuley Aman Siraj (23CO25), Khalife Abdul Sami (23CO33),

Khan Mohd Irfan (23CO40) is approved in the partial fulfillment of the

requirement for the degree of B. E. in Computer Engineering.

Examiners

1……………………………………
(Internal Examiner Name & Sign)

2……………………………………
(External Examiner Name & Sign)

Date:

Place:
 Abstract
In the wake of increasing challenges from urban traffic congestion, delays in response
to emergencies, and traffic violation, the world needs smart traffic management
systems. With this in mind, this project aims at an intelligent traffic signal system that
dynamically responds to real-time traffic conditions to optimize traffic flow and
maximize safety while giving priority to emergency vehicles.

The system aims at easing congestion through dynamic signal adaptation relative to
real-time vehicle counts, eliminating bottlenecks in high-density areas and improving
the smoothness of traffic flow. Real-time violation detection is built into the system
while allowing the automatic identification of speeding vehicles and red-light
violations, with instant alerts going to law enforcement agencies. Helmet detection for
motorcyclists is also covered to enforce helmet use.

For instance, the emergency vehicles such as ambulances receive priority by the green
light of traffic lights automatically in the required direction so that response time is
increased. Another feature includes AI-based accident detection. It continuously
searches and responds to accident scenarios in real-time so that overall incident
management is improved.

This methodology uses sophisticated technologies like AI and machine learning models
for object detection like YOLO to obtain the real-time data. Sensors and cameras are
used for obtaining the real-time data, which is further stored in MongoDB. It further
utilizes the communication hardware to communicate with the traffic signal and
emergency vehicle.

The successful implementation of the Smart Traffic Signal System portrays an

improvement of managing urban traffic, especially road safety, and emergency
response, which will give the process to smarter city traffic infrastructure.

i
Contents

Abstract i

List of Figures iii

1 Introduction

1.1 Overview 1

1.2 Motivation 1

1.3 Objectives 2

1.4 Organization of Report 2

2 Literature Survey

2.1 Survey of Existing System 3

2.2 Limitations of Existing System 4

3 Proposed System

3.1 Problem Statement 5

3.2 Proposed Methodology 6

3.3 System Design 9

3.4 Details of Hardware and Software Requirements 12

4 Results and Discussion

4.1 Implementation Details 13

4.2 Results 17

5 Conclusion and Future Work 22

References 24

iii
List of Figures

3.1: System Architecture of the Smart Traffic Signal System, illustrating the 6
interaction between various components.

3.2: Traffic Signal Control Flowchart illustrating the system's real-time processes 10
including traffic monitoring, vehicle classification, violation detection, and
dynamic signal adjustment.

3.3: Main homepage of the MongoDB database displaying traffic data, violation 11
records, and system logs for efficient real-time management in the Smart Traffic
Signal System.

3.4: Graph displaying the number of violations per road in the MongoDB database, 12
visualizing traffic infractions across different intersections.

4.1: Sample Image I from the Ambulance Detection Dataset. 14

4.2: Sample Image II from the Ambulance Detection Dataset. 14

4.3: Sample Image I from the Helmet Detection Dataset. 15

4.4: Sample Image II from the Helmet Detection Dataset. 15

4.5: Sample Image I from the Accident Detection Dataset. 16

4.6: Sample Image II from the Accident Detection Dataset. 16

4.7: Traffic Signal Control Dashboard showing real-time traffic camera feed, 18
violation alert, and signal status.

4.8: Violation Display Dashboard showcasing real-time alerts for traffic violations, 18
including details of infractions and corresponding camera feeds.

4.9: Dashboard showing real-time traffic data and past 24-hour traffic data. 19

iii
4.10: Car Counting Model Output showing real-time vehicle detection, speed 19
measurement, and counting.

4.11: Accident Detection Model Output showcasing real-time identification of 20

traffic incidents, highlighting affected vehicles.

4.12: Ambulance Detection Model Output demonstrating the system's capability to 20

identify emergency vehicles in real-time.

