Exudate Regeneration for Automated Exudate Detection in Retinal Fundus Images

This paper presents a framework for the automated detection of Exudates, an early sign Diabetic Retinopathy. The introduces classification-extraction-superimposition (CES) mechanism enabling generation representative exudate samples based on limited open-source samples. demonstrates how manipulation Yolov5M output vector can be utilized extraction and super-imposition, segueing into development custom CNN architecture focused classification in retinal fundus images. performance proposed is compared with various state-of-the-art image architectures wide range metrics, including simulation post deployment inference statistics. A self-label presented, endorsing high developed architecture, achieving 100% test dataset.