Detection of a Mentally Challenged Person from their Expressions

Abstract--- The objective of the work is to identify the emotions of the special people with facial expression, speech
and body gestures. The emotion in mentally challenged people and the emotional disturbance are of same weightage
which has been computed with the psychiatric disorder. Recent research and theory in emotion strongly suggests
that there are two approaches namely the rational approach and physiotherapy which are particularly applicable to
the study of mental retardation. The existing model available for identifying the emotions of the mentally challenged
people uses separate models to understand the facial expressions, Speech and body gestures. The setup for these
models need more economical support as individual models has to be designed and implemented. In this paper, the
emotions of the special people is identified with their facial expressions, speech and body gestures. The model
proposed in this approach is using a Bayesian classifier, using a multimodal corpus with four emotions. From the
analysis the mentally challenged person’s expressions are entirely different from normal human’s expressions. The
quality of recognizing the speech will improve when the input speech is free from noise. The noise elimination could
be done using the voice activity detection (VAD) technique.