4.13: Helmet Detection Model Output showcasing the system's effectiveness in 21

identifying motorcyclists wearing helmets.

iv
 Chapter 1
Introduction

Such problems in urban traffic systems have been bringing attention towards the basic
day-to-day life dealings of commuters. The problems such as congestion, traffic
violations, delayed responding to emergency, and hazardous driving practices require
more efficient and better traffic management solutions to deal with these problems
dynamically. The existing traffic management system is failing to handle such potential
issues dynamically that affect common people by undergoing inconvenience and posing
potential hazards.

1.1 Overview
The increasing number of vehicles on roads and inefficient traffic management
creates huge congestion and delay problems as mentioned in the article [1]. They
are unable to adjust in real-time; they usually cause multiple delays at intersections,
and with slower enforcement for emergency vehicles such as ambulances.
Furthermore, there are cases of speeding and jumping the red light during traffic,
and also the motorcyclists fail to wear helmets, which are truly grave safety
concerns. It is clear that a far more intelligent system, AI-based, which is able to
govern traffic flow, violation detection, and prioritization of emergency vehicles, is
advisable in order to uplift the present levels of safety.

1.2 Motivation
This project was motivated by real-time observations of traffic inefficiencies.
Residents of urban areas often encounter long delays due to poorly timed traffic
lights, which cannot perform according to real road conditions. Emergency
vehicles, for example, emergency vehicles, like ambulances, often get stuck in
traffic, which can have life-threatening consequences, as mentioned in the article
[2]. Lastly, there are human traffic violations and unsafe practices such as
motorcyclists not wearing helmets contributing to the increase in road accidents and
deaths. Critical issues that motivated us to find a solution that could dynamically
manage traffic flow, violations detection, and safety improvements are as follows:

1
1.3 Objectives
The primary objectives of this project are to:
o Optimize Traffic Flow: Adjust signal timings based on real-time vehicle
counts to reduce congestion.
o Detect Violations: Identify and alert authorities about speeding, red-light
jumping, and helmet violations.
o Prioritize Emergency Vehicles: Ensure faster passage for ambulances by
adjusting signals to clear their path.
o Enhance Road Safety: Implement helmet usage for roadsters.
o Accident Detection: Employ AI to detect accidents in real-time and notify
for prompt response.

1.4 Organization of Report

The report is organized as follows:

o Chapter 2 provides a literature survey, covering the existing systems for traffic
management, their features, limitations, and technologies used, along with a
survey of user requirements.
o Chapter 3 outlines the proposed system, detailing the problem statement,
proposed methodology, system design, and hardware/software requirements for
solving the identified issues.
o Chapter 4 presents the results and discussion, comparing the outcomes of the
proposed system with existing solutions, including implementation details and
project outcomes.
o Chapter 5 concludes the report with an evaluation of the project's purpose,
achievements, and future work that could enhance the solution further.

2
Chapter 2
Literature Survey

2.1 Survey of Existing System

There are several developed advanced traffic management systems worldwide

towards addressing traffic congestion, safety, and violations. Among them includes
the adaptive traffic control systems under which traffic light timings change in
accordance with the real-time data in vehicle flow: for instance, Siemens SCATS
and IBM's Smarter Traffic. This has proved to be quite efficient with a high
improvement in traffic flow- especially in the congested urban areas as mentioned
in the paper [3].

Some of the popular methods include Violation Detection Systems. This involves
using a high-resolution camera to monitor traffic behavior where red-light
violations and speeding incidents are recorded. According to the paper [4], such
technologies are already used in a wide coverage of the U.S. and Germany enforcing
rules about traffic flow End.

Predictive Traffic Systems are also high-quality solutions using historical and real-
time data to predict the traffic patterns and then taking proactive adjustments. The
paper [5] mentions how they apply data analytics in enhancing traffic management
with regards to smart cities.

While these systems possess special efficiencies and automation in most sectors,
they aren't particularly strong in real-time prioritization of an ambulance, helmet
compliance for motorcyclists, and accident detection. The growing demand for such
complex systems that answer these deficits has led to the quest for more advanced
AI-based solutions as pointed in the paper [6], which remarks the rising demand for
deep learning-based system development.

