Autism Action Recognition and Severity Analysis to Avoid Overdosage of Drugs Based on IP Webcam App through Deep Transfer Learning

Authors

  • B.Prabha, N.R.Shanker ,M.Priya ,E.Ganesh

Abstract

Child with Autism Spectrum Disorder (ASD) consumes too much or too little of drugs leads to severe side effects such as kidney failure, irregular heart rate, and fluctuating blood pressure. However, clinical literate/therapist could not continuously monitor autistic child behavior for drug prescription or to avoid over/under dosage. In this paper, we propose IP web camera based continuous monitoring system for ASD children action recognition to assist therapist for accurate dosage of drug prescription. The proposed method consists of pre-trained model with deep convolution neural network (CNN) and transfer learning for autistic child behavior analysis. IP camera system based Autismdata.Net image dataset is created, from ASD children in the natural environment. The dataset consists of ASD children images with 10 different behaviors, due to adverse drug effects.

Published

2020-06-30

Issue

Section

Articles