Diabetic retinopathy is a condition related to diabetes mellitus that damages the eyes. The retina, which is the light-sensitive tissue in the rear of the eye, sustains damage to its tiny blood vessels, which leads to the condition. Diabetes- related elevated blood sugar levels have the potential to weaken and damage these blood vessels over time, which can result in a variety of alterations to the structure and function of the retina. A common chronic illness that affects people all over the world, diabetes mellitus, is typified by high blood sugar levels. Damage to the blood vessels in the retina is the main cause of diabetic retinopathy (DR), which manifests as a serious consequence. In the early stages, DR may not cause any symptoms, but if left untreated, it can result in irreversible blindness. The manual study of cases by doctors in traditional detection methods takes a lot of time and is prone to inaccuracy. DR detection and diagnosis have become more efficient with the use of machine learning and deep learning approaches, which take advantage of technological advances. DR can be detected and classified at several phases, such as mild, moderate, severe, and proliferative, with the help of convolutional neural networks (CNNs), in particular, which have proven effective in interpreting retinal eye scans. In this study, the ResNet-152 architecture trained on the APTOS BLINDNESS DATASET 2019, consisting of fundus images, demonstrated a noted accuracy of 86% in predicting and classifying stages of diabetic retinopathy. These findings highlight the potential of deep learning approaches, particularly ResNet-152, in improving diabetic retinopathy diagnosis. This research offers a promising avenue for enhancing patient outcomes and reducing the risk of vision loss associated with diabetes.