This paper presents an efficient technique for real-time recognition of human activities using accelerometer signals alone. It is based on amplitude modulation frequency modulation (AM-FM) decomposition for feature extraction and support vector machine (SVM) algorithm for classification. Due to the nature of signals, and being the proposed technique independent from the orientation of the inertial sensor, this methodology is particularly suitable for implementation in smartwatches or other wearable sensors in order to recognize the exercise being performed. In order to demonstrate the validity of this methodology, it has been successfully applied to accelerometer data related to four dynamic activities and belonging to a free available database and compared with results in the literature.
Wearable Acceleration-Based Human Activity Recognition Using AM-FM Signal Decomposition / Biagetti, Giorgio; Crippa, Paolo; Falaschetti, Laura; Alessandrini, Michele; Turchetti, Claudio. - 309:(2022), pp. 429-439. [10.1007/978-981-19-3444-5_37]