CDADNet: Context-guided dense attentional dilated network for crowd counting
作者：
Highlights：
• Crowd counting in image is easily affected by complex background.
• Context-guided module provides rich contextual information.
• Euclidean loss, cross-entropy loss and adaptive density-level loss are combined.
• The attention mechanism and dilated convolution architecture are fused.
• The dense attentional dilated module generates high-quality density maps.
摘要
•Crowd counting in image is easily affected by complex background.•Context-guided module provides rich contextual information.•Euclidean loss, cross-entropy loss and adaptive density-level loss are combined.•The attention mechanism and dilated convolution architecture are fused.•The dense attentional dilated module generates high-quality density maps.
论文关键词：Crowd counting,Density map,Dense dilated,Attention
论文评审过程：Received 27 November 2020, Revised 24 May 2021, Accepted 5 July 2021, Available online 10 July 2021, Version of Record 14 July 2021.
论文官网地址：https://doi.org/10.1016/j.image.2021.116379
