Saliency detection in human crowd images of different density levels using attention mechanism

作者：

Highlights：

• The result of knowledge-driven gaze in the human visual system is used to predict saliency in crowd.

• Self-attention models the relationship between elements in the high-level features.

• Multi-scale feature extraction with self-attention is effective in high-density crowd image.

摘要

•The result of knowledge-driven gaze in the human visual system is used to predict saliency in crowd.•Self-attention models the relationship between elements in the high-level features.•Multi-scale feature extraction with self-attention is effective in high-density crowd image.

论文关键词：Saliency,Human crowd,Deep neural network,Attention mechanism

论文评审过程：Received 8 September 2019, Revised 5 August 2020, Accepted 7 August 2020, Available online 15 August 2020, Version of Record 20 August 2020.

论文官网地址：https://doi.org/10.1016/j.image.2020.115976