top of page

Vehicle Plate Number Classifier Engine

Problem Statement
The need to distinguish between regular license plate numbers and those belonging to various government agencies, such as the Army, Navy, Police, and other official vehicles, presents a critical challenge. This project aimed to develop a system capable of accurately classifying license plate numbers into predefined categories, addressing the need for automated categorization in domains such as law enforcement, toll collection, and parking management.
Project Description

To solve the classification problem, a multi-classifier engine was designed and implemented using a Naive Bayes Classifier. The solution leveraged Python for model development and FastAPI for deployment. Key steps in the project included:

  1. Data Collection:

    • A dataset comprising over 4,000 unique license plate numbers labeled with their respective categories was gathered.

    • Data retrieval was accomplished using MySQL techniques.

  2. Data Preprocessing:

    • Data cleaning, duplicate removal, and formatting were performed to prepare the dataset.

    • The dataset was split into training and testing sets for robust model evaluation.

  3. Feature Engineering:

    • License plate numbers were converted into numerical formats using Count Vectorizer, enabling the algorithm to interpret text data effectively.

  4. Model Development:

    • A Naive Bayes Classifier, known for its simplicity and effectiveness in text-based multi-class classification, was trained on the preprocessed data.

    • The model established relationships between plate numbers and their associated categories during the training phase.

  5. Model Evaluation:

    • The model was evaluated using metrics such as accuracy, precision, recall, and F1-score.

    • The classifier achieved an impressive 93% accuracy on the testing dataset.

  6. Deployment with FastAPI:

    • The trained model was deployed using FastAPI, allowing real-time predictions via API calls.

    • Users can input license plate numbers and receive their corresponding categories instantly.

Project Outcomes

The project successfully delivered the following results:

  • High Accuracy: The model achieved a 93% accuracy score, demonstrating its reliability in license plate classification.

  • Real-Time Predictions: FastAPI integration enabled real-time category predictions, enhancing user accessibility and system usability.

  • Scalability: The solution can be scaled by incorporating additional data, improving feature engineering, and experimenting with other machine learning models.

  • Practical Applications: Potential use cases include:

    • Automated toll collection systems.

    • Parking management solutions.

    • Law enforcement tools for vehicle tracking and categorization.

  • Future Enhancements: Proposed improvements include integrating new data sources, enabling real-time updates, and continuously monitoring model performance to adapt to evolving requirements.

Tech Stacks

  • Programming Language: Python

  • Libraries: Scikit-learn, Pandas, FastAPI

  • Data Preprocessing: Count Vectorizer

  • Database: MySQL

  • Deployment Framework: FastAPI

  • Evaluation Metrics: Accuracy, Precision, Recall, F1-score

bottom of page