Improving Virtual Keyboards When All Finger Positions Are Known
summary
Claims that assigning each key column to a finger and using correlations among all ten finger positions yields faster, more accurate typing; implementation uses a laser-light-plane touch surface plus fingertip color markers and a multinomial logistic regression classifier trained on ~40,000 labeled touches from 10 skilled typists. A keyboard with pre-allocation of keys plus multinomial logistic regression (PC) boosts speed from about 23.27 to 31.62 wpm and cuts total error from 10.24% to 7.59% in a 9-participant follow-up.
AI-written summary · about
chapters (9)
- 0:00Introduction to research
- 1:10Key observations for improvement
- 3:24Experiment setup details
- 6:58Correlations between fingers
- 8:24Implementing new keyboards
- 9:31Second experiment design
- 11:18Performance results
- 13:48Conclusion and summary
- 14:57Q&A session
related talks
21mDualKey: Miniature Screen Text Entry via Finger Identification
08mTypeBoard: Identifying Unintentional Touch on Pressure-Sensitive Touchscreen Keyboards
20mTipText: Eyes-Free Text Entry on a Fingertip Keyboard
21mThe Impact of Word, Multiple Word, and Sentence Input on Virtual Keyboard Decoding Performance
18mQWERTY or DVORAK? Debunking the Keyboard Layout Myths
23mIJQwerty: What Difference Does One Key Change Make? Gesture Typing Keyboard Optimization Bounded ...
20mDriftBoard: A Panning-Based Text Entry Technique for Ultra-Small Touchscreens
20mExpressive Keyboards: Enriching Gesture-Typing on Mobile Devices