Prediction of Autism Spectrum Disorder in Children using Face Recognition
Sajeev Ram Arumugam, R. Balakrishna, Rashmita Khilar, S. Oswalt Manoj, C.S. Shylaja
Abstract
Autism Spectrum Disorder (ASD) is a disability in neurological development, leading to delay in communication and behavioural issues at the two years of children lifetime and continuous until adulthood. This paper has used Convolution Neural Network [CNN] to classify the facial images into two classes namely ASD affected and Normal images, which helps us to identify whether the child has ASD. The early the detection of ASD, the more the issues being rectified soon by giving therapy to such children to improve their social and behavioural issues. This research work has used a publicly available dataset from Kaggle website, and both training and testing was performed in the ratio 70:30. Finally, the developed neural network based model has gained the ability to achieve an accuracy rate of 91% and the loss value is identified to be 0.53 respectively.