FARES: Fast and Accurate Recognition of Exact Scenes on Mobile Devices

Mobile devices represented by smartphones have been continuously evolving to support various applications. In this paper, we study a new mobile application, named exact scene recognition, where a user can identify a particular place by comparing two images taken there. This application framework enables users to annotate a scene supporting more descriptive image-based interactions such as mobile augmented-reality applications. We enhance the regular approaches with the assistance of the angle-of-view (AOV) information obtained from the smartphone. Our experimental results show a significant improvement on accuracy compared to the existing solutions.