Dual Mutual Learning Network with Global-local Awareness for RGB-D Salient Object Detection

TL;DR

This paper introduces GL-DMNet, a dual mutual learning network with global-local awareness for RGB-D salient object detection, effectively addressing modality discrepancies and leveraging global and local information for improved performance.

Contribution

The paper proposes a novel dual mutual learning network with specific modules for inter-modality fusion and a transformer-based decoder, advancing RGB-D SOD performance.

Findings

Outperforms 24 state-of-the-art methods on six datasets

Achieves ~3% average improvement over second-best model

Demonstrates effective fusion of global and local features

Abstract

RGB-D salient object detection (SOD), aiming to highlight prominent regions of a given scene by jointly modeling RGB and depth information, is one of the challenging pixel-level prediction tasks. Recently, the dual-attention mechanism has been devoted to this area due to its ability to strengthen the detection process. However, most existing methods directly fuse attentional cross-modality features under a manual-mandatory fusion paradigm without considering the inherent discrepancy between the RGB and depth, which may lead to a reduction in performance. Moreover, the long-range dependencies derived from global and local information make it difficult to leverage a unified efficient fusion strategy. Hence, in this paper, we propose the GL-DMNet, a novel dual mutual learning network with global-local awareness. Specifically, we present a position mutual fusion module and a channel mutual fusion module to exploit the interdependencies among different modalities in spatial and channel dimensions. Besides, we adopt an efficient decoder based on cascade transformer-infused reconstruction to integrate multi-level fusion features jointly. Extensive experiments on six benchmark datasets demonstrate that our proposed GL-DMNet performs better than 24 RGB-D SOD methods, achieving an average improvement of ~3% across four evaluation metrics compared to the second-best model (S3Net). Codes and results are available at https://github.com/kingkung2016/GL-DMNet.