Child autism diagnosis using deep learning-based facial expression analysis

Ch.Lavanya Susanna , U.Dhanalakshmi , A.Anusha , P.Harshitha , B.Harsha

Machine Learning models, Deep Learning models, Autism Spectrum Disorder, Facial features extraction

Autism Spectrum Disorder (ASD) is a group of neurodevelopmental diseases associated with behavior, social interaction, and communication. Within the first two years of life, or during developmental phases, the disease's symptoms typically manifest. There are two approaches to ASD diagnosis and rehabilitation. The first is the manual method, based on observation or interviews that primarily entails the analysis of behavioral symptoms. The other approach makes use of EEG readings, brain MRIs, and conventional machine learning (ML) for automatic diagnosis. ASD cannot currently be diagnosed using a diagnostic test, which makes the diagnosis difficult. This paper has explored the early detection of Autism Spectrum Disorder (ASD) by identifying autistic children by face feature recognition using a Convolutional Neural Network. The accuracy of the suggested method is 99%, which is higher than the results of existing systems like SVM and mobileNet algorithms, which only provide 70% accuracy. With the use of a Deep Learning-based strategy that incorporates face analysis, our findings should greatly help researchers, therapists, psychologists, and other pertinent stakeholders in the advancement of ASD screening, monitoring, and diagnosis.

Volume 9 Issue 2, February-2024

Pages : 208 - 216