Automated Classifier Development Process for Recognizing Book Pages from Video Frames

One of the latest developments made by publishing companies is introducing mixed and augmented reality to their printed media (e.g. to produce augmented books). An important computer vision problem that they are facing is classification of book pages from video frames. The problem is non-trivial, especially considering that typical training data is limited to only one digital original per book page, while the trained classifier should be suitable for real-time utilization on mobile devices, where camera can be exposed to highly diverse conditions and computing resources are limited. In this paper we address this problem by proposing an automated classifier development process that allows training classification models that run real-time, with high usability, on low-end mobile devices and achieve average accuracy of 88.95% on our in-house developed test set consisting of over 20 000 frames from real videos of 5 books for children. At the same time, deployment tests reveal that the classifier development process time is reduced approximately 16-fold.