Exploring geometric information in CNN for image retrieval

被引:0
作者
Ying Li
Xiangwei Kong
Haiyan Fu
机构
[1] Dalian University of Technology,School of Information and Communication Engineering
[2] Zhejiang University,Department of Data Science and Engineering Management
来源
Multimedia Tools and Applications | 2019年 / 78卷
关键词
Image retrieval; Spatial pooling; Feature weighting;
D O I
暂无
中图分类号
学科分类号
摘要
Convolutional Neural Network (CNN) has brought significant improvements for various multimedia tasks. In contrast, image retrieval has not yet benefited as much since no training database is available. In this paper, we propose an unsupervised weighting scheme for pre-trained CNN models to adaptively emphasize image center. Different from the general preference for fully connected layers which represent abstract semantics, we aggregate the activations of convolutional layers on image patches to depict local patterns in details. It is an empirical observation that the target of searching is naturally the focus of an image. Thus we pooling the features with respect to their positions, since they innately maintain the geometric layout of an image. Experimental results on two benchmarks prove the effectiveness of our methods.
引用
收藏
页码:30585 / 30598
页数:13
相关论文
共 51 条
  • [1] Bai S(2016)Sparse contextual activation for efficient visual re-ranking IEEE Trans Image Process 25 1056-1069
  • [2] Bai X(2017)Improving large-scale image retrieval through robust aggregation of local descriptors IEEE Trans Pattern Anal Mach Intell 39 1783-1796
  • [3] Husain SS(2010)Improving bag-of-features for large scale image search Int J Comput Vis 87 316-336
  • [4] Bober M(2012)Aggregating local image descriptors into compact codes IEEE Trans Pattern Anal Mach Intell 34 1704-1716
  • [5] Jégou H(2004)Distinctive image features from scale-invariant keypoints Int J Comput Vis 60 91-110
  • [6] Douze M(2017)Local residual similarity for image re-ranking Inform Sci 417 143-153
  • [7] Schmid C(2017)Effective multi-query expansions: Collaborative deep networks for robust landmark retrieval IEEE Trans Image Process 26 1393-1404
  • [8] Jégou H(2017)Unsupervised metric fusion over multiview data by graph random walk-based cross-view diffusion IEEE Trans Neural Netw Learn Syst 28 57-70
  • [9] Perronnin F(2018)Deep adaptive feature embedding with local sample distributions for person re-identification Pattern Recogn 73 275-288
  • [10] Douze M(2015)Query specific rank fusion for image retrieval IEEE Trans Pattern Anal Mach Intell 37 803-815