In diabetics, a condition known as diabetic retinopathy (DR) damages the retina and over timemay result in blindness. Ophthalmologists are now manually evaluating DR, which is a labor-intensive process. Furthermore, this work (project), which is a subset of artificial intelligence(AI), will now focus on studying distinct DR stages. To identify the DR stage and categorizethe 3662 training photos into high resolution fundus images, we trained a network called CNN on a sizable dataset. The APTOS dataset that we are using is hosted by Kaggle. Five DRphasesare available: zero, 1, 2, 3, and four. The enter parameters for this task are patient-provided fundus eye picture