Image Classification Project

This project is a deep learning-based image classification system built with TensorFlow and Keras. It classifies images into two categories: "happy" and "sad". The workflow includes data preprocessing, model training, evaluation, and prediction.

Project Structure

├── data/
│   ├── happy/
│   └── sad/
├── logs/
│   ├── train/
│   └── validation/
├── models/
│   └── imageclassifier.h5
├── notebook.ipynb
├── requirements.txt
└── venv311/

• data/: Contains subfolders for each class ("happy", "sad") with images for training and validation.
• logs/: Stores TensorBoard logs for training and validation.
• models/: Contains the saved Keras model.
• notebook.ipynb: Jupyter notebook with step-by-step code for data processing, model building, training, evaluation, and testing.
• requirements.txt: List of required Python packages.
• venv311/: Python virtual environment.

Main Steps

1. Install Dependencies
   • TensorFlow, OpenCV, Matplotlib, and other required packages.
2. Data Cleaning
   • Remove corrupted or unsupported images.
3. Data Loading & Preprocessing
   • Load images using tf.keras.utils.image_dataset_from_directory.
   • Scale pixel values to [0, 1].
   • Split data into training, validation, and test sets.
4. Model Building
   • Build a CNN using Keras Sequential API.
5. Training
   • Train the model with TensorBoard logging.
6. Evaluation
   • Evaluate model performance using accuracy, precision, and recall.
7. Testing
   • Test the model on new images.
8. Saving & Loading Model
   • Save the trained model and demonstrate loading for inference.

Usage

1. Clone the repository and navigate to the project directory.
2. (Optional) Create and activate a Python virtual environment.
3. Install dependencies:

   pip install -r requirements.txt

4. Place your images in the data/happy/ and data/sad/ folders.
5. Open and run the Getting Started.ipynb notebook to train and test the model.

Requirements

• Python 3.11+
• TensorFlow
• OpenCV
• Matplotlib
• NumPy

All dependencies are listed in requirements.txt.

Model

• The model is a simple CNN with Conv2D, MaxPooling2D, Flatten, and Dense layers.
• Trained model is saved as models/imageclassifier.h5.

Logs

• Training and validation logs are saved in the logs/ directory for visualization with TensorBoard.

License

This project is for educational purposes.