摘要Diabetes is a significant issue in the medical field.The detection and identification of the human eye diseases caused by excessive blood sugar levels in diabetes mellitus are important.The main objective of this study is to propose a viable solution for diagnosis using fundus images.This study presents a stage by stage implementation methodology.The original fundus image is first preprocessed,then the blood vessels are segmented,and finally the features are extracted and classified.This work uses an effective way to introduce a meta-heuristic algorithm.Blood Vessel Segmentation(BVS)is vital in DR(Diabetic Retinopathy)detection;hence,this research proposes a Firefly-Optimized Frangi based Filter(FOFF).Categorizing the disease is the last procedure.The classifier K-Nearest Neighbour(KNN)has an accuracy of 91.62％,while the SVM does well with an accuracy score of 95.54％.