Natural activation for gesture recognition systems

Abstract

Gesture recognition is becoming a popular way of interaction, but still suffers of important drawbacks to be integrated in everyday life devices. One of these drawbacks is the activation of the recognition systemtrigger gesture -which is generally tiring and unnatural. In this paper, we propose two natural solutions to easily activate the gesture interaction. The first one requires a single action from the user: grasping a remote control to start interacting. The second one is completely transparent for the user: the gesture system is only activated when the user's gaze points to the screen, i.e. when s/he is looking at it. Our first evaluation with the 2 proposed solutions plus a default implementation suggests that the gaze estimation activation is efficient enough to remove the need of a trigger gesture in order to activate the recognition system.

Kinect-based Powered Wheelchair Control System

“…In this system, the disabled elder can use gesture [4][5] to control the wheelchair. The way of fulfilling this mission is to recognize different skeleton via Kinect. There are three gestures recognition in this system: normal gesture, wheelchair call gesture, and wheelchair parking gesture.…”

On the relation of ordinary gestures to TV screens

“…These movements could be used to interact with the TV and the disc player. The success of gesture interaction in gaming has yet to be realised in interaction with television content [1,5,15]. The question remains as to how to ensure smooth communication, whether by making the TV able to interpret viewers' 'natural' gestures and relate them to specific commands or by designing new kinds of gestures currently not found in TV viewing practices.…”

“…This has proved to be a surprisingly challenging task [1]. The design experiments also focus on individual use situations [2,3,5,6,7,15,17,19,21,28,38], although it is argued that television viewing is often a collaborative activity [4,11,12]. Promising to deliver gesture interaction to viewers, often in social settings, further extends the de-mands of technologies that promise 'natural' gesture interaction.…”