Diabetic retinopathy (DR) is derived from diabetes is a serious eye disease and in the most common cause of blindness. DR is an eye disease that is a result of chronic disease of diabetes. A microaneurysm is a small red spot in the retina that raises from the fragile part of the blood vessel, hard exudates, and abnormal growth of blood vessels are the diabetic retinopathy. Predicting the presence of microaneurysmsas an early indication of fundus imaging and diabetic retinopathy has been a major challenge for decades. DR is suffering from a person's chronic high blood sugar levels, microvascular problems, and irreversible vision loss leads.To solve this problem, the rapid advancement of Deep Learning (DL) makes it an effective technique for providing interesting solutions to analytical problems in medicine.The proposed system DL algorithms are Retinal Hidden Linear Selection (RHLS) algorithm and Quick Convolutional Diagnosis (QCD)algorithm. Retinal Hidden Linear Selection (RHLS) algorithm used to pre-process to remove noise from the fundus image, and excessive-performance. Then, a quick convolution diagnosis (QCD) algorithm is a classification to determine whether the fundus image is subjected to a normal or influence. The results obtained are the proposed method, from a very effective and color fundus image shows a successful diabetic retinopathy diagnosis.