Suppress and Balance: A Simple Gated Network for Salient Object Detection

TL;DR

This paper introduces GateNet, a gated network architecture for salient object detection that effectively manages information flow between encoder and decoder, improving detection accuracy across multiple datasets.

Contribution

The paper proposes a simple gated network with multilevel gate units and a dual branch structure to enhance feature cooperation and saliency map detail restoration.

Findings

Outperforms most state-of-the-art methods on five datasets

Effectively controls information interference between encoder and decoder

Accurately localizes salient objects of various scales

Abstract

Most salient object detection approaches use U-Net or feature pyramid networks (FPN) as their basic structures. These methods ignore two key problems when the encoder exchanges information with the decoder: one is the lack of interference control between them, the other is without considering the disparity of the contributions of different encoder blocks. In this work, we propose a simple gated network (GateNet) to solve both issues at once. With the help of multilevel gate units, the valuable context information from the encoder can be optimally transmitted to the decoder. We design a novel gated dual branch structure to build the cooperation among different levels of features and improve the discriminability of the whole network. Through the dual branch design, more details of the saliency map can be further restored. In addition, we adopt the atrous spatial pyramid pooling based on the proposed "Fold" operation (Fold-ASPP) to accurately localize salient objects of various scales. Extensive experiments on five challenging datasets demonstrate that the proposed model performs favorably against most state-of-the-art methods under different evaluation metrics.