Triaxial accelerometer-based human activity recognition using 1D convolution neural network
Wahyu Andhyka Kusuma, Agus Eko Minarno, Mochamad Satriyo Wibowo · 2020
Deep learning has been instrumental for human activity recognition (HAR). In spite of its strong potential, significant challenges exist, wherein the real case, deep learning model requires a massive dataset for training. However, existing research require an improvement to classify static and dynamic activity with more significant achievement. To address such challenges, we proposed a model utilizing 1-dimensional Convolution Neural Network (CNN) to classify static and dynamic activity using public dataset. The proposed scheme in this study has been conducted (through experiments), in which the result denotes the state-of-the-art methods, obtaining better performance than others.