Download PDFOpen PDF in browser

Real Time Implementation of Emotion Detection and Classification system for Autism Patients

EasyChair Preprint 2207

7 pagesDate: December 19, 2019

Abstract

Autism is a neurodevelopment disorder characterized by impaired social interaction with verbal and nonverbal communication. Autism is usually diagnosed in the children of age group 4 to 15 years. 1500 out of 1 million children are diagnosed with autism problems. Many autistic patients have impairment in understanding other people's emotions which hinder their interpersonal communication. There is no effective solution to aid autistic patients in understanding the expression of the people with whom they are communicating. Problem with expression detection system is to be able to make a person independent model and to be able to make an efficient system in classifying the emotions into one of the following expressions such as anger, smile, surprise, neutral, sad etc.

In this paper appearance based method is used for extracting the facial features. This paper describes the different facial expressions of humans, real-time face detection, extraction of features from the face and finally a decision based system is built to classify the different facial expressions. The methodology involves resizing the video in to 256*256 pixels using Lanczos resizing filter, grey scale conversion is carried out by luminosity method, from which face is detected using Viola and Jones algorithm it uses Haar features for face detection, further Eigen values are extracted from the detected face which is used for classification.  In this system 215 images have been trained and the range of Eigen value for classification is determined. Hence, from the range of values obtained, efficiency of facial expression detection for smile, anger, surprise, sad and neutral are 90%, 83%, 80%, 80% and 90% respectively. Thus, the autistic patients will be notified with expression detected which will be effective for their communication.

Keyphrases: Clustering, Eigen values, Grey Scale, Vila Jones, luminosity method

BibTeX entry
BibTeX does not have the right entry for preprints. This is a hack for producing the correct reference:
@booklet{EasyChair:2207,
  author    = {Mohana and H. V. Ravish Aradhya},
  title     = {Real Time Implementation of Emotion Detection and Classification system for Autism Patients},
  doi       = {10.29007/7d8x},
  howpublished = {EasyChair Preprint 2207},
  year      = {EasyChair, 2019}}
Download PDFOpen PDF in browser