Glaucoma Detection using Deep Learning & NSGA-II

This project implements a deep learning pipeline for Glaucoma detection using retinal fundus images.
It combines Convolutional Neural Networks (CNNs), Transfer Learning, and NSGA-II (Genetic Algorithm) for hyperparameter optimization.
A Flask web app is also included for real-time predictions and Grad-CAM visualizations.

---

🎥 Project Demo

▶️ Watch the Demo Video

---

Model Performance

Grad-CAM++ Visualization

---

Project Structure

• main.py → Core training script with:

  • Custom Enhanced CNN with CBAM attention
  • Transfer Learning support (ResNet18, MobileNetV3)
  • K-Fold cross-validation
  • NSGA-II for hyperparameter optimization
  • Metrics, plots, and Grad-CAM generation

• app.py → Flask web application for:

  • Uploading fundus images
  • Running inference using trained model
  • Viewing Grad-CAM heatmaps for predictions

• results/ → Trained models, metrics, and plots will be saved here.

• datasets/ → Acrima, Drishti-GS1, HRF, ORIGA, Rim-One.

---

Requirements

Install dependencies:

pip install -r requirements.txt