An Unsupervised Game-Theoretic Approach to Saliency Detection

TL;DR

This paper introduces an unsupervised, game-theoretic method for salient object detection that leverages multiple features and an iterative random walk to produce accurate saliency maps without labeled data.

Contribution

It formulates saliency detection as a non-cooperative game and combines color and deep features through an iterative random walk, advancing unsupervised detection methods.

Findings

Outperforms state-of-the-art supervised algorithms on 6 datasets

Effectively combines multiple features for improved detection

Demonstrates robustness across challenging scenarios

Abstract

We propose a novel unsupervised game-theoretic salient object detection algorithm that does not require labeled training data. First, saliency detection problem is formulated as a non-cooperative game, hereinafter referred to as Saliency Game, in which image regions are players who choose to be "background" or "foreground" as their pure strategies. A payoff function is constructed by exploiting multiple cues and combining complementary features. Saliency maps are generated according to each region's strategy in the Nash equilibrium of the proposed Saliency Game. Second, we explore the complementary relationship between color and deep features and propose an Iterative Random Walk algorithm to combine saliency maps produced by the Saliency Game using different features. Iterative random walk allows sharing information across feature spaces, and detecting objects that are otherwise very hard to detect. Extensive experiments over 6 challenging datasets demonstrate the superiority of our proposed unsupervised algorithm compared to several state of the art supervised algorithms.