Human Detection with Non Linear Classification Using Linear SVM

Abstract

Human detection is one of the major topics of vision research and has lots of application such as surveillance, robotics, pedestrian detection etc. Detecting human beings expeditiously is the major problem faced in many research works. Support vector machine is one of the techniques used for human detection and among them linear and non linear SVM are most popular. The performance of linear SVM decreases significantly for large set data. Non linear SVM are more promising but they are more computationally expensive. This paper presents the study of classifying the human posture like non linear SVM classification using number of linear SVM. The target space is divided into sub space by each linear SVM which result in non linear classification boundary of human posture. This paper is a review of pricewise linear support vector machine for detecting human being in various environment.