Human body segmentation in a static image with multiscale superpixels

Abstract

In this work, we propose a new method for human body accurate detection in a static image with multiscale superpixels. The main contribution of this work is as follows: (1) A new framework for human body accurate detection using multiscale superpixels and classifier with autothreshold is proposed. (2) Some effective constraints for human body detection are given. Experimental results show the framework can accurately detect the human body in most situations. The dataset we use is a collection of photos of humans, varying in pose and clothing with complex backgrounds.