Table 9.
Our real-time WGAS gait classification system vs. the literature.
| Reference | Algorithms (*) | Data Acquisition | Overall Accuracy |
|---|---|---|---|
| [22] | SVM | Three IMUs (Opal, APDM, Inc., Portland, OR, USA) featuring a tri-axial accelerometers and a tri-axial gyroscope | 90.5% |
| [23] | ANN, SVM | 37 markers were placed on body traced by an infrared camera during walking on the two embedded force plates | 96.9%, 98.2% |
| [24] | ANN, SVM | 2 markers on the shoe and traced by cameras during walking on a treadmill using the PEAK MOTUS motion analysis system | 75%, 83.3% |
| [25] | ANN, SVM | Synchronized PEAK 3D motion analysis system and a force platform during normal walking | 83.3%, 91.7% |
| [26] | ANN, SVM | Gait video sequence captured by a static camera during normal walking | 98%, 98% |
| [27] | SVM with PCA | Integrated a pressure sensor, a tilt angle sensor, three single-axis gyroscopes, one tri-axial accelerometer and a bend sensor inside a small module in a shoe with an RF transmitter | 98% |
| [28] | ANN | 29 retro reflective markers placed on the body, with 3D marker trajectories captured with an 8-camera motion analysis system during normal walking | 89% |
| [29] | ANN | 25 reflective markers placed on the body and data acquired from the Vicon Nexus 3D motion capture system | 95% |
| [30] | SVM | One tri-axial ACC worn on the subject waist while walking | 98% |
| This work | ANN, SVM | tri-axial gyroscopes and tri-axial accelerometers integrated on a PCB forming a WGAS with an MSP430 microcontroller and an RF transmitter embedded on the PCB | 100%, 98% |
* SVM: support vector machine; ANN: artificial neural network; PCA: principle component analysis.