Visual aesthetic understanding: Sample-specific aesthetic classification and deep activation map visualization

作者：

Highlights：

•

• A sample-specific classification model to weight the importances of different samples is proposed.

• The model simultaneously conducts classification and visualization of aesthetic information.

• Aesthetic belief map is used to explain what we have learned in aesthetic understanding, and also can be used to study image cropping.

摘要

•A sample-specific classification model to weight the importances of different samples is proposed.•The model simultaneously conducts classification and visualization of aesthetic information.•Aesthetic belief map is used to explain what we have learned in aesthetic understanding, and also can be used to study image cropping.

论文关键词：Visual aesthetic quality assessment,Aesthetic understanding,Sample-specific weighting

论文评审过程：Received 18 July 2017, Revised 8 May 2018, Accepted 8 May 2018, Available online 29 May 2018, Version of Record 29 May 2018.

论文官网地址：https://doi.org/10.1016/j.image.2018.05.006