センサ行動認識のための機械学習を用いた加速度データシミュレーション

Shingo Takeda, Paula Lago, Tsuyoshi Okita, Sozo Inoue,
SOFT九州支部学術講演会
(Not Available)
(Not Available)
49-50
2018-12-01
Kagoshima
In this paper, we simulate accelerometer from motion capture data for sensor-based activity recognition. We propose to use machine learning to add noises caused by shaking of sensors. Experiments with public dataset showed that F-Score estimated by our method is 2% closer to that obtained with the real accelerometer data than that estimated by the conventional method that not considered shaking of sensors.

Data Files