1. Technical description
Python language, TensorFlow, convolutional neural network CNN algorithm, PyQt5 interface, Django framework, deep learning
Contains: training prediction code, data set, PyQt5 interface + Django framework web interface
2. There are training codes and data sets in the training prediction folder
3. carnum_check is Django web version qt_check is QT version
4. Version description:
TensorFlow uses the latest version 2.11.0, django uses the latest version 4.1.7, pyqt5 uses the latest version
5. Model: more than 10,000 pictures in 25 iterations
The license plate recognition system uses the machine learning library tensorflow as the model building framework, uses the CNN convolutional neural network to build the model, and divides the test set and the training set through the processing of the data set, and obtains the trained model through multiple rounds of iterations, and then the model Carry out packaging, and develop a WEB interface system for user operations, and finally realize that the user enters a license plate picture on the WEB page, and the system displays the license plate information (regional Chinese characters, letters, numbers) in the form of a pop-up window. At the same time, the pictures entered by the user, the predicted results, and the operation time will all be stored in the database, and the administrator can dynamically view these information in the background management system.
The following will mainly explain the realization of each design function module of this system. It includes system environment configuration, system operation interface, function module relationship and system operation flow chart, etc., and introduces the implementation process of this system in detail.
2. Environment configuration
In the functional design and implementation of the license plate recognition system, in order to facilitate the display and provide users with a good visual operation platform, the system uses the WEB web page interface as the operation platform, and uses jupyter notebook as the algorithm research platform. The development environment configuration is shown in Table 1.1.
Table 1.1 Development environment configuration information
Configuration Name Parameter Remarks
CPUIntel(R) Core(TM) i5-8300H CPU8 core
Memory 16GB None
Memory 2GB None
Operating systemWindowsWindows10 version
This system uses Pycharm as the development software, uses Python as the programming language, and builds a development environment on the Windows system. The Python third-party package information mainly used in this system: tensorflow is version 2.8.0 and Django is version 4.0.
3. The main interface of the system
The main interface of the license plate recognition system is divided into three, the system description home page, the user input picture detection page, and the management background information management page
By starting the Django application, open the webpage and enter 127.0.0.1:8000 to enter the home page of the system description. The system description home page is shown in Figure 2.1.
Figure 2.1 System home page
On the home page of the system, the HTML language is used to build the page skeleton, the CSS language is used to adjust the page style, and the main introduction functions of the system are displayed. Provide users with an overview of the system, etc. The system home page index.html file is located in the template front-end folder.
In order to enable users to input and upload a picture to predict picture information, a detection page is designed in this system, and the page picture is shown in Figure 2.2.
Figure 3.2 Flow chart of backend event processing
Four method classes are defined in the back-end processing view, namely index method, check method, upload_img method, and check_img method. The index method is responsible for rendering the homepage page event when the user clicks the home page, the check method is responsible for rendering the detection page event when the user clicks the detection page, and the upload_img method is responsible for uploading the license plate image entered by the user to the specified folder location. The main implementation steps of this method are as follows: 1. The user clicks the upload picture button, selects a picture and clicks upload. 2. Trigger the upload event in JavaScript By obtaining the button ID uploaded by the user, extract the picture uploaded by the user to form an object and save it in a variable. 3. Initiate a POST request to the backend and carry image variable information. 4. The backend calls methods such as with to save the picture information in the media folder, and returns the URL information of the picture. 5. After the front end receives the URL information of the picture, it changes the default picture IMG tag attribute and displays it. The image upload flow chart is shown in Figure 3.3 below.
5. Algorithm Module Realization
This module mainly implements the model training part of the license plate recognition system. Through processing the data set, building a network model, iterative training and other stages, an accurate prediction model is finally obtained. Its overall flow chart is shown in Figure 4.1.
Figure 4.1 Overall flow chart of the algorithm module
A good model first needs to provide a good data set. This system uses an open source license plate data set, which contains 13675 pictures. Figure 4.2 is part of the picture information displayed in jupyter notebook. As shown in Figure 4.3, it is a partial picture of the data set downloaded in this machine.
From the last figure 4.10, it can be seen that after 25 rounds of training, the prediction accuracy has increased from 0.06 in the first round to 0.57, and the accuracy has increased by 100 times. Models have certain practicability. Next, draw the ACC curve and Loss curve during model training as shown in Figure 4.11.
Figure 4.11 ACC and LOSS curves during training
Finally, save the training model as the H5 model, and customize a picture for prediction, as shown in Figure 4.12.
Other study items:
OpenCV+dlib face recognition access control management system Python language, dlib, OpenCV, Pyqt5, sqlite3 database
OpenCV+dlib face recognition attendance management system Python language, dlib, OpenCV, Pyqt5, sqlite3 database
OpenCV+dlib face recognition system Python language, dlib, OpenCV, Pyqt5, sqlite3 database
Flask+LSTM music double recommendation algorithm + crawler + visualization system (deep learning)
Flask+LSTM movie double recommendation algorithm + crawler + visualization system (deep learning)
Flask+LSTM recruitment double recommendation algorithm + crawler + visualization system (deep learning)
deepsort road vehicle tracking + speed real-time analysis system
License plate recognition + vehicle management + billing system
2023 license plate recognition system (Django framework + convolutional neural network CNN + PyQt5 interface)
License plate recognition system based on convolutional neural network
License plate recognition system 1
License plate recognition system 2
OCR license plate recognition YOLOv5
Employee face attendance system
Animal recognition system based on Django
UAV target recognition + target tracking (OpenCV+YOLO)
Face recognition OpenCV face detection pyqt5 framework, face management
YOLO target detection algorithm real-time target detection pytorch
pyqt5+yoloV5 target detection platform
gesture control mouse
gesture control volume
Gestures to drag and drop objects
Face recognition intelligent attendance management system based on opencv
Python face recognition sign-in and attendance system (opencv+pyqt5)
Old photo restoration system based on deep learning
Character Relationship Visualization and Question Answering System Knowledge Graph Character Relationship
Python+Django realizes the access control management system based on face recognition
Python face recognition sign in
Python deep learning - weather recognition
python deep learning - face recognition
Python deep learning--attention detection
Python deep learning - verification code recognition
Python deep learning--garbage classification and recognition
Python deep learning - handwriting recognition
Python deep learning--traffic sign recognition
Python deep learning - pet recognition
Python deep learning - vehicle classification
Python deep learning--commodity recognition
Python deep learning--recognition system
Python deep learning--bird recognition
python deep learning - face recognition
Python deep learning--attention detection
Python deep learning - fruit recognition
Python deep learning - garbage classification
Python deep learning - handwritten digit recognition
Python deep learning - gesture recognition
Python deep learning - mask recognition
Python deep learning--news classification
Python deep learning - license plate recognition
Django project, Flask project, machine learning, data analysis, big data, Java project (nearly a thousand sets of project source code, all of which can be shared and learned)