To make motion gestures more widely adopted on mobile devices it is important that devices be able to distinguish between motion intended for mobile interaction and every- day motion. In this paper, we present DoubleFlip, a unique motion gesture designed as an input delimiter for mobile motion-based interaction. The DoubleFlip gesture is dis- tinct from regular motion of a mobile device. Based on a collection of 2,100 hours of motion data captured from 99 users, we found that our DoubleFlip recognizer is extreme- ly resistant to false positive conditions, while still achieving a high recognition rate. Since DoubleFlip is easy to perform and unlikely to be accidentally invoked, it pro- vides an always-active input event for mobile interaction.
