Hierarchical Spatial Sum-Product Networks for Action Recognition in Still Images

TL;DR

This paper introduces hierarchical spatial sum-product networks that model spatial configurations of body parts for action recognition in still images, improving accuracy by capturing complex part relationships.

Contribution

It proposes a novel hierarchical spatial SPN framework that models spatial relationships and high-order correlations of parts for action recognition.

Findings

Effective on two benchmark datasets

Models pairwise spatial relationships of parts

Captures high-order correlations of parts

Abstract

Recognizing actions from still images is popularly studied recently. In this paper, we model an action class as a flexible number of spatial configurations of body parts by proposing a new spatial SPN (Sum-Product Networks). First, we discover a set of parts in image collections via unsupervised learning. Then, our new spatial SPN is applied to model the spatial relationship and also the high-order correlations of parts. To learn robust networks, we further develop a hierarchical spatial SPN method, which models pairwise spatial relationship between parts inside sub-images and models the correlation of sub-images via extra layers of SPN. Our method is shown to be effective on two benchmark datasets.