Content extraction from video lectures for low power display

Abstract

The exponential growth and proliferation of the MOOC platform in recent years has facilitated easy access to advanced learning materials in the form of video lectures across almost all major specializations. The ever expanding outreach of internet has further boosted the utility and attraction of this medium of learning across the globe. Handheld devices like i-pad and e-notebooks can be used to access such platforms. Such portable devices however, are constrained by their battery capacity and hence need frequent charging. This can limit the access period for streaming or playing saved video-lectures, while traveling or while in a remote/rural set-up. Limited bandwidth may also obstruct the download/streaming of such material. In this work we present techniques that can drastically reduce the power budget for playing such video lectures on portable devices. Using image processing techniques we extract only the essential features from a video lectures, such that there is minimal loss in the information conveyed by the instructor, while the total amount of data transaction and processing is reduced by a large amount. The proposed method can therefore facilitate streaming/playing of video lectures for longer duration for a single charge.