Performance evaluation in content-based image retrieval:: overview and proposals

被引:333
作者
Müller, H [1 ]
Müller, W [1 ]
Squire, DM [1 ]
Marchand-Maillet, S [1 ]
Pun, T [1 ]
机构
[1] Univ Geneva, Comp Vis Grp, CH-1211 Geneva 4, Switzerland
关键词
content-based image retrieval; performance evaluation; information retrieval;
D O I
10.1016/S0167-8655(00)00118-5
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Evaluation of retrieval performance is a crucial problem in content-based image retrieval (CBIR). Many different methods for measuring the performance of a system have been created and used by researchers. This article discusses the advantages and shortcomings of the performance measures currently used. Problems such as defining a common image database for performance comparisons and a means of getting relevance judgments (or ground truth) for queries are explained. The relationship between CBIR and information retrieval (IR) is made clear, since IR researchers have decades of experience with the evaluation problem. Many of their solutions can be used for CBIR, despite the differences between the fields. Several methods used in text retrieval are explained. Proposals for performance measures and means of developing a standard test suite for CBIR, similar to that used in IR at the annual Text REtrieval Conference (TREC), are presented. (C) 2001 Elsevier Science B.V. All rights reserved.
引用
收藏
页码:593 / 601
页数:9
相关论文
共 42 条
  • [1] Aksoy S., 1999, Proceedings. 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No PR00149), P63, DOI 10.1109/CVPR.1999.786918
  • [2] *ANN, 1999, ANNOTATED GROUNDTRUT
  • [3] [Anonymous], 1966, Factors determining the performance of indexing systems
  • [4] Vols
  • [5] BELONGIE S, 1998, P INT C COMP VIS BOM
  • [6] Efficient content-based retrieval: Experimental results
    Berman, AP
    Shapiro, LG
    [J]. IEEE WORKSHOP ON CONTENT-BASED ACCESS OF IMAGE AND VIDEO LIBRARIES (CBAIVL'99) - PROCEEDINGS, 1999, : 55 - 61
  • [7] BORGMAN CL, 1989, INFORMATION PROCESSI, V25, P225
  • [8] Retrieval performance improvement through low rank corrections
    Comaniciu, D
    Meer, P
    Xu, K
    Tyler, D
    [J]. IEEE WORKSHOP ON CONTENT-BASED ACCESS OF IMAGE AND VIDEO LIBRARIES (CBAIVL'99) - PROCEEDINGS, 1999, : 50 - 54
  • [9] *COR, 1999, COREL CLIPART PHOTOS
  • [10] Cox IJ, 1996, ADV DIGITAL LIB ADL, P66