| Title |
Gesture Recognition of Writing Numbers using Pressure Sensor based on CNN-LSTM Combination Model |
| Authors |
김서현(Seo-Hyeon Kim) ; 신하림(Ha-Rim Shin) ; 한영선(Youngsun Han) ; 장원두(Won-Du Chang) ; 민영재(Yong-Jae Min) |
| DOI |
https://doi.org/10.5573/ieie.2023.60.12.52 |
| Keywords |
Deep learning; Numeric gesture classification; CNN; LSTM; Pressure sensor |
| Abstract |
This paper presents a method for acquiring and automatically recognizing user gestures through a pressure sensor. The pressure sensor used in this study was attached to the back of a cell phone, which allows a certain range of movements to be observed using only a single sensor. Since gesture recognition using a pressure sensor can be configured at a low cost compared to gesture recognition using a touch panel or other sensors, it is expected to increase the freedom of hardware device configuration if it can recognize complex patterns. To validate the model, 5400 numeric gesture data were collected, and a model combining a convolutional neural network and a long short-term memory layer was proposed for automatic classification of the data. The experimental results showed an accuracy of 79.9% for 10 digit patterns and 89.0% for 7 patterns, showing the potential of gesture recognition using pressure sensors. |