/TrafficSignClassifier

A web app for classifying traffic signs using a deep learning model trained on the GTSRB dataset.

Primary LanguagePythonMIT LicenseMIT

⚠️ Traffic Sign Classifier ⚠️

A web app for classifying traffic signs using a deep learning model trained on the German Traffic Sign Recognition Benchmark (GTSRB) dataset.

Traffic Sign Classifier Preview

🚦 Overview

The Traffic Sign Classifier is a web-based application that allows users to classify traffic signs by uploading an image. It uses a Convolutional Neural Network (CNN) trained on the GTSRB dataset to identify and classify different types of traffic signs, providing the predicted label along with the confidence level.

💻 Features

  • Real-time Traffic Sign Classification: Upload an image of a traffic sign, and the model instantly predicts its label and confidence score.
  • Drag & Drop Interface: Users can drag and drop or select images directly from their device for classification.
  • Clear Button: A simple option to reset the input and upload a new image.
  • Responsive Design: Optimized for various screen sizes, from desktop to mobile.
  • Light/Dark Mode: Seamless switch between light and dark modes to suit your preference.

🧠 Model

The classifier is built using a deep learning model, specifically a Convolutional Neural Network (CNN), trained on the GTSRB dataset. The dataset consists of over 50,000 images across 43 different traffic sign categories.

  • Dataset: GTSRB Dataset
  • Model Architecture:
    • 3 Convolutional Layers
    • 2 Fully Connected Layers
    • Softmax for output classification

The model achieves 99.05% accuracy in recognizing traffic signs from the test set.

📷 How It Works

  1. Upload an Image: Click on the upload area or drag & drop an image of a traffic sign.
  2. Preview: The app displays a preview of the uploaded image.
  3. Classify: Once uploaded, the classifier will process the image and return the predicted traffic sign label along with a confidence score.
  4. Result Display: The app presents the classified traffic sign with the associated confidence level and metadata.

🚀 Getting Started

Prerequisites

Ensure you have the following installed:

  • Python==3.12.3+
  • Django==5.1.1+
  • torch==2.4.1+cu118
  • pillow>=10.3.0

Installation

  1. Clone the Repository:

    git clone https://github.com/FazleLabib/TrafficSignClassifier.git
cd TrafficSignClassifier

  2. Install Dependencies::

    pip install -r requirements.txt

  3. Run Migrations::

    python manage.py migrate

  4. Start the Django Development Server::

    python manage.py runserver

  5. Open the Application::

    Visit http://127.0.0.1:8000/ in your browser to access the Traffic Sign Classifier app.