Image Search Result Summarization with Informative Priors

Though current commercial image search engines provide effective ways to retrieve the relevant images, they are ineffective for users to find the desired from the retrieved hundreds of results, especially for ambiguous queries In this paper, we propose to summarize the search results by several representative images We argue that the relevance and image quality are two important measures for a user friendly summarization since image search results are normally noisy with some low-quality images The two factors, which can be regarded as informative prior of whether an image is a good summary candidate, are modeled into Affinity Propagation framework User studies demonstrate that our proposed method is able to produce a user friendly summary, in terms of relevance, diversity, and coverage.