AI-Driven Diabetic Retinopathy Detection: IoT and Deep Learning for Retinal Fundus Image Analysis

Diabetic retinopathy is one of the main causes of blindness in diabetics and early detection and classification are essential for both prevention and successful treatment. The subjective and time-consuming nature of traditional diagnostic techniques emphasizes the need for automated, accurate, and efficient systems. Using IoT devices that capture real-time retinal fundus images and combining them with patient metadata including age, gender, and blood glucose levels, this study suggests a novel deep learning-based method for the detection and classification of DR. The suggested approach combines a number of advanced techniques, including U-Net for image segmentation, InceptionV3 for feature extraction, and Vision Transformer for classification. To improve quality and eliminate noise, the images are preprocessed using Median Filtering and CLAHE. Evaluation was conducted using the IDRiD, which consists of 516 fundus images with DR grades ranging from 0 (no DR) to 4 (proliferative DR). With an accuracy of 98.55%, precision of 97.89%, recall of 98.01%, and F1 score of 97.94%, the experimental results show the effectiveness of the model. This study offers a possible way to prevent blindness in diabetic patients by demonstrating the potential of integrating deep learning with IoT devices for the early and precise diagnosis of DR.