中北大学学报（自然科学版）2018，Vol.39Issue(4)：467-473,7.DOI:10.3969/j.issn.1673-3193.2018.04.017
基于深层卷积神经网络的图像美感度分类
Classification of Image Aesthetic Degree Based on Deep Convolution Neural Network
摘要
Abstract
Aiming at the problems such as low accuracy and poor description of aesthetic features in im-age aesthetic classification,an algorithm based on deep convolutional neural network for image aesthet-ics classification was proposed.Firstly,the image was input into 5 5-layer convolution neural network to automatically learn ,and obtained more detailed and deep-level aesthetic characteristics,and then the classification of image aesthetic degree was carried out by softmax classifier to obtain the best classifica-tion results.The experimental results show that the proposed algorithm achieves 80.13% and 87.32% accuracy in A1and A0databases,respectively,and achieves better classification accuracy in six scenarios of CUHKPQ database.关键词
卷积神经网络/特征提取/图像美感度分类Key words
convolution neural network/feature extraction/classification of image aesthetic degree分类
通用工业技术引用本文复制引用
杨国亮,曾建尤,王志元..基于深层卷积神经网络的图像美感度分类[J].中北大学学报（自然科学版）,2018,39(4):467-473,7.基金项目
国家自然科学基金资助项目(51365017) （51365017）
