Database Reference
In-Depth Information
outside the “O” region plays a constructive role in improving the recognition.
State-of-the-art algorithms were used as comparison benchmark, and it has been
demonstrated that the proposed method outperformed both the conventional CBIR
using original image query and the CIRM algorithm. Moreover, a recommendation
system is built upon an initial visual query input, where neither the text nor the
voice has the strength in describing the visual intent. Once the context metadata is
associated with the intent, more reliable contextual text and GPS features are taken
advantage of in searching and re-ranking. Ultimately, interesting and related social
activities are recommended to the users.
Search WWH ::
Custom Search