Hybrid content recognition from bitmap images

The Ypopsei project proposes the development of innovative hybrid object/place recognition methods from bitmap images using state-of-the-art machine learning approaches; Bag-of-Visual Words and Convolutional Neural Networks. In particular, within the framework the project tuning of hyperparameter is performed in order to define the optimum ones based on objectively quantifying and measuring the classification accuracy. Ultimately, the best performing in terms of accuracy hybrid method will be integrated in a pilot application that will be used within the cultural heritage domain. More specifically, the project’s case study is the Folklore Museum of Xanthi, where a personalized content-enriched tour will be offered using the visitor’s mobile phone to capture an image of an exhibit and content provision referring to the specific exhibit.