Immersion In Slideworld: A Methodological Case Study For Technology Selection And Evaluation

In the context of SlideWorld, a research project aiming at creating an immersive experience in videoconferencing, two video signal processing technologies have been developed and evaluated. A Smile Detector is used to increase the feeling of social presence and a Keyword Extractor allows focusing the attention on the video message. Those technologies have been evaluated for their intrinsic performances, but also in their contextual use in immersion with respect to users' feedbacks. Index Terms-- Keyword Extraction, Smile Detection, Video, Immersion 1. INTRODUCTION SlideWorld is an Alcatel-Lucent Bell Labs' research project, aiming at creating an immersive experience for end-users during videoconferences. Our idea is to identify the key moments of the videoconference and to emphasise them in order to maximize the attendees' attention. To do so, some signal processing technologies had to be identified, contextualized and evaluated. In this paper, we present some sociology and cognitive psychology state of the art that enable us to define the high level objectives of those technologies (see §2). Then an early evaluation of the users' needs is presented, using a qualitative methodology from ergonomics (see §3). The selected signal processing technologies ­smile detection and keyword extraction- are presented (see §4) and evaluated in order to refine the way they are used in SlideWorld to support the immersive experience (see §5). Some further enhancements for the technologies and their use in an immersive videoconference system are presented in the conclusion (see §6).