Abstract: Hashing and Feature Detection methods have demonstrated to be useful for a variety of tasks and have appealed extensive tending in recent years. Various feature detection and hashing approaches have been purported to capture similarities between textual, visual, and cross-media information. However, most of the existing works use a bag-of-words methods to represent textual information and for feature detection use the SIFT algorithm. Since words with different forms may have similar meaning, semantic level text similarities cannot be well processed in these methods. To address these challenges, in this paper, we propose a novel method called semantic cross-media hashing (SCMH), which uses uninterrupted word representations to entrance the textual similarity at the semantic level and use a deep belief network (DBN) to fabricate the correlation between different sense modality and a technical report on feature detection and carrying out a SURF algorithm. To manifest the potency of the proposed method, we evaluate the proposed method on three commonly used cross-media data sets are used in this work. Experimental results show that the proposed method attains significantly better performance and speed than state-of-the-art approaches. Moreover, the efficiency of the proposed method is comparable to or better than that of some other feature detection and hashing methods.
Keywords: Hashing method, SURF, Word Embedding, Fisher Vector