Cross-Language and Cross-Media Image Retrieval: An Empirical Study at ImageCLEF2007

This paper summarizes our empirical study of cross-language and cross-media image retrieval at the CLEF image retrieval track (ImageCLEF2007). In this year, we participated in the ImageCLEF photo retrieval task, in which the goal of the retrieval task is to search natural photos by some query with both textual and visual information. In this paper, we study the empirical evaluations of our solutions for the image retrieval tasks in three aspects. First of all, we study the application of language models and smoothing strategies for text-based image retrieval, particularly addressing the short text query issue. Secondly, we study the cross-media image retrieval problem using some simple combination strategy. Lastly, we study the cross-language image retrieval problem between English and Chinese. Finally, we summarize our empirical experiences and indicate some future directions.