Local object detection and recognition in mobile devices

Applications which required powerful machines can now be executed in small devices, such as mobile phones or Personal Digital Assistants (PDA). The increasing number of mobile devices with embedded digital cameras introduces new possibilities and challenges, particularly, in the area of computer vision. This paper proposes a method for real-time object recognition using mobile devices that runs locally without the need to communicate with a server. Both object detection and identification algorithms are performed by the mobile device avoiding the communication with the server. These algorithms are based on variance comparisons in near regions and on Support Vector Machines (SVM) and Center- Symmetric Local Binary Patterns (CS-LBP). The paper presents results that show the effectiveness of the method.