A Deep Learning Approach for Enhanced Detection of Diabetic Retinopathy Using a Modified Fast RCNN

In the present work a deep learning model for the detection of diabetic retinopathy has been proposed using faster RCNN algorithm. To address the challenges in detecting abnormalities within the human eye suffering from. Diabetic retinopathy which is a multi-stage disease prevalent in long-standing diabetes, complicates diagnosis due to its intricate nature. A total of 20 data set images were taken from Kaggle databases and subjected to the Fast RCNN algorithm to detect the efficiency of prediction. The results indicated a modal loss of 0.6 and an accuracy of 89.9. The training and testing models were able to converge after 12 epochs indicating the efficient prediction after applying deep learning techniques.