Cross-modality Complementary Interaction Network

Salient Object Detection (SOD) aims to identify the most conspicuous objects and regions in the human vision. This paper presents the Cross-Modality Complementary Interaction Network(CCINet), a novel approach for obtaining complementary features between different modalities. We introduce feature complementarity Module (FCM) to capture inter-modality complementary features. Subsequently, the Multi-Scale Aggregation Module (MAM) is utilized to acquire multilevel information across different resolutions. Finally, we employ a decoder to generate saliency maps. Experimental results demonstrate the superiority of our model over previous methods, as it outperforms them across four evaluation indicators on five challenging datasets.