Semantic Part Segmentation using Compositional Model combining Shape and Appearance

TL;DR

This paper introduces a compositional model that combines shape and appearance cues for semantic part segmentation of animals, addressing challenges of similar appearance and shape variability, with a new learning algorithm and efficient inference method.

Contribution

It proposes a novel mixture of compositional models integrating multiple cues and a learning algorithm for animal part segmentation under various poses and viewpoints.

Findings

Effective segmentation of horse and cow parts demonstrated on Pascal VOC 2010 dataset.

The model achieves accurate boundary delineation despite appearance similarities.

Efficient inference algorithm reduces computational complexity.

Abstract

In this paper, we study the problem of semantic part segmentation for animals. This is more challenging than standard object detection, object segmentation and pose estimation tasks because semantic parts of animals often have similar appearance and highly varying shapes. To tackle these challenges, we build a mixture of compositional models to represent the object boundary and the boundaries of semantic parts. And we incorporate edge, appearance, and semantic part cues into the compositional model. Given part-level segmentation annotation, we develop a novel algorithm to learn a mixture of compositional models under various poses and viewpoints for certain animal classes. Furthermore, a linear complexity algorithm is offered for efficient inference of the compositional model using dynamic programming. We evaluate our method for horse and cow using a newly annotated dataset on Pascal VOC 2010 which has pixelwise part labels. Experimental results demonstrate the effectiveness of our method.