Attention-guided chained context aggregation for semantic segmentation

作者：

Highlights：

• Proposing a series-parallel hybrid structure of multiple information flows to enrich pixel representations.

• Adopting an extended attention model for feature aggregation.

• Constructing a general semantic segmentation model and achieving SOTA or competitive performance on benchmarks.

摘要

Highlights•Proposing a series-parallel hybrid structure of multiple information flows to enrich pixel representations.•Adopting an extended attention model for feature aggregation.•Constructing a general semantic segmentation model and achieving SOTA or competitive performance on benchmarks.

论文关键词：Semantic segmentation,Multi-scale contexts,Series-parallel hybrid streams,Convolutional networks

论文评审过程：Received 16 June 2021, Revised 16 August 2021, Accepted 11 September 2021, Available online 15 September 2021, Version of Record 22 September 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104309