Research Article
You Can Write Numbers Accurately on Your Hand with Smart Acoustic Sensing
@INPROCEEDINGS{10.1007/978-3-319-78078-8_21, author={Mingshi Chen and Panlong Yang and Ping Li}, title={You Can Write Numbers Accurately on Your Hand with Smart Acoustic Sensing}, proceedings={Quality, Reliability, Security and Robustness in Heterogeneous Systems. 13th International Conference, QShine 2017, Dalian, China, December 16 -17, 2017, Proceedings}, proceedings_a={QSHINE}, year={2018}, month={4}, keywords={Smartwatch Wavelet transformation CNN}, doi={10.1007/978-3-319-78078-8_21} }- Mingshi Chen
Panlong Yang
Ping Li
Year: 2018
You Can Write Numbers Accurately on Your Hand with Smart Acoustic Sensing
QSHINE
Springer
DOI: 10.1007/978-3-319-78078-8_21
Abstract
Although smartwatch has drawn many attentions in recent years, small and inconvenient interaction mode limits the prevalence of smartwatches. Writing numbers with hands will naturally extend the input interface for smart watch. In this work, we design a passive acoustic sensing, where smart watches are collecting the ambient sound during writing. First of all, we use the wavelet transformation to mitigate the surrounding noise, and devise the time-frequency figures for AI enabled processing. After that, we apply the CNN(Convolutional Neural Network) model for number recognition, where three layers of convolution and three layers of max pool are incorporated. The number recognition accuracy rate could be above 95% when single person is well trained, and be around 92% when 7 to 9 persons are incorporated.
