🧠 CNN-Based Image Classification Project

This repository contains a deep learning project implementing three powerful CNN architectures layer by layer for multi-class image classification.

🚀 Project Overview

• Models implemented:

  • GoogleNet
  • EfficientNet
  • Vision Transformer (ViT)

• Dataset: 2,484 images categorized into 7 classes

• Achieved 97.423% accuracy

• Ranked 1st (tied) on the Ain Shams AI department Kaggle leaderboard

---

📂 Repository Structure

├── notebooks/
│   └── cnn_classification.ipynb      # Main Jupyter notebook with model training and evaluation
├── testing_scripts/
│   ├── test.py                       # Script for testing trained models on new data
│   ├── output/                       # Folder where CSV results will be saved
│   ├── models/                       # Folder where trained models are saved
│   └── Test/                        # Folder where test photos should be placed
├── README.md  

🏆 Result

Achieved 97.423% accuracy on 2484 unseen test samples and ranked 1st (tied) on the Ain Shams CS department Kaggle leaderboard

📋 How to Use

1. Open and run the notebook in the notebooks/ folder to train and evaluate models.
2. Use the testing script in testing_scripts/ to run inference on new images.

---