DATA-662 - MDCH-615

🧠 Final Project — Data 622

Machine Learning Classification Using Transfer Learning and X-Ray Images

This project explores the use of transfer learning for multi-class classification of chest X-ray images into three clinically meaningful categories: Normal, Viral Pneumonia, and Bacterial Pneumonia. Leveraging the publicly available CoronaHack Chest X-Ray Dataset, we benchmarked various state-of-the-art CNN architectures and fine-tuned them for medical image classification tasks.

📌 Best model: EfficientNetV2-S, achieving 92.79% test accuracy and 0.9780 micro-AUC.

---

📂 Repository Structure

Final_Project_Data622/
│
├── Dataset/               # Metadata and label files (images downloaded separately from Kaggle)
├── Final Code/            # Final version of training and classification scripts using EfficientNetV2-S
├── Testing Models/        # Experiments with multiple pre-trained CNN architectures
│   ├── ResNet101/
│   ├── Dense121/
│   ├── Xception/
│   └── ... (etc.)
│
├── README.md              # Main project README (this file)