 Hypertensive retinopathy, HR, is a serious eye disease caused by high blood pressure. It leads to changes in the retinal arteries, resulting in cotton wool patches, bleeding in the retina, and constricted retinal arteries. Ophthalmologists can make a diagnosis of HR by analyzing fundus images to identify its stages and symptoms. Early detection of HR can reduce the risk of vision loss. Computer aided diagnostic, CADX, systems have been developed to detect HR using machine learning, ML, and deep learning, DL, techniques. However, these systems face challenges such as class imbalance, high computational complexity, and lack of lightweight feature descriptors. To address these issues, we propose a pre-trained transfer learning, TL, based mobile net architecture integrated with dense blocks to optimize the network for the diagnosis of HR-related disease. Our proposed mobile HR system achieves an accuracy of 99% and an F1 score of 0.99 on different data sets. The results were verified by an expert ophthalmologist, indicating that our proposed mobile HR system produces positive outcomes.