Earticle

초록

영어

In this paper, we introduce a simple but efficient method to construct powerful image representation via spatially weighted deep convolutional feature aggregation for image retrieval. First, the convolutional features are extracted from the convolutional layer of the CNNs and proposed spatial weights are applied to the convolutional features. A sum pooling is performed to aggregate the spatially weighted convolutional features into the global image representation. We carry out extensive experiments on Oxford and Paris Building datasets and experiment results show that the proposed method achieves competitive performance compared to current state-of-the-art methods.

목차

Abstract
1. Introduction
2. Related Work
2.1 Traditional image retrieval
2.2 CNN-based image retrieval
3. Spatially Weighted Convolutional Feature Aggregation
4. Experiments
4.1 Datasets
4.2. Evaluation metrics
4.4 Implementation details
4.5 Experiment results
5. Conclusion
Acknowledgement
References

키워드

저자

• Enkhbayar Erdenee [ Department of Computer and Information Engineering, Inha University, Incheon, South Korea ]
• Sanggil Kang [ Department of Computer and Information Engineering, Inha University, Incheon, South Korea ]

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  정보화연구 [정보화연구(구 정보기술아키텍처연구)]

• 간기

  계간

• pISSN

  1738-382X

• 수록기간

  2004~2026

• 등재여부

  KCI 등재

• 십진분류

  KDC 325 DDC 658