Early Diagnosis of ASD based on Facial Expression Recognition and Head Pose Estimation.

Autism Spectrum Disorder (ASD) is one of the most common developmental disorders characterized by impairment of social interaction and communication skills, as well as stereotype behaviors. The early diagnosis has been focused on the analysis of EEG and MRI, which requires sophisticated medical equipment and the data collection process is cumbersome. Based on the differences of appearance characteristics between ASD and Typical Development (TD) children, we propose an efficient and effective method to diagnose autistic patients via facial expression recognition and head pose estimation. We apply the Conformer network to facial expression and head pose, respectively, to extract both local features and global features. In addition, we process the extracted features with accumulative histogram and adopt Long Short-Term Memory (LSTM) for classification. We verify the performance of the proposed method on our self-collected ASD video dataset (ACVD) and achieve a classification accuracy of 97.59%.