Cat Classification Project

Overview

This project implements machine learning models to classify images containing Lacta, Karamela, or both cats. The system addresses challenges like class imbalance, visual complexity, and limited training data through a systematic approach including baseline CNNs, transfer learning, and Bayesian optimization.

Quick Start (View Only)

No installation required! You can read all notebooks directly on GitHub:

• Simply open any .ipynb file in the notebooks folder
• GitHub will render the notebooks with all code, markdown, and outputs visible
• Perfect for reviewing methodology and results without running code

Full Setup (Interactive Execution)

1. Clone Repository

git clone https://github.com/Raumfahrer007/cat_classifier.git
cd cat-classifier

2. Install Jupyter Notebook

pip install jupyter notebook

3. Download Project Assets

From Google Drive: https://drive.google.com/drive/folders/1Rt7a59YWzPHRuB3BATFKmCIZ0MHnRrpD?usp=sharing

• Download the models/ folder and place it in the project root
• Download the raw/ folder and place it inside data/

Final structure should be:

text cat_classifier/
├── models/ # ← Download from Google Drive
├── data/
│ └── raw/ # ← Download from Google Drive
├── notebooks/
├── src/
└── ...

4. Launch Jupyter Notebook

jupyter notebook

5. Execute Notebooks in Order

Notebook 1: 01_data_exploration_and_preparation

• Run all cells to split raw data into train/val/test sets
• This creates the structured dataset for training

Notebook 2: 02_machine_learning

• Run all cells EXCEPT training cells (models already provided)
• Execute evaluation cells to reproduce results and analyses
• Explore the pre-trained models and their performance

Important Notes

• Training is optional: All trained models are included in the downloaded models/ folder
• Data privacy: The raw images are provided separately via Google Drive

Project Structure

• notebooks/01_data_exploration_and_preparation.ipynb - Dataset splitting and analysis

• notebooks/02_machine_learning.ipynb - Model development, training, and evaluation

• models/ - Pre-trained models (download from Google Drive)

• data/raw/ - Original images (download from Google Drive)

• data/processed/ - Generated by Notebook 01

• src/ - additional code required for the notebooks

Requirements Core dependencies are listed in each notebook and will be installed as needed. Main packages include TensorFlow, scikit-learn, Optuna, and standard data science libraries.