3
2.2 Limitations of Existing System
Despite the improvements provided by the above-mentioned systems, several
limitations persist:

1. Inadaptability: Most systems don't change, especially for unexpected traffic

surges or real-time emergencies like accidents or arrival of the emergency
vehicle.
2. Non-identification of Emergency Vehicles: The current solutions fail
to detect and prioritize ambulances or other emergency vehicles, thus
experiencing delays that could be critical.
3. Lack of Accident Detection: The available systems lack the actual
mechanism of real-time accidents detection, hence delays emergency response
and worsens traffic conditions.
4. Helmet Compliance: There is a lack of helmet compliance checks in very
few safety-ensuring systems for motorcyclists as an essential element of
enforcement concerning road safety.

4
Chapter 3
Proposed System

3.1 Problem Statement

The key issues to be dealt with according to the proposed Smart Traffic Signal System
are the following:

1. Dynamic Traffic Management: The system aims to adjust signal timings

based on real-time vehicle counts to alleviate congestion, especially in high-
density areas.
2. Emergency Vehicle Prioritization: The system will identify ambulances
and toggle traffic lights green in the direction of the emergency vehicle.
3. Real-time Violation Detection: The system will detect speeding
automobiles and red light violations, capture the license plates and alert law
enforcement agencies in real-time.
4. Helmet Compliance Monitoring: The system will ensure that
motorcyclists comply with helmet regulations, alerting authorities in case of
violations.
5. Accident Detection: The system will employ AI-based technologies to detect
accidents in real-time, improving incident management and response times.

5
Figure 3.1: System Architecture of the Smart Traffic Signal System, illustrating the interaction between
various components.

3.2 Proposed Methodology

The algorithms and techniques to be used in the Smart Traffic Signal System are as
follows:
1. Vehicle Detection and Counting:
o YOLO (You Only Look Once): This is a very advanced framework of real-
time object detection systems for identifying any type of vehicle and will
include but not limited to cars, buses trucks and motor cycles as mentioned
in paper [7]. The ability of YOLO to utilize the image effectively accounts
for the speedy rate of detection that is vital for the accurate monitoring of
traffic volume. This element of the system expects that the cameras will be
constantly working and adjusting the timing of signals according to the
traffic in real time thereby alleviating the risks of major traffic jams and
enhancing the control of the floods of vehicles.

6
2. Signal Timing Optimization:
o Adaptive Signal Control Technology (ASCT): This algorithm will be
customized to control real-time traffic signals based on the conditions of the
traffic environment so that delay can be minimized and flow may be
improved. The green period for traffic signals was calculated using the
following formula:
𝑽𝒆𝒉𝒊𝒄𝒍𝒆 𝑪𝒐𝒖𝒏𝒕
Green Duration = Base Duration + (𝑴𝒂𝒙 𝑽𝒆𝒉𝒊𝒄𝒍𝒆 𝑪𝒐𝒖𝒏𝒕) x

(𝑴𝒂𝒙 𝑫𝒖𝒓𝒂𝒕𝒊𝒐𝒏 − 𝑩𝒂𝒔𝒆 𝑫𝒖𝒓𝒂𝒕𝒊𝒐𝒏)

3. Violation Detection:
o License Plate Recognition (LPR): This component will utilize machine
learning models to perform number plate recognition on video footage from
the cameras constructed into the system. It offers great prospects for the
automatic reporting of traffic infringements like red light abuse or illegal
parking using the developed LPR system. The system will also lead to a
timely response to violation of laws and therefore increase road safety by
keeping a record of the number plates recognized.

4. Helmet Detection:
o Helmet Detection with AI Model: Development of an advanced AI model
designed exclusively for the purpose of detecting helmets on motorcycle
riders is done. This model will employ deep learning methods on a large
database filled with images of riders with and without helmets. The AI will
perform an analysis of images collected from traffic cameras and determine
whether the helmet laws are being followed or not. The system also has the
ability to send notifications to police departments whenever an infraction is
detected, enhancing cyclist safety.

7
5. Accident Detection:
o Custom AI Model for Accident Detection: A traffic condition monitoring
and event detection computer vision aided system is presented. Such model
is trained on a rich dataset containing various kinds of accidents including
vehicle impact, sudden stop, and abnormal traffic motion. This AI model
watches footage captured by traffic cameras and changes in movement,
orientations and/or more stimuli, finds the omission of movement or cars in
oncoming lanes when s/he/she is not supposed to be. And after an accident
happens, emergency services are notified automatically so that help arrives
as fast as possible to reduce the effects of the accident and increase the level
of road safety.

