A novel co-attention computation block for deep learning based image co-segmentation

作者：

Highlights：

• A novel co-attention block is proposed to compute the correlation between images.

• Top-k average pooling is presented to grasp the semantics in the feature responses.

• We achieve the currently best results of image co-segmentation on testing datasets.

摘要

•A novel co-attention block is proposed to compute the correlation between images.•Top-k average pooling is presented to grasp the semantics in the feature responses.•We achieve the currently best results of image co-segmentation on testing datasets.

论文关键词：Visual co-attention,Image co-segmentation,Deep learning,Correlation calculation,Average pooling

论文评审过程：Received 18 December 2019, Revised 23 June 2020, Accepted 25 June 2020, Available online 7 July 2020, Version of Record 18 July 2020.

论文官网地址：https://doi.org/10.1016/j.imavis.2020.103973