Visual complexity analysis using deep intermediate-layer features
作者：
Highlights：
• Unsupervised extraction of information from convolutional layers of deep neural networks.
• Unsupervised Activation Energy (UAE) metric to quantify visual complexity.
• SAVOIAS, a dataset for the analysis of visual complexity.
• High correlation between our UAE method and ground truth.
• Within context of category, visually more complex images are more memorable to human.
摘要
•Unsupervised extraction of information from convolutional layers of deep neural networks.•Unsupervised Activation Energy (UAE) metric to quantify visual complexity.•SAVOIAS, a dataset for the analysis of visual complexity.•High correlation between our UAE method and ground truth.•Within context of category, visually more complex images are more memorable to human.
论文关键词：Visual complexity,Convolutional layers,Deep neural network,Feature extraction,Convolutional neural network,Activation energy,Memorability,Object classification,Scene classification
论文评审过程：Received 14 May 2019, Revised 9 January 2020, Accepted 11 March 2020, Available online 2 April 2020, Version of Record 9 April 2020.
论文官网地址：https://doi.org/10.1016/j.cviu.2020.102949
