Slide Icera

ICERA 2022
RESEARCH ON BUILDING THE SYSTEM OF THE IDENTIFICATION AND

DETECTION TRAFFIC VIOLATIONS VEHICLES AT VIETNAMESE INTERSECTIONS
Presenter: Trinh Luong Mien & Vu Van Duy – University of Transport and Communications
INTRODUCTI
Vietnam usesON
many motorbikes and many types of transportation, so the traffic flow in Vietnam
is very complicated. Public transport in Vietnam is not efficient. Traffic congestion at rush hour,
especially in big cities such as Hanoi and Ho Chi Minh City
The state of urban traffic in Vietnam. The state of urban traffic in Vietnam.
The state of urban traffic in Vietnam.

Identification
.
Detection of vehicle number

plates
Allow monitoring
Managing vehicles
Control from the center

SOLUTION FOR BUILDING THE SYSTEM OF THE IDENTIFICATION
AND DETECTION TRAFFIC VIOLATIONS VEHICLES AT
VIETNAMESE INTERSECTIONS
The system of monitoring traffic

vehicle volume and density on the street
The Apps Web software programs The hardware devices
• Traffic cameras.
Cameras IoT The Web program provides : • IoT-based
processing board.
• Information to users. The equipment • Data server
• Warnings to users. devices used : computer.
Developed to : Identify and detect

traffic vehicles.
Traffic violations on
the street.
applied image
processing.
OPERATING PRINCIPLE OF THE SYSTEM
image data Technical

Internet cabinets.
…
Running red Exceeding the

The server analyzes Support traffic lights allowed speed
and processes the management and
detection of vehicle traffic
offending vehicles. statistics.
Calculate the traffic vehicle volume and

density through the intersection on the street
Data storage
server.
SYSTEM STRUCTURE BLOCK DIAGRAM
Monitoring and Monitoring and

warning on Apps warning Web on
smatphone. Laptop, Tablet.
Internet
The App and Web programs
The hardware devices Data Web program

server
IoT-based
processing board
Apps program
Traffic Cameras
THE ALGORITHM FLOWCHART OF THE SYSTEM OF
MONITORING TRAFFIC VEHICLE VOLUME
Begin Begin
False
Initial
Vehicles in motion
Connect traffic cameras

Operate the processing board
Detection
Image processing Apps line
True
Transmit the data to server
Counting Vehicles exiting the
intersection
Warn, show traffic situation

On Web
End
End
DESIGN THE IOT-BASED PROCCESSING BOARD
Based on the requirements of the system, the author team designed the principle circuit,
the printed circuit of the processing board on Altium Designer software.
The principle circuit The printed circuit board

DEVELOPMENT OF THE APP PROGRAM
Image, label in text formatlist of labels trained on

the new dataset.
List of files used for training

and evaluation
Configuration file
with architecture Training model YOLOv5
classes YOLOv5
The previous training file of
the model YOLOv5
Trained file *.pt
The Yolov5 model training process

TRAINING DATA SET
The training data set is collected through traffic cameras mounted at the
intersection on Le Duan - Kham Thien streets (Hanoi), and images shot from
specialized cameras on Nguyen Trai, Nguyen Xien, Le Van Luong streets (Hanoi).
The number of collected

images
20%
80%
Trained images: 25600 Pictures for testing: 6400

Total : 32000 Regression, classification, and object loss changes of the
bounding box during model training.
DEVELOPMENT OF THE APP PROGRAM
System management interface Traffic counting interface
Violation vehicle management interface Traffic violation detection interface

DEVELOPMENT OF THE WEB PROGRAM
The Web program is developed to manage and monitor vehicles.
Violation vehicle management Interface extract images of

System management interface
interface vehicles violating traffic
The screen of the graph of the

The statistical interface of traffic
number of each vehicle passing
through the intersection
through the intersection
IMPLEMENTATION AND RESULTS
Result of detecting movement direction and

Results of speed detection and vehicle
counting vehicles at low point of time on
counting on Vietnam's streets
Vietnam's streets
Experiment to adjust the traffic light cycle at the intersection Experiment to adjust the traffic light cycle at the intersection
LeDuan-KhamThien str. at off-peak hours LeDuan-KhamThien str. at rush hour
Monitor web interface

The results of recognition and counting traffic vehicles at
intersections on the Vietnamese streets
Accuracy
Bus
Truck
Car
Motor bike
92 93 94 95 96 97 98
Accuracy
Evaluation of the adaptive control effect of signal lights over different periods.
Time to move through the node before and after applying Time to move through the node before and after applying
the method. the method.
Thank you !

Slide Icera

Uploaded by

Copyright:

Available Formats

Slide Icera

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Slide Icera

Uploaded by

Copyright:

Available Formats

ICERA 2022

RESEARCH ON BUILDING THE SYSTEM OF THE IDENTIFICATION AND

The state of urban traffic in Vietnam.

Detection of vehicle number

Control from the center

The system of monitoring traffic

The Apps Web software programs The hardware devices

Developed to : Identify and detect

image data Technical

Running red Exceeding the

Calculate the traffic vehicle volume and

Monitoring and Monitoring and

The hardware devices Data Web program

Connect traffic cameras

Warn, show traffic situation

The principle circuit The printed circuit board

Image, label in text formatlist of labels trained on

List of files used for training

Trained file *.pt

The Yolov5 model training process

The number of collected

Trained images: 25600 Pictures for testing: 6400

System management interface Traffic counting interface

Violation vehicle management interface Traffic violation detection interface

The Web program is developed to manage and monitor vehicles.

Violation vehicle management Interface extract images of

The screen of the graph of the

Result of detecting movement direction and

Monitor web interface

You might also like