Table 5.
Comparison between the reported studies and proposed methods.
| Author | The Sensor Type and Number | Sensor Position | Number of Subjects | Classification Method | Number of Postures | Refresh Rate | Accuracy |
|---|---|---|---|---|---|---|---|
| Cheng et al. [25] | textile pressure-sensing matrix 80 × 80 | On floor | 11 | KNN | 7 | 40 Hz | 78.7% |
| Costilla-Reyes et al. [44] | piezoelectric sensors 88 × 88 | On floor | 127 | CNN + SVM | 3 | 1.6 kHz | 90.60% |
| Zhou et al. [35] | fabric sensor mat □ 120 × 54 | On floor | 13 | RNN (Recurrent Neural Network) | person identification | 25 Hz | 76.9% |
| Zhang et al. [45] | Force Sensing Resistors: 504 × 384 | On floor | 2 | Mean-Shift Clustering | 7 | 44 Hz | 95.59% |
| Proposed method | Pressure Thin Film Sensor □ 32 × 32 | On floor | 10 | Improved CNN | 9 | 40 Hz | 96.41% |
| D–S fusion | CNN–SVM–KNN | 10 | D-S theory | 9 | 40 Hz | 99.96% |