Key points are not available for this paper at this time.
This paper introduces a web image search reranking approach that explores multiple modalities in a graph-based learning scheme. Different from the conventional methods that usually adopt a single modality or integrate multiple modalities into a long feature vector, our approach can effectively integrate the learning of relevance scores, weights of modalities, and the distance metric and its scaling for each modality into a unified scheme. In this way, the effects of different modalities can be adaptively modulated and better reranking performance can be achieved. We conduct experiments on a large dataset that contains more than 1000 queries and 1 million images to evaluate our approach. Experimental results demonstrate that the proposed reranking approach is more robust than using each individual modality, and it also performs better than many existing methods.
Building similarity graph...
Analyzing shared references across papers
Loading...
Meng Wang
Tongji University
Hao Li
Guizhou University
Dacheng Tao
Nanyang Technological University
IEEE Transactions on Image Processing
Chinese Academy of Sciences
University of Chinese Academy of Sciences
University of Vermont
Building similarity graph...
Analyzing shared references across papers
Loading...
Wang et al. (Thu,) studied this question.
synapsesocial.com/papers/6a10fb30d06b5b96589feff1 — DOI: https://doi.org/10.1109/tip.2012.2207397