Accelerometer Data Embedded in a Wristwatch
Dataset Descriptions
(see our publication at AIMS 2002)
Dataset1: Glancing at the watch.
It can be useful for the watch to know when it is looked at, for instance for triggering a power-saving mode if the watch is not being glanced at. This dataset contains typical values of accelerometer data, captured using an array of the off-the-shelf Crossbow ADXL evaluation boards, of the movements someone makes when glancing at his/her watch.
The dataset is in Excel format, with the different scenarios spread over different worksheets (plus one worksheet containing graphs).
Dataset2: Gestures and activities.
Instead of concentrating on tiny buttons or touch-screens, watches can also get input from gestures. This dataset contains accelerometer data of several gestures and actions that can be used to alter the behaviour of the watch (e.g. enlarge the watch's font when the user is running). The same setup was used as in the first dataset.
The dataset is in Excel format, with the different scenarios spread over different worksheets (plus one worksheet containing graphs).
Dataset3: Watching the watch testing sets.
Similar to dataset 1, but now multiple similar datasets for training-testing evaluation of desk-watch-desk.
The dataset is in Excel format, with the different scenarios spread over different worksheets (plus one worksheet containing graphs).
Dataset4: Watching the watch testing sets.
Similar to dataset 1, but now multiple similar datasets for training-testing evaluation of standing-watch-standing.
The dataset is in Excel format, with the different scenarios spread over different worksheets (plus one worksheet containing graphs).
Source code for the PIC
SC