Research Article
Position-Based Feature Selection for Body Sensors regarding Daily Living Activity Recognition
Table 6
Recognition performance for thirteen activities in terms of F1-score metric.
| Activity | Leutheuser et al. | Zdravevski et al. | Proposed method |
| Sitting | 94.10% | 93.07% | 96.83% | Lying | 100% | 100% | 99.74% | Standing | 90.93% | 74.40% | 97.35% | Washing dishes | 96.50% | 79.21% | 99.83% | Vacuuming | 85.13% | 100% | 96.88% | Sweeping | 88.57% | 100% | 98.74% | Walking | 98.89% | 94.48% | 99.81% | Ascending stairs | 95.55% | 98.50% | 99.73% | Descending stairs | 94.80% | 91.50% | 98.65% | Treadmill running (8.3 km/h) | 100% | 94.80% | 100% | Bicycling on an ergometer (50 W) | 63.50% | 99.60% | 98.62% | Bicycling on an ergometer (100 W) | 58.75% | 99.55% | 99.01% | Rope jumping | 100% | 98.43% | 100% | Overall accuracy | 89.00% | 93.00% | 99.13% |
|
|