Leveraging Deep Learning for Early Stage Diabetic Retinopathy Detection: a Novel CNN and Transfer Learning Comparision

Diabetic retinopathy (DR) is a contributor, to vision impairment highlighting the need, for accurate and efficient screening methods. While traditional techniques are prevalent, deep learning approaches offer promising alternatives. In this research paper, A new Convolutional Neural Network (CNN) design was proposed and that focuses on detecting Diabetic Retinopathy (DR) from retinal images. The evaluation of the effectiveness of the proposed CNN is comparing it with two known trained models, namely VGG 19 and ResNet 101 which are commonly used in image classification tasks. The proposed CNN model has the highest performance with respect to the other two models was based on the performance metrics(Accuracy, Precision, Recall, F1-score) . The paper gives an overview of how the proposed CNN model is built.