Review on Pupil Segmentation Using CNN-Region of Interest

Abstract

Segmentation is a core concept in image-processing applications. Its applications were widely used in to computer vision, surveillance, robotic applications and medical imaging. Due to the rapid growth of deep learning techniques and their efficiency, a number of applications started building on these algorithms. Due to the pandemic situation caused by COVID-19, most of the industries are looking forward to developing applications based on irises for various reasons. In our literature, we focused on traditional techniques and convolutional neural network (CNN) based techniques to show accuracy. The pupil is the smallest part of the eye. In an image it is a barely traceable object. The current traditional systems are well versed with circle detection but have proven to have high loss of accuracy in many implementations. Our intension is to focus on traditional methods of pupil detection and CNN-based segmentation to provide accuracy and to minimize the false negative rate (FNR).