RLP-AGMC: Robust label propagation for saliency detection based on an adaptive graph with multiview connections

作者：

Highlights：

• An adaptive graph is constructed with multiview connections.

• An efficient and robust mechanism is presented to calculate seeds.

• Robust label propagation mechanism is devised to refine final result.

• A new BSOS salient object detection dataset is conducted.

• Performances outperform existing methods on various datasets using six metrics.

摘要

•An adaptive graph is constructed with multiview connections.•An efficient and robust mechanism is presented to calculate seeds.•Robust label propagation mechanism is devised to refine final result.•A new BSOS salient object detection dataset is conducted.•Performances outperform existing methods on various datasets using six metrics.

论文关键词：Deep features,Multiview connections,Graph affinity matrix,Label propagation,Salient object detection

论文评审过程：Received 19 September 2020, Revised 2 July 2021, Accepted 3 July 2021, Available online 14 July 2021, Version of Record 14 July 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116372