Presentation 2021-03-01
Fall Detection Based on LSTM Using Accelerometer
Yoshiya Uotani, Chen Ye, Kohei Yamamoto, Tomoaki Ohtsuki,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In recent years, the increase in fall accidents among the elderly has become a problem with the aging of the population. A fall detection method using an accelerometer has been researched and developed. However, conventional fall detection methods do not support the positions of multiple accelerometers. Besides, there is a problem that the detection accuracy is low because they use a classification algorithm that is not suitable for time series prediction, such as Support Vector Machine (SVM) and Random Forest (RF). In this report, we propose fall detection based on LSTM (Long short-term memory) using an accelerometer. In the proposed method, behaviors are learned at multiple sensor installation positions, and the behaviors are classified into multiple classes by LSTM, and then classified into two classes, fall and non-fall. Experiments show that the proposed method achieves higher fall detection accuracy than the conventional method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Long short-term memory (LSTM) / 3-axis accelerometer / fall detection
Paper # SeMI2020-59
Date of Issue 2021-02-22 (SeMI)

Conference Information
Committee SeMI / IPSJ-MBL / IPSJ-UBI
Conference Date 2021/3/1(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Online
Topics (in Japanese) (See Japanese page)
Topics (in English) Mobile Computing, Ubiquitous Computing, etc.
Chair Susumu Ishihara(Shizuoka Univ.)
Vice Chair Kazuya Monden(Hitachi) / Koji Yamamoto(Kyoto Univ.)
Secretary Kazuya Monden(Kyoto Univ.) / Koji Yamamoto(Cyber Univ.) / (Hitachi) / (Waseda Univ.)
Assistant Yuki Katsumata(NTT DOCOMO) / Yu Nakayama(Tokyo Univ. of Agri. and Tech.) / Akira Uchiyama(Osaka Univ.)

Paper Information
Registration To Technical Committee on Sensor Network and Mobile Intelligence / Special Interest Group on Mobile Computing and Pervasive Systems / Special Interest Group on Ubiquitous Computing System
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Fall Detection Based on LSTM Using Accelerometer
Sub Title (in English)
Keyword(1) Long short-term memory (LSTM)
Keyword(2) 3-axis accelerometer
Keyword(3) fall detection
1st Author's Name Yoshiya Uotani
1st Author's Affiliation Keio University(Keio)
2nd Author's Name Chen Ye
2nd Author's Affiliation Keio University(Keio)
3rd Author's Name Kohei Yamamoto
3rd Author's Affiliation Keio University(Keio)
4th Author's Name Tomoaki Ohtsuki
4th Author's Affiliation Keio University(Keio)
Date 2021-03-01
Paper # SeMI2020-59
Volume (vol) vol.120
Number (no) SeMI-382
Page pp.pp.7-12(SeMI),
#Pages 6
Date of Issue 2021-02-22 (SeMI)