Detecting Autism Spectrum Disorder Using DenseNet

Abstract

Autism spectrum disorder (ASD) is a degenerative sickness which affects brain development that ultimately destroys the physical appearance of the face. ASD is one of the most frequent acute neurodevelopmental illnesses in the world today (ASD). It is a lifelong disorder that hampers an individual’s conduct and communication abilities. The paper suggests a unique approach for identifying ASD using a machine classifier. Additionally, machine learning (ML) classifier models provide ASD class types coupled with evaluation criteria. DenseNet was established partly to address the worsening accuracy in high-level neural networks caused by the removal of the gradient. ASD is a sort of mental condition that may be discovered through the study of social media data and biological imaging. A recent study has shown that ASD may be detected merely by utilizing facial pictures. DenseNet machine learning models are pre-trained in this work to categorize face photos as either healthy or perhaps autistic. Autism may be diagnosed by traits such as eye, nose, and lip distance in an image and layout. Machine learning techniques may be used to recognize such landmarks, but pinpoint technology for retrieving and reconstructing the right patterns from data given. By deploying a simple web tool based on a machine learning algorithm, our study supports the medical facilities in recognizing autism based on the features of the face.