Salient object detection network with multi-scale feature refinement and boundary feedback

作者:

Highlights:

• We propose a novel multi-scale refinement network for edge-aware saliency detection.

• We design a feature interaction module to represent multi-scale features.

• A feature refinement module is proposed to enhance feature discriminability.

• We design a boundary feedback method to effectively learn edge information.

• The proposed method performs competitive against the recent representative methods.

摘要

Highlights•We propose a novel multi-scale refinement network for edge-aware saliency detection.•We design a feature interaction module to represent multi-scale features.•A feature refinement module is proposed to enhance feature discriminability.•We design a boundary feedback method to effectively learn edge information.•The proposed method performs competitive against the recent representative methods.

论文关键词:Salient object detection,Saliency detection,Multi-scale features,Boundary,Attention

论文评审过程:Available online 24 October 2021, Version of Record 9 November 2021.

论文官网地址:https://doi.org/10.1016/j.imavis.2021.104326