Abstract Retinal images are a key factor for ophthalmologists in the diagnosis of several eye diseases, with a sign of a microvascular retina, which occurs in response to high blood pressure and diabetes. Periodic ophthalmoscopy is the best approach for eye disease screening. However, the number of ophthalmologists available is a limiting factor in initiating screening. With the availability of digital fundus cameras, for automatic analysis of digital images, it can help ophthalmologists diagnose eye disease. On the other hand, computer science, especially in-depth learning has the ability to identify objects in many domains, one of which is the identification of objects in images including images of the retina of the eye. This chapter explains the application of depth learning approach to classification of eye diseases based on color fundal images as inputs. Subsequently, in the Fundus Imaging subchapter, we will discuss the acquisition of relevant Images, major vascular and nonvascular abnormalities, assessment of hypertensive retinopathy and diabetes. In the last section, we implemented convolutional neural networks (CNN) as classifiers for retinal images. The public database used in this study is retinal color images from the STARE dataset which are resized to 61 × 70, 46 × 53, and 31 × 35 pixels. This dataset is categorized into 15 classes of eye diseases. Our experiments show that the highest training accuracy is obtained from input images measuring 61 × 71 and 31 × 35 pixels. The highest test accuracy obtained from the network configuration used is the size of the input image with the smallest resolution of 31 × 35 pixels with an accuracy of 80.93%.