Recognizing Unintentional Touch on Interactive Tabletop
summary
Uses gaze and head orientation plus touch to classify accidental contacts on a capacitive tabletop, extracting five spatio-temporal features (distance from gaze, side distance, touch clustering, finger count, temporal history). Trains a Gradient boosting model and ships a real-time filter; it achieves F1 = 91.3%, about 4.3% better than prior work. A 12-participant usability test shows the filter runs live and measurably reduces unintentional activations, letting users rest arms and lean naturally.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 0:12Background
- 1:30Intuition
- 1:45Research Questions
- 3:04Findings
- 4:11Viability
- 4:53Summary
- 5:38Outro
related talks
14mSmart Touch: Improving Touch Accuracy for People with Motor Impairments with Template Matching
24mGaze and Touch Interaction on Tablets
19mEveryday Eye Contact Detection Using Unsupervised Gaze Target Discovery
08mTypeBoard: Identifying Unintentional Touch on Pressure-Sensitive Touchscreen Keyboards
20mImproving Gesture Recognition Accuracy on Touch Screens for Users with Low Vision
14mTransform intent into action through your eyes
18mOrbits: Gaze Interaction for Smart Watches using Smooth Pursuit Eye Movements
10mTeethTap: Recognizing Discrete Teeth Gestures using Motion and Acoustic Sensing on an Earpiece