Vehicle Number Plate Detection

The client needs an OCR (optical character recognition) scanning mobile application for vehicle number plate detection that will be able to detect the number on the number plate of cars and store it in the database.

Designing a vehicle number plate detection software involves several steps, including:

1. 01

   Defining the problem:

   The first step is to define the problem we want to solve. In this case, the client experienced problems detecting the numbers on a blurry vehicle number plate available in images or videos captured by a camera.

2. 02

   Gathering Data:

   As this is a huge problem, the client wants to solve it by designing software that can easily detect the vehicle number from blurry images or fast videos. For this, we gathered the data to train our software. We require a large dataset of images or videos that contain vehicles and their number plates. We sourced most of the data from publicly available sources and some by ourselves.

3. 03

   Preprocessing the Data:

   After gathering the data, we preprocessed the images and videos by resizing them, converting them to grayscale, and applying filters to remove noise and enhance edges.

4. 04

   Training the software:

   In the next step, we trained the software using machine learning techniques and deep learning algorithms such as convolutional neural networks (CNNs) to detect number plates.

5. 05

   Testing the software:

   After training the software, we tested it on a separate dataset to evaluate its performance. We also used metrics such as precision, recall, and F1 score to measure the accuracy of our software.

6. 06

   Optimizing the software:

   We also optimize our software by tweaking the hyperparameters or adding more training data to enhance its performance.

7. 07

   Deploying the software:

   After getting satisfied with the performance of our software, we deployed it to the server.

• OpenCV
• .Net Framework
• C#
• MSSQL

Adequate Infosoft has developed a desired OCR scanning mobile application. This application is used to identify the license plate of a car and get its vehicle number.

How this app works:

• There are sensors installed at various points, and when any vehicle passes through one of those sensors, the camera takes a picture and uploads it to an FTP server.
• After that, our application will analyze that image to locate the license plate number.
• The application recognizes the license plate and generates a new image of the plate that is distinct and clear than the previous one.
• After the image has been identified, it is parsed to obtain the correct plate number.

The software can automatically detect the number plate in any image.