Classification of Uncertain ImageNet Retinal Diseases using ResNet Model

Abstract

An automated method is required to detect retinal diseases. The goal of this study is to build automated screening methods for retinopathy (DR) and other eye diseases using deep learning. In this study present an approach for detecting retinal illness using a CNN. In order to find these patterns, CNN may be employed by extracting features from the data. A color Fundus image-based technique is proposed by the authors of this study to detect retinal disorders. Without prompt diagnosis and treatment, retinal disorders may cause permanent vision loss. The condition must be diagnosed at an early stage in order to get the right therapy and cure it. Deep learning models may be used to train and test the data in order to classify different retinal disorders, where several common retinal diseases and conditions are classified and normalized. This study looked at how retinal pictures may be used to classify eye illnesses using CNN. Over eight different retinal illnesses are included in this dataset, which applied a CNN model to Pretrained on Classification of uncertain ImageNet Retinal Diseases using ResNet Model has been implemented, CNN is built for retina pictures with varying task functions and depths. Various filtering and pooling strategies are tried and shown to have a significant impact on network performance. This is possible because employing a convolutional neural network to process the retinal pictures. It has been discovered that this suggested method has an accuracy rate of more than 80%.