Research Article
Gait Recognition Using Spatio-Temporal Information of 3D Point Cloud via Millimeter Wave Radar
Table 2
A comparison of various radar-based personal identification studies with our method.
| | Method | [17] | [18] | [19] | [20] | [14] | [15] | [13] | Our method |
| | Single-fixed | NA | 91% | NA | NA | 90% | NA | NA | 96.7% | | Single-random | 93% | 90% | 85.8% | 68.88% | 45% | 78.46% | 89% | 94.9% | | Two-fixed | NA | NA | NA | NA | 86% | NA | NA | 90.2% | | Two-random | NA | NA | NA | 75.51% | NA | NA | NA | 87.4% | | Number of people identified | 8 | 10 | 4 | 7 | 10 | 5 | 12 | 13 | | Number of coexisting persons | 1 | 1 | 1 | 2 | 5 | 1 | 1 | 2 | | Type of input data | Point cloud | Gait spectrum | Point cloud | Point cloud | Point cloud | Micro-Doppler | Point cloud | Point cloud | | Number of training data required | 40,000 frames | NA | 100,000 frames | 75 minutes | 900 MB | 100 minutes | NA | 127 MB | | Number of experimental scenes | 3 | 1 | 1 | 3 | 2 | 2 | 1 | 3 |
|
|
