Table 1 A tabular summary of different HAR methodologies and their corresponding performances (in terms of accuracy) achieved till date for HARTH, KU-HAR and HuGaDB datasets.
Dataset | Study | Method | Accuracy(%) |
|---|---|---|---|
HARTH | Logacjov et al.14 | SVM-based classification model | 81 |
KU-HAR | Sikder and Nahid15 | RF-based classification model | 89.67 |
HuGaDB | Filtjens et al.41 | Multi-stage spatial-temporal graph convolutional network (MS-GCN) | 83.8 |
Gochoo et al.36 | Hierarchical feature-based technique with SGD | 92.5 | |
Javeed et al.37 | A Hybrid features selection model using deep belief networks | 92.5 | |
Fang et al.38 | CNN model | 79.24 | |
Sun et al.39 | ANN based method for real-time gait analysis | 88 | |
Kumari et al.40 | LSTM model | 91.1 |
