Social media data and biological pictures may be used to diagnose autism spectrum disorder (ASD), a kind of mental disease. Autism spectrum disorder (ASD) is a neurodevelopmental illness that alters the facial appearance as a person matures. Autism spectrum disorder (ASD) children are easily distinguishable from normally developing (TD) youngsters due to noticeable differences in facial landmarks. The planned study is novel since it would use facial recognition and social media to identify people with autism spectrum disorder. While deep learning approaches hold promise for identifying such landmarks, doing so would need very accurate technology for extracting and creating the appropriate patterns of facial data. This research aids communities and psychiatrists in experimentally identifying autism using face traits by using a simple web application built on a deep learning system, namely, a convolutional neural network with transfer learning and the flask framework. The classification challenge was performed using the pretrained models Xception and Resnet50. The few face photos used to evaluate these models were gathered via the Kaggle site.