Retinal pictures are essential for a number of uses, including the identification of people and the discovery of diseases. Detection of the blood vessels that are present in the retinal image is very helpful for an ophthalmologist to easily diagnose the fundus image. Also based on the status of the retinal blood vessels, early identification of diabetes can be achieved. This paper presents an empirical analysis of three supervised models—K Nearest Neighbour (K-NN), Support Vector Machine (SVM), and Decision Tree (DT) in detecting retinal blood vessels Grey Scaling, Gaussian Blur, and Gamma Correction methods are used to extract features. To measure the efficacy performance of the supervised models evaluation metrics—Precision, Recall and F1-Score are used. The highest accuracy is obtained in case using Decision Tree Algorithm as 95.75%.