6. Emergency Vehicle Detection:

o Custom AI Model for Ambulance Detection: Thanks to the noticeable
advancement of image processing and the application of machine learning
techniques, a dedicated AI model has been developed that is capable of
recognizing the approaching of emergency vehicles, in particular,
ambulances. The optimization of this model is based on a dataset which
contains information of an array of emergency response vehicles under
various traffic settings. The AI interprets the video and detects the
ambulance’s moving parts, blinking lights, sirens, and general silhouette.
With this functionality, traffic lights are turned green to allow the ambulance
vehicle to pass the traffic signals with minimum or no stops at all. This
feature is important in speeding the provision of medical services during
emergencies, which can save lives.

8
3.3 System Design
The design of the Smart Traffic Signal System will encompass the following
elements:

1. Object-Based Design: The system will consist of various modules,

including:

o Traffic Management Module: Responsible for controlling traffic

signals based on real-time data.
o Violation Detection Module: Manages the detection and reporting of
traffic violations.
o Emergency Response Module: Prioritizes emergency vehicles and
manages traffic lights accordingly.
o Data Logging Module: Records all events, violations, and system
performance metrics.

2. Inter-module Relations: The modules will communicate with each other

through a centralized database, ensuring that real-time data is accessible for
decision-making. For instance, the Traffic Management Module will receive
inputs from the Detection Module to optimize signal timings.

3. Platforms: The proposed system will be built using Python with Flask for the
backend and JavaScript for the frontend, allowing for a responsive web
interface.

9
4. Flowchart:

Figure 3.2: Traffic Signal Control Flowchart illustrating the system's real-time processes
including traffic monitoring, vehicle classification, violation detection, and dynamic signal
adjustment.

10
5. Database Design: A MongoDB database is utilized to store traffic data,
violation records, and system logs. The choice of MongoDB allows for flexible
and scalable data management, accommodating the diverse and dynamic nature
of traffic information.

o Database Structure: The database design includes the following key

collections:

1. Traffic Data Collection: This collection records real-time vehicle

counts, signal timings, and traffic conditions. Each entry includes
timestamps, vehicle types, and corresponding signal states.
2. Violation Records Collection: This collection maintains logs of
detected traffic violations, including license plate numbers, violation
types (e.g., running red lights, illegal parking), timestamps, and
camera footage links for review.
3. System Logs Collection: This collection stores system logs,
capturing events related to the operation of the traffic signal system,
including errors, adjustments made to signal timings, and system
alerts.

Figure 3.3: Main homepage of the MongoDB database displaying traffic data, violation records, and
system logs for efficient real-time management in the Smart Traffic Signal System.

11
 Figure 3.4: Graph displaying the number of violations per road in the MongoDB database,
visualizing traffic infractions across different intersections.

3.4 Details of Hardware and Software Requirements

Hardware Requirements:

o High-resolution cameras for video feeds

o Servers for hosting the application and running AI models
o Network routers and switches for communication between components

Software Requirements:

o Programming Languages: Python (backend logic, AI models),

JavaScript (frontend functionality)
o Frameworks and Libraries: Flask (web framework), OpenCV (image
processing), YOLO (object detection), TensorFlow/PyTorch (deep
learning)
o Database: MongoDB (for data storage)

12
Chapter 4
Results and Discussion

4.1 Implementation Details

The following are key steps in the implementation of the Smart Traffic Signal
System:

1. System Setup:
o Hardware Installation: High-resolution cameras and traffic flow
sensors were installed at critical intersections to capture real-time traffic
data.
o Software Deployment: The system was developed using Python with
Flask for the backend and JavaScript for the frontend. MongoDB was
used to store traffic data, violation records, and system logs.
2. Algorithm Development:
o YOLO was integrated for vehicle detection and counting, enabling the
system to monitor traffic flow dynamically.
o The Adaptive Signal Control Technology (ASCT) algorithm was
customized to optimize traffic signal timings based on real-time data.
o Implementation of LPR and image processing algorithms was done for
violation detection and for helmet compliance monitoring.
3. Testing and Calibration:
o The system was tested thoroughly in order to allow the possibility of
accurate vehicle detection, signal adjustment, and violation reporting.
Its algorithms had been calibrated against the real traffic pattern.
4. User Interface Development:
o Responsive web interface was developed for traffic managers to
monitor system performance in real-time, view the current traffic
updates, and receive alerts regarding violations and emergencies.

