Salient object detection based on edge‐interior feature fusion

Recently, existing FCNs-based methods have shown their advantages in processing object boundaries. However, these still suffer from false interference, which appears saliency predictions. To solve this problem, an edge-interior feature fusion (EIFF) framework is proposed, consists of internal-boundary decoupled generation structure with receptive field enlargement and attention mechanism enhancement, a salient refinement module. Specifically, the first learns edge features interior through decoupling network, supervised by labels obtained ground-truth image erosion algorithm. Then, module (FRM) designed to purify coarse prediction focusing on ambiguous regions mining strategy generate final map. compensate for shortcomings BCE IU loss, we also introduce weighted loss guide our model focus more error-prone parts. Experimental results five benchmark datasets demonstrate that proposed method performs favorably against 19 state-of-the-art approaches under four standard metrics.