Abstract: Existing salient object detection (SOD) models primarily resort to convolutional neural networks (CNNs) and Transformers. However, the limited receptive fields of CNNs and quadratic ...