Diabetes is a chronic disorder that affects the health of an individual by increasing blood sugar levels. Adverse effects of diabetes can cause severe eye-related problems like Diabetic Macular Edema (DME), leading to vision loss. DME is characterized by the formation of hard exudates in the macular region, causing damage to the retinal microvasculature. Early detection is, therefore, required to prevent vision loss, as DME has no permanent cure. A deep learning framework is utilized in the present work to detect DME. This is followed by preprocessing of the input image, such as cropping and resizing the image. The images of DME are obtained from the DME VQA dataset built from IDRiD and eOphta datasets. The model was trained and tested in the ratio 4:1. The accuracy obtained is 86.5% from the Dense-net model employed in the present work. Motivation: The clinical diagnosis of DME may cause human error as different clinicians have a chance to interpret symptoms differently, leading to variations in results in diagnosis, and limited information from a patient’s medical history can obstruct accurate results. The clinical diagnosis may be time-consuming, leading to patient treatment delays. Therefore, CNN models are used in detecting DME as they can analyze fundus images effectively by capturing abnormalities that indicate the presence of DME. CNN models help healthcare clinicians in many ways, including early detection of DME, providing accurate results with high efficiency, and allowing them to focus on patients’ further treatment planning and other complex tasks.