• Introduction
• Features
• Technologies Used
• Usage
• Dataset & Model Training
• Results

Diabetic Retinopathy Detection is a machine learning project aimed at detecting diabetic retinopathy in retinal images. Diabetic retinopathy is a medical condition where the retina is damaged due to diabetes, which can lead to blindness if not detected early. This project leverages deep learning techniques to assist in the early detection of this condition.

• Preprocessing of retinal images
• Deep learning model for detection
• Evaluation metrics for performance
• Visualization of results

• Python
• TensorFlow/Keras
• OpenCV
• NumPy
• Pandas
• Matplotlib

The dataset used for this project can be obtained from the Kaggle Diabetic Retinopathy Detection competition. And the model used is trained using a ResNet-52 neural network architecture.

The model achieved a training accuracy of 96.01% and a validation accuracy of 93.2%, with a minimal loss of 0.136. On the test set, the accuracy was 92.32% with a loss of 0.149.