CSRNet: Cascaded Selective Resolution Network for real-time semantic segmentation

作者：

Highlights：

• Multiple-stage segmentation network refines the feature maps progressively.

• Multiple context information embedding enlarges the receptive field at each stage.

• Selective Resolution Module aggregates multi-resolution feature maps.

• The proposed network alleviates the problem caused by multi-scale objects.

• The proposed system shows improved accuracy for real-time segmentation.

摘要

•Multiple-stage segmentation network refines the feature maps progressively.•Multiple context information embedding enlarges the receptive field at each stage.•Selective Resolution Module aggregates multi-resolution feature maps.•The proposed network alleviates the problem caused by multi-scale objects.•The proposed system shows improved accuracy for real-time segmentation.

论文关键词：Semantic segmentation,Attention mechanism,Real-time inference,Deep neural networks

论文评审过程：Received 27 June 2021, Revised 15 May 2022, Accepted 11 August 2022, Available online 17 August 2022, Version of Record 27 August 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118537