Scalable feature-based video retrieval for mobile devices

Abstract

The authors present a novel method for video retrieval on mobile devices. The target scenario is the following: features are extracted from a captured query video at the user side and transmitted to a remote server; then, video retrieval is applied within a stored database at the server side and relevant information about the query is returned to the user. In particular, we focus on the user side and propose a scalable method for video feature extraction and encoding taking into consideration the processing capabilities of the device and available bandwidth. Despite these constraints, the first results show a promising accuracy of retrieval.