Smart Touch: Improving Touch Accuracy for People with Motor Impairments with Template Matching
summary
Empirical study of 10 participants on a Microsoft PixelSense tabletop records multi-point contacts (some users produce 5–8 concurrent touches and land/lift errors measured in centimeters, up to ~45 cm). A template‑matching system called Smart Touch extracts stable frames, normalizes ellipse poses, and uses 30 templates for matching to predict intended (x,y) by applying learned offsets to the weighted centroid. In evaluation Smart Touch yields predictions 3x closer than native land-on/lift-off, though residual errors remain in centimeters.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:15The Problem
- 2:25The Solution
- 4:07Concurrent Touches
- 5:08Land or Distance
- 6:14Liftoff Distance
- 7:13Smart Touch
- 11:08Evaluation
related talks
06mRecognizing Unintentional Touch on Interactive Tabletop
20mIt's a Wrap: Mapping On-Skin Input to Off-Skin Displays
06mScraTouch: Extending Interaction Technique Using Fingernail on Unmodified Capacitive Touch Surfaces
24mGaze and Touch Interaction on Tablets
20mImproving Gesture Recognition Accuracy on Touch Screens for Users with Low Vision
19mATK: Enabling Ten-Finger Freehand Typing in Air Based on 3D Hand Tracking Data
19mDesigning Multi-Touch Gestures Using the Space around the Smartwatch as Continuous Input Space
23mRealPen: Providing Realism in Handwriting Tasks on Touch Surfaces using Auditory-Tactile Feedback