Salient object detection in low-light RGB-T scene via spatial-frequency cues mining

Huihui Yue, Jichang Guo, Xiangjun Yin, Yi Zhang, Sida Zheng

Abstract: Highlights•We propose the first network, called SFMNet, to study RGB-T salient object detection (SOD) for low-light images.•We make a pioneering attempt to introduce frequency cues into SOD task for low-light scenes.•We design a spatial-frequency feature exploration (SFFE) module to separate spatial-frequency features and adaptively select high and low-frequency features.•We design a spatial-frequency feature interaction (SFFI) module for cross- modality and cross-domain information aggregation.•We contribute a new dataset as a benchmark to explore RGB-T SOD in low-light images.