An Overview of Retinal Image Classification-Techniques and Challenges

Abstract

Retinal image classification has been into existence since decades. However, the biomedical industry has suffered over benchmark accuracy datasets. The availability of good quality data points is one of the main reasons for the absence of complete automation in retinal image processing. In this paper we try to delineate the different architectures utilised in retinal image processing. The utility of transformer in transplant techniques for data creation and classification is widely recognized. Finally, we compare different models and analyse their merits and demerits. The paper aims to not only provide a literature survey for all the different mechanisms available, but also put forth the promising areas for tomorrow.