Abstract:
Autism Spectrum Disorder (ASD) is neurocognitive disorder which has been on rise worldwide. A person
with such condition have difficulties in responding to speech, social interaction and unable to utter words.
The learning activities for them revolves around inanimate structures, objects and pictures handling. With
such condition it becomes improbable for children with such condition to recognize a face and respond to
it. Taking attendance for such specialized children become a time consuming task. So in this paper we have
proposed a methodology of automated attendance of such specialized child to ensure the security of each
child and time which is saved by such automatic technology is spent in concentrating on teaching them
activities to get acquainted with social interaction capabilities. The proposed methodology is done in three
stages. First is detecting Face from a video using dlib library based on Histogram of Oriented Gradients
(HOG) and then classifying using Support Vector Classifier(SVC).In the next stage, face detected is aligned
using FaceAligner and facial features are extracted based on 68 landmark point face detector and trained
into database. In final stage, faces from live feed webcam is detected, its feature being extracted and
compared with database to record attendance.
