This project aims to develop an AI-based solution for classifying eye diseases from retinal images. Using a dataset of retinal images, the model identifies four types of eye diseases with high accuracy. The project leverages state-of-the-art deep learning models, including ResNet50 , to ensure reliable performance.
The project workflow includes:
1. Data Preprocessing: Cleaning and resizing the images to ensure consistency.
2. Model Development: Training multiple models using the fastai library to achieve the highest accuracy.
3. Deployment: Integrating the trained model into a mobile application using Flutter, with backend support through Dradio or Hugging Face for interactive predictions.
The solution is designed to assist healthcare professionals in diagnosing eye diseases more efficiently, reducing manual errors, and enhancing patient care.
Key Achievements:
• Achieved an impressive accuracy of 91% in classifying eye diseases.
• Successfully deployed the model on platforms like Hugging Face and integrated it with a Flutter-based mobile application.
Future Scope:
• Expanding the dataset for improved accuracy and generalization.
• Implementing a user-friendly interface for seamless interaction.
• Exploring real-time image processing for instant predictions.
