youtouch-teaser

We present YouTouch!, a system that tracks users in front of an interactive display wall and associates touches with users. With their large size, display walls are inherently suitable for multi-user interaction. However, current touch recognition technology does not distinguish between users, making it hard to provide personalized user interfaces or access to private data. In our system we place a commodity RGB + depth camera in front of the wall, allowing us to track users and correlate them with touch events. While the camera’s driver is able to track people, it loses the user’s ID whenever she is occluded or leaves the scene. In these cases, we re-identify the person by means of a descriptor comprised of color histograms of body parts and skeleton-based biometric measurements. Additional processing reliably handles short-term occlusion as well as assignment of touches to occluded users. YouTouch! requires no user instrumentation nor custom hardware, and there is no registration nor learning phase. Our system was thoroughly tested with data sets comprising 81 people, demonstrating its ability to re-identify users and correlate them to touches even under adverse conditions.

Publications

  • YouTouch! – Low-Cost User Identification at an Interactive Display Wall.

    von Zadow, U.; Reipschläger, P.; Bösel, D.; Sellent, A.; Dachselt, R.;

    In Proceedings of the 14th International Working Conference on Advanced Visual Interfaces 2016.AVI '16, Bari, Italy.ACM,2016.

    @inproceedings{Zadow2016_YouTouch,
       author = {Ulrich von Zadow and Patrick Reipschl\"{a}ger and Daniel B\"{o}sel and Anita Sellent and Raimund Dachselt},
       title = {YouTouch! \&\#8211; Low-Cost User Identification at an Interactive Display Wall},
       booktitle = {Proceedings of the 14th International Working Conference on Advanced Visual Interfaces 2016},
       year = {2016},
       month = {6},
       location = {Bari, Italy},
       numpages = {8},
       publisher = {ACM},
       address = {New York, NY, USA},
       keywords = {Display wall, multitouch, RGBD sensor, user identification, multi- user interaction, re-identification, interactive surface}
    }

Associated Student Theses

The following student theses from the Interactive Media Lab Dresden are associated with YouTouch!:

  • Daniel Bösel

    Image-based User Identification at an Interactive Display Wall using Artificial Neural Networks

    Daniel Bösel April 1st, 2016 until September 9th, 2016

    Supervision: Raimund Dachselt, Ulrich von Zadow

  • Daniel Bösel

    Robust Correlation of Interactions at a Display Wall with Users Using a Depth Camera

    Daniel Bösel August 15th, 2014 until November 6th, 2014

    Supervision: Raimund Dachselt, Ulrich von Zadow