Figure 8.

An example of the signal performed by the health subject and the processes of the proposed sub-task segmentation. (a) The accelerometer and gyroscope signals collected from the IMUs placed on the wrist and arm. (b) The divided segments obtained from the process of sliding window. In this example, there are 23 segments (c) The classification results for sub-task A, B and C after the processes of feature extraction and ML-based sub-task segmentation, where the TP, TN, FP, and FN are annotated. (d) The classification results after the processes of rule-based modification, where the modified sliding segments are highlighted in red square (e.g., $w_4$, $w_{10}$, $w_{15}$, $w_{22}$) and the successful modification results are annotated.