Frequency-aware feature aggregation network with dual-task consistency for RGB-T salient object detection

Heng Zhou, Chunna Tian, Zhenxi Zhang, Chengyang Li, Yongqiang Xie, Zhongbo Li

Abstract: Highlights•We propose FFANet to provide a novel frequency perspective for multi-spectral salient feature aggregation and RGB-T SOD.•Specifically, we propose a FMFA to explore the separability and complementarity of two spectra in the frequency domain, generating multi-spectral HF and LF cues.•In addition, to effectively alleviate the rough masks caused by blurred boundaries, we propose a HF-SDM and a dual task consistency learning method.