Classification of Autism Disorder using Deep Learning Techniques: A BlazePose-based Approach

Conditions that include pervasive developmental disorder, autism, childhood disintegrative disorder, and Asperger's syndrome are all included under the umbrella term of autism spectrum disorder. Even today, some people still refer to “Asperger's syndrome,” which is generally considered a lesser form of autism spectrum condition. Early childhood is when autism spectrum disorder usually first manifests itself, and it eventually causes problems with social integration, academic achievement, and professional functioning in society. To develop a method using as Inception V3 algorithm to easily determine the presence of the disease. Video-based classification is used to detect autism spectrum disorder. To help in the effective diagnosis of disease with higher accuracy. A minority of children seem to experience typical development in their first year of life but undergo a phase of regression between 18 and 24 months of age, during which they exhibit the symptoms of autism. This study utilizes a Deep Learning (DL) algorithm such as Inception V3 for the early detection of a disorder. Datasets, comprising videos, have been assembled for identifying autism spectrum disorder. A deep learning algorithm is applied to train the videos of children, resulting in the creation of a model file. When an input image is provided for disease prediction, the model effectively identifies the presence of the disorder. Therefore, the proposed system provides an effective solution to predict the presence of autism spectrum disorder in a more efficient way. The findings show that the LSTM classification model achieved 93% accuracy. Thus, this research work helps in the effective diagnosis of autism spectrum disorder with higher accuracy than existing models.