Research Article
FSR-Based Smart System for Detection of Wheelchair Sitting Postures Using Machine Learning Algorithms and Techniques
Table 2
Previous work comparison.
| Sr # | Reference # | Type of sensors | No. of postures | Classifiers/software | Accuracy | Limitations |
| 1 | [23] | Pressure sensor | 13 | (DT-CART), (RF), (KNN), (LR), (LDA), and (NB) | 98.93% | Subjects for the dataset are minimal |
| 2 | [13] | Wisat mat | 5 | Wisat algorithms in MATLAB | 81% | Only 17 datasets are available |
| 3 | [8] | 3 pressure sensors, 1 sonar | 4 | KNN | 76.05% | Less accuracy and classifier comparison is significantly less |
| 4 | [14] | 12 pressure sensors | 5 | Decision tree (J48), (SVM), (MLP), Naive Bayes, and (-NN) | 99.47% | Can record up to only 12 pressure points simultaneously |
| 5 | [15, 26] | A pressure sensor | 4 | Support vector machine (SVM) | 89.6% | Multiple data collected by only using 10 subjects |
| 6 | [10] | 16 sensors with 16 matrix | 4 | -nearest neighbors (-NN), random forest (RF), decision tree (DT) support vector machines (SVM), and LightGBM | 99.03% | Not proposed a system through which it can correct wrong user posture |
|
|