Image Classifier with Python

A deep learning-based image classification project using Convolutional Neural Networks (CNN) and the CIFAR-10 dataset, featuring an intuitive Tkinter GUI for real-time image classification.

Overview

This project implements a CNN-based image classifier that can recognize and classify images into 10 different categories from the CIFAR-10 dataset. The application provides a user-friendly graphical interface for uploading images and receiving instant classification results.

Features

• Deep Learning Model: Custom CNN architecture with multiple convolutional and dense layers
• CIFAR-10 Dataset: Trained on 60,000 images (50,000 training, 10,000 testing)
• Interactive GUI: Built with Tkinter for easy image upload and classification
• Real-time Classification: Instant prediction results displayed in the interface
• Model Persistence: Trained model saved as HDF5 file for reuse
• Data Preprocessing: Automatic image normalization and resizing

Classified Categories

The model can classify images into the following 10 classes:

• ✈️ Aeroplane
• 🚗 Automobile
• 🐦 Bird
• 🐱 Cat
• 🦌 Deer
• 🐕 Dog
• 🐸 Frog
• 🐴 Horse
• 🚢 Ship
• 🚚 Truck

Model Architecture

The CNN model consists of:

1. Convolutional Layers:

   • Conv2D (32 filters, 3x3 kernel) with ReLU activation
   • Dropout (0.2)
   • Conv2D (32 filters, 3x3 kernel) with ReLU activation
   • MaxPooling2D (2x2 pool size)

2. Fully Connected Layers:

   • Flatten layer
   • Dense (64 neurons, ReLU)
   • Dense (512 neurons, ReLU)
   • Dropout (0.5)
   • Dense (10 neurons, Softmax) - Output layer

3. Optimization:

   • Optimizer: SGD (learning rate: 0.01, momentum: 0.9)
   • Loss function: Categorical Crossentropy
   • Metrics: Accuracy
   • Training: 10 epochs with batch size of 32

Requirements

tensorflow>=2.0.0
keras>=2.0.0
opencv-python
matplotlib
pillow
numpy

Installation

1. Clone the repository:

git clone https://github.com/yourusername/Image-Classifier-with-Python.git
cd Image-Classifier-with-Python

2. Install required dependencies:

pip install tensorflow keras opencv-python matplotlib pillow numpy

Usage

Training the Model

Run the main script to train the model:

python "Image Classifier Project.py"

The script will:

1. Load and preprocess the CIFAR-10 dataset
2. Build the CNN model architecture
3. Train the model for 10 epochs
4. Display training accuracy
5. Save the trained model as model1_cifar_10epoch.h5
6. Launch the GUI application

Using the GUI Application

1. Upload Image: Click the "Upload Image" button to select an image file
2. Classify: Click the "Classify the Image" button to get the prediction
3. View Results: The classification result will be displayed in the interface
4. Data Information: Click "Get Data Information" for dataset details

Project Structure

Image-Classifier-with-Python/
│
├── Image Classifier Project.py    # Main application file
├── README.md                       # Project documentation
└── model1_cifar_10epoch.h5        # Trained model (generated after training)

Performance

The model achieves competitive accuracy on the CIFAR-10 test dataset. The exact accuracy is displayed in the console after training completion.

Technical Details

• Input Shape: 32x32x3 (RGB images)
• Normalization: Pixel values scaled to [0, 1]
• Data Augmentation: None (can be added for improved performance)
• Kernel Constraint: MaxNorm(3) for regularization
• Dropout: Used to prevent overfitting

Author

Enes Günümdoğdu