The development of a new edge computing device based on deep learning YOLOv8 algorithm for diabetic retinopathy Detection

Diabetic retinopathy is a prevalent eye disorder that occurs in individuals with diabetes. The condition is caused by damage to the retinal blood vessels caused by elevated blood sugar levels. The retina is the tissue behind the eye that is sensitive to light. Among the leading causes of blindness worldwide is diabetic retinopathy. However, early detection and prompt treatment can effectively prevent or minimize vision loss. Deep learning algorithms, such as “You Only Look Once” (YOLO) Convolutional Neural Networks (CNNs), have demonstrated state-of-the-art performance on benchmark datasets and computer vision competitions. YOLO is a single end-to-end model that enables real-time object identification with high accuracy. Our proposed YOLOv8-based model integrated with Jetson Nano Developer Kit with CSI camera module shows the highest top1 accuracy of 93.77 on the training dataset and validation dataset with all given classes its measured accuracy top1 accuracy is 72.6 with 2.6ms inference time.