Đồ án tốt nghiệp - Smart parking manugement integrated deep learning for automatic monitoring

HCMC UNIVERSITY OF TECHNOLOFY AND EDUCATION
FACULTY FOR HIGH QUATITY TRAINING
INFORMATION TECHNOLOGY

CAPSTONE PROJECT

SMART PARKING MANAGEMENT INTEGRATED
DEEP LEARNING FOR AUTOMATIC
MONITORING
LECTURER: PhD. Le Vinh Thinh
COURSE YEAR: 2019
STUDENT:
Pham Van Manh Hung 19110081
Nguyen Ngoc Thien
19110148
MAJOR: Information Technology

Ho Chi Minh City, July 2023

SOCIALIST REPUBLIC OF VIETNAM
Independence - Freedom – Happiness

*******
Ho Chi Minh City, 10 month 07 year 2023

TARGET OF THE GRADUATION THESIS
Students first and last name: Pham Van Manh Hung

ID: 19110081

Specialization: Information Technology

Class: 19110CLA3

Students first and last name:Nguyen Ngoc Thien

ID: 19110148

Specialization: Information Technology

Class: 19110CLA2

Instructors: PhD. Le Vinh Thinh

Date of receiving topic: 13/02/2023

Date of submission topic: 10/07/2023

1. Title of the topic: Smart Parking Management integrated Deep Learning for automatic
monitoring.
2. Data, original documents: Survey the current status through searching websites on the
Internet today. In addition, learn about the shortcomings of traditional and smart garage
management in apartments and schools.
3. Content of the topic:
- Built an automatic vehicle monitoring application integrated with deep learning.
- Building a website to manage the parking system.
- Build an application that allows users to register in the system when entering an
apartment or school.
4. Products: Smart garage management software system.
Dean of Faculty

(Full name, signature)

Instructors

(Full name, signature)

SOCIALIST REPUBLIC OF VIETNAM
Independence - Freedom – Happiness
***********

INSTRUCTOR COMMENT SHEET
Student 1: Pham Van Manh Hung
Student ID 1: 19110081
Student 2: Nguyen Ngoc Thien
Student ID 2: 19110148
Instructor: PhD. Le Vinh Thinh
Phone number:
Major: Information technology
Project name: SMART PARKING MANAGEMENT INTEGRATED DEEP LEARNING
FOR AUTOMATIC MONITORING
Comment:
1. About the content of the topic and the block of work done:
................................................................................................................................................
........................................................................................................................................
2. Advantages:
................................................................................................................................................
............
...

...
which
takes advantage of the unique characteristics of the customer such as face and license
plate. This gives the expected results for the actual deployed system.
In addition, the team has completed the management website system to limit the
error of the models and fully meet the proposed management functions.
7.1.2. Disadvantages
Due to the limited project development time, some functions of the system cannot be
completed, such as:
- Cross-platform application integration
- The system interface is still rudimentary and not familiar to users
7.1.3. Convenience
- There are many tutorials available for beginners with Deep learning models.
- There are many smart and traditional garage management models on the market to
facilitate surveying user needs.
- The system includes many functions to meet the management capacity of the system.
- Enthusiastic instructors and professional knowledge are effectively put into practice.
7.1.4. Difficulties
Some difficulties and solutions:

STT

Difficulties

Solutions

108 | Page

CHAPTER 7: INSTALL AND TEST

1

Difficulty in finding and

- Search from the parking lots of Ho Chi

collecting data.

Minh City University of Technical
Education
- Collect data from the Roboflow site

2

3

The system has many

Divide specific work among members and

functions and limited

conduct daily meetings to report on regular

ution time

work

The transformation between

Learn how to convert the model into a C#

training models into real

application using the

projects

Microsoft.ML.OnnxRuntime library to read
the Onnx file of the deeplearning model
Table 7.1.1. Some difficulties and solutions

7.2. Oriented development
- Built-in phone alert if someone illegally enters the parking location.
-

Improve processing speed and performance of training models.

-

Build a revenue management website so that managers can check the total income
and timekeeping of all employees.

109 | Page

REFERENCE

REFERENCE
[1]: Introduction of MS SQL Server - GeeksforGeeks
[2]: License plate detection dataset : SampleLP Dataset Overview (roboflow.com)
[3]: License plate detection dataset : SampleLP Dataset Overview (roboflow.com)
[4]: vietocr · PyPI
[5]: https://huyhoang17.github.io/128D-Facenet-LFW-Embedding-Visualisation
[6]: https://viblo.asia/p/data-visualization-voi-thuat-toan-t-sne-su-dung-tensorflowprojector-924lJAAzZPM
[7]: https://towardsdatascience.com/using-transformers-for-computer-vision6f764c5a078b

110 | Page