Autism Spectrum Disorder Detection using Facial Expression

Abstract

Autism Spectrum Disorder (ASD) is a complex neurological disorder that has an impact on communication, language, and social skills. Early identification of ASD patients, particularly in children, could make it easier to design and implement the best therapy approach at the appropriate time. Analyzing facial characteristics, eye contact, and other aspects of human faces can be used to detect ASD. To better accurately identify children with ASD in the early stages, an improved transfer-learning-based autism face recognition framework is proposed in this paper. This study will improve the accuracy of ASD detection and classification of normal and autistic, using machine learning and deep learning approaches. This study detects and classifies Autistic and non-autistic human faces, Using the Deep learning-based CNN model, the study also analyzes the pre-trained transfer learning approaches with the proposed model. Results reveal that the proposed model has better detection and classification results having 99 % Accuracy. Based on our accuracy we propose that the diagnosis of autism spectrum disorders can be done effectively using facial images.