Ophthalmologists use retinal images to determine eye disease, for example glaucoma disease. This diagnosis is made by measuring changes in the cup-to-disc ratio. Eye Cup Segmentation Helps Fossil Ophthalmologists Calculate CDRs of Retinal Background Images. In this article, we propose a machine learning approach to perform segmentation tasks using the U-Net architecture. This proposed technique was evaluated using 650 colour retinal fundus images. Next, U-Net has 160 opportunities. Additionally, we compared the segmentation results with ground reality images. According to the test results, the eyecup's segmentation performance reaches 98.42% of the membrane, with a loss of 0.15. Fundus images help detect glaucoma. In the past, several experiments have been conducted to detect this disease. This paper proposes an ideal framework aimed at building intelligent solutions that help identify glaucoma using MATLAB software tools. The diagnosis of glaucoma is made by measuring changes in the bowl-to-disc ratio. Eye Cup Segmentation Helps Fossil Ophthalmologists Calculate CDRs of Retinal Background Images. As we propose a deep learning method using U-Net architecture to perform segmentation tasks, we collected many fundus images as data. Subsequent pre-processing of the occlusion figure is accompanied by classification and segmentation. The pre-processing module uses several filters such as Gaussian filter and guided image filter to remove noise. The division into healthy and sick eyes helps with further administration. © 2024 Elsevier B.V., All rights reserved.