基于深度学习的图像艺术属性分类
|
摘要
针对图像分类中各类图像艺术属性人工特征描述能力的有限性,提出了一种基于深度残差网络和Inception学习模型的图像艺术属性分类的算法。该算法主要借鉴ResNet50和Inception V3模型,构建"一次性"深度学习网络和多类二叉树分层分类深度学习网络,分别对不同艺术属性图像进行分类测试。实验结果表明,"一次性"深度学习网络与传统的卷积神经网络构建的分类器相比,图像分类正确率有了一定的提升;同时,在相同的多类二叉树分层分类顺序下,该深度学习网络与采用SVM作为分类器所构成的分类模型相比,分类正确率明显提升。
According to the fact that the characteristics extracted manually has limited description ability in traditional image classification with all kinds of artistic attributes, an algorithm of image recognition with different kinds of artistic attributes based on deep residuals network and Inception model is proposed. The algorithm uses ResNet50 and Inception V3 models to make the one trial deep learning network and binary tree multi-class classification deep learning network. Experimental result shows that compared with the traditional convolutional neural network method, one trial deep learning network has an better classification performance. At the same time, the classification accuracy of the binary tree multi-class classification algorithm we proposed is greatly improved compared with the algorithm using SVM as classifier in the same binary tree frame.
According to the fact that the characteristics extracted manually has limited description ability in traditional image classification with all kinds of artistic attributes, an algorithm of image recognition with different kinds of artistic attributes based on deep residuals network and Inception model is proposed. The algorithm uses ResNet50 and Inception V3 models to make the one trial deep learning network and binary tree multi-class classification deep learning network. Experimental result shows that compared with the traditional convolutional neural network method, one trial deep learning network has an better classification performance. At the same time, the classification accuracy of the binary tree multi-class classification algorithm we proposed is greatly improved compared with the algorithm using SVM as classifier in the same binary tree frame.
引文
[1] BERG T,BELHUMEUR P N.Poof:part-based one-vs.-one features for fine-grained categorization,face verification,and attribute estimation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2013:955-962.
[2] YANG S L,BO L F,WANG J,et al.Unsupervised template learning for fine-grained object recognition[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems-Volume 2.New York:ACM,2012:3 122-3 130.
[3] YAO B,BRADSKI G,LI F F.A codebook-free and annotation-free approach for fine-grained image categorization[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2012:3 466-3 473.
[4] HINTON G E,SALAKHUTDINOV R R.Reducing the dimensionality of data with neural networks[J].Science,2006,313(5 786):504-507.
[5] HINTON G E,OSINDERO S,TEH Y W.A fast learning algorithm for deep belief nets[J].Neural Computation,2006,18:1 527-1 554.
[6] GONG Y C,WANG L W,GUO R Q,et al.Multi-scale orderless pooling of deep convolutional activation features[C]//European Conference on Computer Vision.Berlin:Springer International Publishing,2014:392-407.
[7] 石祥滨,阳柯,张德园.基于对称神经网络的图像分类[J].小型微型计算机系统,2017,38(3):578-583.
[8] 陈小娥,陈昭炯.多类 SVM 在图像艺术属性分类中的应用研究[J].智能系统学报,2009,4(2):157-162.
[9] 盛家川.中国传统水墨画的计算机分类与算法研究[D].天津:天津大学,2013.
[10] HE K M,ZHANG X Y,REN S Q,et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2016:770-778.
[11] SZEGEDY C,VANHOUCKE V,IOFFE S,et al.Rethinking the inception architecture for computer vision[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2016:2 818-2 826.
[12] SZEGEDY C,LIU W,JIA Y Q,et al.Going deeper with convolutions[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2014:1-9.
[2] YANG S L,BO L F,WANG J,et al.Unsupervised template learning for fine-grained object recognition[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems-Volume 2.New York:ACM,2012:3 122-3 130.
[3] YAO B,BRADSKI G,LI F F.A codebook-free and annotation-free approach for fine-grained image categorization[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2012:3 466-3 473.
[4] HINTON G E,SALAKHUTDINOV R R.Reducing the dimensionality of data with neural networks[J].Science,2006,313(5 786):504-507.
[5] HINTON G E,OSINDERO S,TEH Y W.A fast learning algorithm for deep belief nets[J].Neural Computation,2006,18:1 527-1 554.
[6] GONG Y C,WANG L W,GUO R Q,et al.Multi-scale orderless pooling of deep convolutional activation features[C]//European Conference on Computer Vision.Berlin:Springer International Publishing,2014:392-407.
[7] 石祥滨,阳柯,张德园.基于对称神经网络的图像分类[J].小型微型计算机系统,2017,38(3):578-583.
[8] 陈小娥,陈昭炯.多类 SVM 在图像艺术属性分类中的应用研究[J].智能系统学报,2009,4(2):157-162.
[9] 盛家川.中国传统水墨画的计算机分类与算法研究[D].天津:天津大学,2013.
[10] HE K M,ZHANG X Y,REN S Q,et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2016:770-778.
[11] SZEGEDY C,VANHOUCKE V,IOFFE S,et al.Rethinking the inception architecture for computer vision[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2016:2 818-2 826.
[12] SZEGEDY C,LIU W,JIA Y Q,et al.Going deeper with convolutions[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2014:1-9.