Saliency maps on image hierarchies

TL;DR

This paper introduces two novel hierarchical image segmentation-based saliency models that effectively generate high-quality saliency maps, achieving state-of-the-art results on benchmark datasets.

Contribution

The paper presents two new hierarchical saliency models that leverage image hierarchies for improved salient object segmentation.

Findings

Achieve state-of-the-art performance on benchmark datasets

Produce high-quality saliency maps

Effectively utilize hierarchical image structures

Abstract

In this paper we propose two saliency models for salient object segmentation based on a hierarchical image segmentation, a tree-like structure that represents regions at different scales from the details to the whole image (e.g. gPb-UCM, BPT). The first model is based on a hierarchy of image partitions. The saliency at each level is computed on a region basis, taking into account the contrast between regions. The maps obtained for the different partitions are then integrated into a final saliency map. The second model directly works on the structure created by the segmentation algorithm, computing saliency at each node and integrating these cues in a straightforward manner into a single saliency map. We show that the proposed models produce high quality saliency maps. Objective evaluation demonstrates that the two methods achieve state-of-the-art performance in several benchmark datasets.