*5.1. Coarse-Grained WD*

Coarse-grained WD is defined to distinguish walk and walk-like activities from walk-related and walk-related activities, which is a coarse classification of walk.

In Table 5, we could find that machine learning methods could easily distinguish walk and walk-like activities from walk-related and walk-unrelated activities. STFT performs better than DWT since walk features are better discriminated in the frequency domain than the time domain.

**Table 5.** Accuracy of coarse-grained walk detection.

