作者: Roohullah , J. Jaafar
DOI: 10.1007/978-3-642-25200-6_10
关键词:
摘要: The evolution in the digital technology from last few decades and different multimedia sources like Broadcast news, Movies, Videos, images, etc. have increased size volume of media daily. Due to this explosive growth volume, it's strongly urged for system that efficiently effectively compiles user demand retrieving relevant images. In paper, query will be expanded through an opensource knowledge base WordNet ConceptNet retrieves images on synonyms concepts. This technique covers word mismatch sense disambiguation (WSD) problem. Propose method applied open benchmark image dataset LabelMe. experimental results show propose techniques improved retrieval performance over traditional ones get accuracy up 94% single 83% sentence queries.