Detection of Conjunctivitis with Facial Images Improved Accuracy using a Hessian Matrix with RNN and CNN

Abstract

Convolutional Neural Network (CNN) classifiers are compared to Recurrent Neural Network (RNN) classifiers in the detection of conjunctivitis with facial images for improving accuracy using a Hessian matrix to improve system efficiency. The face data set used in this paper is the FERET face data set, which contains 200 individuals. Every person has two separate photographs of 120 people, taken at different times. For example, CNN with a variety of examples (N = 10) and RNN Classifier with a variety of examples (N = 10) techniques are used to detect conjunctivitis with facial images in order to improve accuracy using a Hessian matrix. CNN has a 95.71% accuracy rate, whereas RNN has a 91.62% accuracy rate. CNN has a precision rate of 95.03%, while the precision rate of recurrent neural networks (RNN) is 90.15%. CNN has a recall rate of 95.03%, while recurrent neural networks (RNN) have a recall rate of 90.34%. CNN has a specificity rate of 95.71%, while recurrent neural networks (RNN) have a specificity rate of 91.37%. The accuracy rate is significantly different (P 0.0581). When compared to RNN Classifier, the CNN Classifier predicts better classification in terms of detecting quality and reliability of conjunctivitis with facial images using the Hessian matrix.