Diabetic Retinopathy (DR) is a frequent complication of diabetes mellitus characterized by the development of lesions or abnormalities on the retina, affecting vision and potentially resulting in irreversible blindness if not detected early. While traditional methods of diagnosing DR through manual examination by ophthalmologists are labor-intensive, resource-intensive, time-consuming and vulnerable to inaccuracies, Computer-Assisted Diagnosis systems offer a more efficient alternative. Techniques in Deep Learning particularly convolutional neural networks (CNNs) have emerged as highly impactful tools in analyzing healthcare images, specifically for detecting and classifying DR from color fundus images. This article presents an extensive review and analysis of contemporary developments in this field, exploring available datasets and addressing key challenges that warrant further investigation.