Recognition of Human Activity using Incremental SVM

Abstract

Human activity detection is recognize the human body activity from different variation in human body size, and different activity scope. This area of research is become more interesting because of various human body movement style. The main goal of human activity detection is to detect normal and abnormal action which helps to make human life more comfortable. The multi-class support vector machine is used in first step for initial classification and then each decision accept from new frame is used to update the classifier. In the field of human activity detection the detection of abnormal activity is still a research problem, with respect to large human movement. As in hospital monitoring patient and give special manpower to each patient is difficult and more costly hence the propose system helps to monitor abnormal patient activity and provide instant help to the patient. As there is lack of patient dataset we prepare a dataset and experiment the propose system.