13
5. Datasets Used for Model Training: For the training of the ambulance
detection, helmet detection, and accident detection models, the following
datasets were utilized, each containing a diverse array of images to ensure robust
model performance:
o Ambulance Detection: To achieve the aim of identifying and
prioritizing emergency vehicles within the model, it was found
necessary to use a dataset consisting of 9013 images of ambulances
taken within various surroundings manually. The photos in question are
showing the ambulatory vehicles under various lightings and traffic
conditions to increase model flexibility, acquired from [8].

Figure 4.1: Sample Image I from the Ambulance Detection Dataset.

Figure 4.2: Sample Image II from the Ambulance Detection Dataset.

14
o Helmet Detection: In order to train the helmet compliance
monitoring model a dataset consisting of 9787 images of motorcycle
riders with and without helmets was used. This diverse collection aids
to enhance the effectiveness of the model on helmet presence detection
in different conditions, as the images were available from [9].

Figure 4.3: Sample Image I from the Helmet Detection Dataset.

Figure 4.4: Sample Image II from the Helmet Detection Dataset.

15
o Accident Detection: To this end, a dataset with 4585 images
showing different traffic accident scenes was used to train the accident
detection system. The collection contains photographs of a variety of
accidents as this will also serve as a ground model in order to understand
and detect accidents in real time, accessed from [10].

Figure 4.5: Sample Image I from the Accident Detection Dataset.

Figure 4.6: Sample Image II from the Accident Detection Dataset.

16
4.2 Results
1. Traffic Flow Efficiency:
o The system revealed average saving in waiting times at the
intersections of about 30% compared to the traditional signals. Real-
time control by means of vehicle counts improved the traffic flow.
2. Emergency Vehicle Response:
o The emergency vehicle priority feature reduced the average time spent
by the emergency vehicles by nearly 40% because traffic signals turn
green in the direction of the approaching ambulance.
3. Violation Detection Accuracy:
o With License Plate Recognition, the accuracy of the detection of traffic
violations reached 92%, and violation alerts are automatically generated.
4. Helmet Compliance Monitoring:
o Helmet detection Algorithm correctly identified non-compliant
motorcyclists 85% of the time in order for law enforcement alerts to
arrive on time.
5. Incident Detection:
o Accident detection feature brought within the ability of a system to
identify incidents in real time, thus improving up to 25% response time
of emergency services.

17
[Following are the Output Images of our Project]

Figure 4.7: Traffic Signal Control Dashboard showing real-time traffic camera feed, violation alert and
signal status.

The Traffic Signal Control Dashboard is a panoramic view of live traffic conditions. This includes the
real-time feeds from cameras placed at the traffic junctions, along with alerts for speeding and accidents.
There are also signals on any vehicles such as ambulances within the vicinity. The dashboard also shows
the present status of the traffic lights.

Figure 4.8: Violation Display Dashboard showcasing real-time alerts for traffic violations, including
details of infractions and corresponding camera feeds.

This display screen indicates the Violation Display Dashboard which gives a real-time view of traffic
violations that happen at monitored junctions. It adds detailed information on the different infractions,
for instance speeding or signal violations

18
Figure 4.9: Dashboard showing real-time traffic data and past 24-hour traffic data.

This is an image of a dashboard showing real-time traffic data against historical data for the last 24
hours. The real-time information shows what is happening as it happens, and the past 24-hour
information shows one how the traffic has been flowing over time.

Figure 4.10: Car Counting Model Output showing real-time vehicle detection, speed measurement, and
counting.

This image presents the output of the Car Counting Model, which performs real-time detection of
vehicles on the road. The model accurately identifies each vehicle, measures its speed, and keeps a
running count of the total number of vehicles passing through a given area.

19
Figure 4.11: Accident Detection Model Output showcasing real-time identification of traffic incidents,
highlighting affected vehicles.

This image depicts the output of the Accident Detection Model, which identifies traffic incidents in real
time. The model highlights the vehicles involved in an accident, providing immediate visual feedback
for monitoring traffic safety, this feature is essential for quick response coordination.

