Salient Objects in Clutter: Bringing Salient Object Detection to the Foreground

TL;DR

This paper critically evaluates existing salient object detection models, identifies dataset biases, and introduces a new comprehensive dataset that better reflects real-world scenes to improve model robustness.

Contribution

The paper reveals biases in current datasets, proposes a new balanced dataset called SOC with challenging attributes, and updates the benchmark for more realistic salient object detection evaluation.

Findings

Existing models perform well on biased datasets but poorly on real-world scenes.

The SOC dataset includes images with both salient and non-salient objects from daily categories.

Attribute-based performance analysis highlights challenges in real-world scenarios.

Abstract

We provide a comprehensive evaluation of salient object detection (SOD) models. Our analysis identifies a serious design bias of existing SOD datasets which assumes that each image contains at least one clearly outstanding salient object in low clutter. The design bias has led to a saturated high performance for state-of-the-art SOD models when evaluated on existing datasets. The models, however, still perform far from being satisfactory when applied to real-world daily scenes. Based on our analyses, we first identify 7 crucial aspects that a comprehensive and balanced dataset should fulfill. Then, we propose a new high quality dataset and update the previous saliency benchmark. Specifically, our SOC (Salient Objects in Clutter) dataset, includes images with salient and non-salient objects from daily object categories. Beyond object category annotations, each salient image is accompanied by attributes that reflect common challenges in real-world scenes. Finally, we report attribute-based performance assessment on our dataset.