Diabetic　retinopathy　(DR)　is　a　major　reason　for　the　increased　visual　loss　globally,　and　it　became　an　important　cause　of　visual　impairment　among　people　in　25-74　years　of　age.　The　DR　significantly　affects　the　economic　status　in　society,　particularly　in　healthcare　systems.　When　timely　treatment　is　provided　to　the　DR　patients,　approximately　90%　of　patients　can　be　saved　from　visual　loss.　Therefore,　it　becomes　highly　essential　to　classify　the　stages　and　severity　of　DR　for　the　recommendation　of　required　treatments.　In　this　view,　this　paper　introduces　a　new　automated　Hyperparameter　Tuning　Inception-v4　(HPTI-v4)　model　for　the　detection　and　classification　of　DR　from　color　fundus　images.　At　the　preprocessing　stage,　the　contrast　level　of　the　fundus　image　will　be　improved　by　the　use　of　contrast　limited　adaptive　histogram　equalization　(CLAHE)　model.　Then,　the　segmentation　of　the　preprocessed　image　takes　place　utilizing　a　histogram-based　segmentation　model.　Afterward,　the　HPTI-v4　model　is　applied　to　extract　the　required　features　from　the　segmented　image　and　it　subsequently　undergoes　classification　by　the　use　of　a　multilayer　perceptron　(MLP).　A　series　of　experiments　take　place　on　MESSIDOR　(Methods　to　Evaluate　Segmentation　and　Indexing　Techniques　in　the　field　of　Retinal　Ophthalmology)　DR　dataset　to　guarantee　the　goodness　of　the　HPTI-v4　approach　and　the　obtained　results　clearly　exhibited　the　supremacy　of　the　HPTI-v4　model　over　the　compared　methods　in　a　significant　way.