Figure 4.12: Ambulance Detection Model Output demonstrating the system's capability to identify
emergency vehicles in real-time.

This image illustrates the output of the Ambulance Detection Model, showcasing its ability to identify
emergency vehicles in real-time. The system effectively recognizes ambulances on the road, enabling
rapid responses to medical emergencies.

20
Figure 4.13: Helmet Detection Model Output showcasing the system's effectiveness in identifying
motorcyclists wearing helmets.

This image displays the output of the Helmet Detection Model, highlighting the system's effectiveness
in identifying motorcyclists who are wearing helmets. The model utilizes advanced computer vision
techniques to analyze real-time video feeds and accurately detect helmet usage among riders.

21
 Chapter 5
Conclusion and Future Work

The main aim of this project is to help design an intelligent traffic management system,
addressing key problems that arise from real-world urban traffic systems, including
congestion, violation detection, and emergency vehicle prioritization. Our achievement
of this main aim was based on the implementation of real-time vehicle counting,
violation detection, and emergency response mechanisms within this system. The
ability of the system to adjust traffic signal timings on real-time data bases gives a good
improvement in the traffic flow as well as safety at junctions.

The key learnings from this project include incorporating various technologies, like
computer vision, machine learning, and sensor data, that come together to make an
effective solution in managing traffic. The development process was also underscored
by the necessity of adopting a user-centric design that is responsive to the needs both
of the traffic authorities and the road users themselves. Observations during the course
of the project indicated that real-time data analysis can lead to more responsive traffic
systems, which will ultimately contribute to reduced congestion and improved road
safety.

Future Scope

Looking ahead, there are several avenues for future enhancement of the Smart Traffic
Signal System:

1. Scalability: Expanding the system to cover larger urban areas or integrate

with existing city traffic management systems could further improve traffic flow
and safety.
2. AI-Powered Predictive Analytics: Implementing machine learning
algorithms to predict traffic patterns based on historical data could enhance
decision-making capabilities and optimize traffic signal timings even further.

22
 3. Integration with Smart City Initiatives: Collaborating with other smart
city technologies, such as connected vehicles and smart parking systems, could
lead to a more holistic approach to urban mobility.
4. User Feedback Mechanism: Incorporating a user feedback system could
provide valuable insights into the effectiveness of the traffic management
strategies and help refine them over time.
5. Environmental Considerations: Developing algorithms that also
account for environmental factors, such as reducing emissions during peak
traffic hours, could make the system more sustainable.

In conclusion, the Smart Traffic Signal System has all the ingredients to make urban
traffic management the best it can be with a foundation for the future in this area to
innovate upon. With continued development and expansion of the system designed
here, we also have further opportunities to enhance the productivity and safety of the
urban transportation infrastructure.

23
References

[1] [Link]
commuters-wait-an-average-of-128-hours-in-
traffic/articleshow/[Link]
[2] [Link]
jam-trafic-clogged-for-3-hours/articleshow/[Link]
[3] A. Hassan, S. Omer, and Y. Chen, “Adaptive Traffic Signal Control: A
Review of Current Technologies” IEEE Transactions on Intelligent
Transportation Systems, vol. 20, no. 3, pp. 904-915, 2019.
[4] L. Böcker and K. M. Kockelman, “Evaluating the Effectiveness of Red-
Light Cameras in Reducing Crashes” Transportation Research Part A:
Policy and Practice, vol. 124, pp. 1-13, 2019.
[5] A. García, J. Huerta, and R. Ceballos, “Predictive Traffic Control
Systems: A Review” Journal of Traffic and Transportation Engineering,
vol. 8, no. 3, pp. 234-248, 2020.
[6] A. H. Alavi, V. Raghavan, and S. Nadarajah, “AI-Based Solutions for
Enhancing Traffic Management Systems: A Review” Journal of
Intelligent Transportation Systems, vol. 25, no. 2, pp. 1-15, 2021.
[7] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You Only Look
Once: Unified Real-Time Object Detection” Proceedings of the IEEE
Conference on Computer Vision and Pattern Recognition (CVPR), vol.
2016, pp. 779-788, 2016.
[8] [Link]
[9] [Link]
ntbfz
[10] [Link]
jtx9t

24