Improving Sensor-based Activity Recognition Using Motion Capture as Additional Information

Reviewed
Paula Lago, Shingo Takeda, Tsuyoshi Okita, Sozo Inoue,
ACM Int'l Conf. Pervasive and Ubiquitous Computing (Ubicomp) Poster
(Not Available)
(Not Available)
4 pages
2018-10-09
Singapore
http://ubicomp.org/ubicomp2018/
We propose a new method for human activity recognition
using a single accelerometer sensor and additional sensors
for training. The performance of inertial sensors for complex

activities drops considerably compared with simple activ-
ities due to inter-class similarities. In such cases deploy-
ing more sensors may improve the performance. But such

strategy is often not feasible in reality due to costs or pri-
vacy concerns among others. In this context, we propose

a new method to use additional sensors only in training
phase. We introduce the idea of mapping the test data to
a codebook created from the additional sensor information.
Using the Berkeley MHAD dataset our preliminary results
show this worked positively; improving by 10.0% both the
average F1-score and the average accuracy. Notably, we
achieved an average improvement of 143.1% for the stand,
sit and sit to stand activities, typical activities for which the
inertial sensor is less informative when using the wrist-worn
accelerometer.

Data Files

No records to display