FingMag: Finger Identification Method for Smartwatch

Keunwoo Park, Geehyuk Lee

CHI 2019 LBW

Abstract

Interacting with a smartwatch is difficult owing to its small touchscreen. A general strategy to overcome the limitations of the small screen is to increase the input vocabulary. A popular approach to do this is to distinguish fingers and assign different functions to them. As a finger identification method for a smartwatch, we propose FingMag, a machine-learning-based method that identifies the finger on the screen with the help of a ring. For this identification, the finger's touch position and the magnetic field from a magnet embedded in the ring are used. In an offline evaluation using data collected from 12 participants, we show that FingMag can identify the finger with an accuracy of 96.21% in stationary geomagnetic conditions.

Images

FingMag identifying a touching finger.

demo

Documents

You can read our extended abstract from ACM DL. We also share the poster we presented at CHI 2019.