A Novel Bag-of-Object Retrieval Model To Predict Image Relevance

Text based image search is now a days a routine work for which generally, image search re-ranking and image result summarization these two effective approaches are used. But these approaches are not suitable for the object queries. In this system, a novel bag-of-object retrieval model is designed to predict image relevance, which is specifically effective for object queries. In this approach, first, an object vocabulary is constructed which contains query-relative objects based on expanded query sets which considers the frequent object patches in the result image collection. Then training model is constructed on the basis of frequent object patches. The efficiency of the proposed model is demonstrated by comparing the results with image search re-ranking and image search result summarization.