90 likes | 255 Views
Improving Human Computer Interaction in a Classroom Environment using Computer Vision. by Joshua Flachsbart, David Franklin and Kristian Hammond in Proceedings of the 5th international conference on Intelligent User Interfaces 2000. Overview.
E N D
Improving Human Computer Interaction in a Classroom Environment using Computer Vision by Joshua Flachsbart, David Franklin and Kristian Hammond in Proceedings of the 5th international conference on Intelligent User Interfaces 2000
Overview • Trying to improve HCI with multiple (visual) input modes • Discusses chosen imput modes, their constraints, contexts and utilisation Mika P. Nieminen
Environment: Intelligent Classroom • Confined space enables better sensing, plan (task) recognition and plan execution (helping) • Current supported functions: • Automatic filming of a lecture • Powerpoint presentation aids • VCR control • Person tracking, icon recognition Mika P. Nieminen
Tasks out of bulk data • Using contextual information, mostly from visual sensors, to locate the user and figure out their goals/tasks • Hand locations important for pointing and drawing command icons Mika P. Nieminen
Person tracking methods 1/2 • Background Subtraction Input Mode • Slower • Does not recognisedifferent users • Loses immobile users Mika P. Nieminen
Person tracking methods 2/2 • Color Histogram Input Mode • Users histogram must be taught • Does not track hands • Same engine as BSIM Mika P. Nieminen
Icon Recognition • Drawn on board • Remembers learnt icons • Recognition by matchingSalient Features • Video ’smart’ zooms inusing context Mika P. Nieminen
Future Work • Show that multiple input modes function better • Improvements under development • Allow switching modes on the run • Allowing input modes to adapt on the run (Example ’fading’ persons) Mika P. Nieminen
Conclusions • Sounds nice, might work, needs more study • Limited applications Mika P. Nieminen