Human activity recognition method based on inertial sensor and barometer
Lili Xie, Junfang Tian, Genming Ding, Qian Zhao · 2018
In this paper, we propose a human activity recognition (HAR) method based on inertial sensors and barometer. The proposed method recognizes eight human activities following a multi-layer strategy. Activities are classified into two categories: dynamic and static activities; then explicit activity recognition is taken individually in the two categories. Three classifiers are adopted for different classification, including random forest (RF) and support vector machine (SVM). Different feature sets have been selected for different classifiers which are more targeted and effective. In addition, the classifier result is further verified by additional parameters and previous recognition results to decide the final recognition result. Experiments have shown the effectiveness and good performance of the proposed HAR method.