Classification of Idol Photography Based on Pose Guide Ontology

Abstract

The objective of this work is to classify poses of idols wearing swimsuits in the photograph. We propose classification system, which takes unannotated idol photos as inputs and classifies their poses based on the spatial layout of idol in the photo. Our system has two phases; the first phase is to estimate the spatial layout of ten body parts (head, torso, upper/lower arms/legs) based on Eichner's Stickman Pose Estimation method. The second phase is to classify the poses of idols in the photo using Bayesian Network classifier. In order to improve the accuracy of classification, we introduce Pose Guide Ontology (PGO). PGO contains useful background knowledge such as semantic hierarchies and constraints related to the orientation and positional relationship between the body parts.