AI based Classification for Autism Spectrum Disorder Detection using Video Analysis

Abstract

Autism spectrum Disorder(ASD) is a complex neurobehavioral disorder that affects a person's ability to communicate and interact with others. It is also characterized by repetitive behaviors and restricted interests. There is no one-size-fits-all approach to autism, but early intervention and treatment can make a big difference in a person's life. Machine learning and deep learning are two promising areas of research that may help to improve our understanding of autism and lead for better treatments. Machine learning and Deep Learning approaches of artificial intelligence allows computers to learn from data without being explicitly programmed. These models could potentially be used to improve our ability to communicate with, and understand people with autism. Various machine-learning techniques are used to predict autism at an early stage. Support Vector Machine (SVM), Decision tree, Naïve Bayes, Random Forest, Logistic Regression, and K-Nearest Neighbour are some of the machine learning techniques used in this research area. Various advancement in the field of machine learning and Artificial Intelligence (AI) has helped in the development of ASD Detection using Machine learning and Deep Learning. In this research work, the prediction of Autism Spectrum Disorder has been performed on a video dataset. The video dataset contains the video of Autistic and Non-Autistic kids performing four different actions. The video features have been extracted through Convolutional Neural Network(CNN) models such as Inception V3and Resnet50 and are trained through long Short Term Memory(LSTM) based models by using this we get 91 % accuracy.