Python/Vk: An Efficient Video Handling Development System

Abstract

We developed a programming environment Python/Vk for efficient research of video handling technologies. Python/Vk provides various video event detectors such as shot-boundary detector and camera motion estimator. We can easily develop applications using Python/Vk that enables to decode frame images and audio data from encoded video files, to make indexes for structuring video and content-based retrieval, and to edit summaries of the video. Combination with open source libraries of Python such as OpenCV, NumPy and SciPy gives more advanced functionalities to the applications. We describe the architecture of Python/Vk, the productivity comparing with C++, and example applications.