Object localization and edge refinement network for salient object detection

作者：

Highlights：

• Building a detection network combining a transformer and a CNN.

• Designing an Information Multiple Selection module to capture useful information.

• The network is trained with two edge labels to further preserve object boundaries.

• The network has higher accuracy and efficiency than most recent networks.

摘要

•Building a detection network combining a transformer and a CNN.•Designing an Information Multiple Selection module to capture useful information.•The network is trained with two edge labels to further preserve object boundaries.•The network has higher accuracy and efficiency than most recent networks.

论文关键词：Transformer features,Information multiple selection,Object localization and edge refinement,Dual edge labels

论文评审过程：Received 10 June 2022, Revised 25 September 2022, Accepted 2 October 2022, Available online 8 October 2022, Version of Record 20 October 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118973