Automated detection of Diabetic Retinopathy (DR) could be a potential solution for the detection of DR at the early stages, thereby minimizing the requirement of expensive and complicated surgeries. Recent advancement in the field of artificial intelligence and computing devices has fueled the research in these areas. Large databases (KAGGLE) available resulted in the application of convolution neural networks (CNN) and its variants (ImageNet) for accurate DR detection. In this paper implementation of deep learning architecture in Raspberry Pi is proposed to classify DR. Proposed system classifies DR into five different categories - No DR, Mild, Moderate, Severe and Proliferative depending on the severity of the problem. 35126 images are used for training (90%) and validation (10%) of the model. The model gave high accuracy during the testing phase and system implementation. The proposed scheme using the Raspberry PI system to test the images acquired, thereby reducing the overall cost and this facilitates the detection of diabetic retinopathy by the deployment of the system in primary health centers.
Detection and Classification of Diabetic Retinopathy using Raspberry PI
2020-11-05
1045877 byte
Conference paper
Electronic Resource
English
INTRODUCTION TO DIABETIC RETINOPATHY AND EARLY DIABETIC RETINOPATHY
British Library Conference Proceedings | 2006
|VISUAL FIELDS AND DIABETIC RETINOPATHY
British Library Conference Proceedings | 2006
|LASER TREATMENT OF DIABETIC RETINOPATHY
British Library Conference Proceedings | 2006
|Detection and decision-support diagnosis of diabetic retinopathy using machine vision
British Library Online Contents | 2011
|