Abstract | ||
---|---|---|
ABSTRACTNowadays, smart devices have become increasingly essential in humans' life. However, traditional input methods may not work effectively due to the tiny screen of the devices. Moreover, most learning-based gesture input schemes only perform well in terms of certain metrics such as accuracy, without considering other aspects like response time and user training overhead which are essential for real-world usage scenarios. Without taking the trade-off between different metrics into account, existing learning-based gesture input systems suffer from severe performance degradation in practice. In this paper, we investigate the trade-off between evaluation metrics of mobile interaction systems and then report our attempt towards a more practical digits input system based on our previous work. We propose an acoustic-based device-free digits input system named MetaDigit which achieves over 85% accuracy of real-time digits recognition with even zero-shot from new users. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1145/3417313.3429377 | SENSYS |
DocType | Citations | PageRank |
Conference | 0 | 0.34 |
References | Authors | |
0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Shicong Hong | 1 | 3 | 1.41 |
Zhihong Xiao | 2 | 0 | 0.34 |
Zishuo Guo | 3 | 0 | 0.34 |
Yongpan Zou | 4 | 118 | 9.06 |
kaishun wu | 5 | 1059 | 94.59 |