Refurbished and improvised model using convolution network for autism disorder detection in facial images

Abstract

The main quality of deep learning over conventional machine learning (ML) techniques empowers firsthand uses in processing of images, speech recognition, medical imaging, machine translation and robotics, computer vision, and numerous other fields. The purpose of this study is to assess algorithms of deep Learning for person with the disorder of autism. This disorder is developing disorder that causes significant communicative, social and behavioral difficulties in those who have it. In this research paper, the Enhanced version of convolution network is discussed. Visual geometry group (VGG), is one of model of the convolution neural network which has essential features of convolution neural network (CNN). The VGG 16 net is employed to calculate the processes that can be used to classify this disorder with improved accuracy. The preprocessing of the image data is done. The VGG 16 convolution network is used to classify between autism spectrum disorder (ASD) and Non-ASD. Finally, the algorithm's efficacy is considered based on its accuracy performance. The VGG 16 net algorithm produces better results with an accuracy of 68.54%, compared with the normal CNN algorithm.