基于多层特征描述及关系学习的智能图像情感识别  

Image emotion recognition based on multi-levels features representation and relationship learning

作　　者：杨文武 普园媛[1] 赵征鹏[1] 徐丹[1] 钱文华[1] 阿曼 YANG Wenwu;PU Yuanyuan;ZHAO Zhengpeng;XU Dan;QIAN Wenhua;A Man(School of Information Science and Engineering, Yunnan University, Kunming 650000, Yunnan, China)

机构地区：[1]云南大学信息学院

出　　处：《陕西师范大学学报（自然科学版）》2019年第5期40-48,共9页Journal of Shaanxi Normal University：Natural Science Edition

基　　金：国家自然科学基金(61163019,61271361,61462093,61761046,U180227);云南省科技厅项目(2014FA021,2018FB100);云南省教育厅科学研究项目(2018JS011)

摘　　要：针对缺少标记的情感图像数据会严重影响卷积神经网络(convolutional neural network,CNN)性能的问题,利用半监督动态学习的方法建立了大规模的图像情感数据集--Large-scale deep emotion(LSDE)数据集。为了有效弥补图像特征和人类情感之间的差异,先将图像目标与背景进行分离,之后使用关系学习网络获得基于前景和背景图像的不同层级间的关系。在LSDE数据集、Twitter2数据集以及ArtPhoto数据集上的实验结果表明,关系学习网络能够有效地提取图像的多层级特征并学习到不同层级特征之间的关系,弥补图像特征和人类情感的差异,提高图像情感识别的准确率。Aimed at the problem of lacking of emotions images will influence the performance of CNN seriously, a large image emotion dataset, Large-scale deep emotion(LSDE) dataset, is built using semi-supervised dynamic method to ensure emotion labels of images. In order to bridge the gap and find the relationship between different levels features effectiveness, at first, objects of images are divided from complete images using salient object detection, then a relationship learning network is adopted based on foreground images and background images in this paper to learn relationships between different levels features of images to bridge the gap between image features and emotions of human beings. Experimental results in LSDE dataset, Twitter2 dataset and ArtPhoto dataset show that relationship learning network can extract different level features from foreground and background images and can learn the relationship between different level features and bridge the gap between features of images and emotions of human beings, the image emotion recognition system also can recognize emotions of images accurately.

关 键 词：图像情感识别 多层级图像特征 关系学习网络 CNN 人工智能 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]