Micro aneurysms Detection Using Faster R-CNN network

Authors

  • Manojkumar S B, Dr. H S Sheshadri

Abstract

Prolong Diabetic Retinopathy (DR) disease leads for visual loss all over world. Regular screening and an early detection and treatment for this disease are necessary. Among many types and classifications of diabetic retinopathy (DR), Microaneurysms (MA’s) perceives during the early stage of diabetic retinopathy and hence it is very difficult process to detection the Microaneurysms (MA’s). During the period of one decade, many approaches applied for the detection and identification of the DR using some mathematical and image processing algorithms, features extraction techniques, and artificial neural network classification. In which few are failed during the pre-processing stage, feature extraction stage, vessels extraction and during the classification stage. In this paper it is clearly identified Microaneurysms during the early stage by developing the regions of a fundus image to show the particular region in terms of its severity level by collected the large data from kaggle, DIARETDB1dataset and then pre-trained the models and applied to deep learning classifiers using faster R-CNN model will successfully achieve a better performance on Diabetic retinopathy & MA detection and getting the performance merit of 82% accuracy.

Published

2020-11-01

Issue

Section

Articles