Edge Refinement Exploration for Salient Object Detection
Yinjie Wang, Yu Qiu, Jing J. Xu · 2023
Recently, salient object detection (SOD) has made significant progress. However, it is still challenging to make accurate predictions around the boundary regions. We introduce an Edge-Guided Salient Refinement Module (ESRe) to increase the detection accuracy near object boundaries, which extends Domain transform (DT) technology from the field of one-dimensional (1-D) signal processing to two-dimensional (2-D) image processing. By feeding into the coarse saliency map and the edge probability map, ESRe performs a convolution operation on the coarse saliency map to calculate the correction value. And then, ESRe refines the coarse saliency map to be better aligned with object boundaries, guided by the edge information which can be seen as the prior knowledge. Specifically, the edge probability map controls the relative contribution of the coarse saliency map and correction value to the refined saliency map. Extensive experiments on popular benchmarks demonstrate the E-Sal achieve competitive performance compared with state-of-the-art methods.