Bridging the gap: effect of text query reformulation in multimodal retrieval

摘要: Multimodal Retrieval provides new paradigms and methods aimed at effectively searching through the enormous volume of data. retrieval is a well studied problem often used in image retrieval. Most existing works under pretext multimodality stress on bridging semantic gap by using both textual visual features. In this paper, we use relevance feedback from user-generated documents associated with images for expanding query study its effect text We employ topic decomposition based keyphrase extraction technique to expand queries. Our results articulate fact that an insightful expansion always improves performance or Also, adopt optimum weight learning scheme combine modalities privileged way. perform comparative two established techniques which are expansion. A detailed set experiments standard real world dataset also carried out same.