This document describes research using a convolutional neural network to detect diabetic retinopathy from fundus images. The researchers trained a CNN model on a dataset of over 35,000 fundus images to classify images into five stages of diabetic retinopathy severity. The CNN model extracts features from input fundus images and uses activation functions and optimization algorithms to output a classification. The classification along with patient details will generate a standardized report on diabetic retinopathy detection and diagnosis.