Table 2.

Smart cushions based on fewer individual pressure sensors.

Author	Number of Sensors	Placement of the Sensors	Postures Recognized	Classification Techniques	Accuracy
Hu et al. [35]	6	2 on the seat and 4 on the backrest	Sit straight, lean left, lean right, lean back	SVM	N/A
Benocci et al. [36]	5	4 on the seat and 1 on the backrest	Normal posture, right side, left side, right/left/both legs extend forward	kNN	92.7%
Bao et al. [37]	5	5 on the seat	Normal sitting, forward, backward, lean left, lean right, swing, shake	Density-based cluster	94.2%
Diego et al. [38]	4	4 on the seat	N/A	Threshold-based	N/A
Min et al. [39]	6	4 on the seat and 2 on the backrest	Crossing left leg, crossing right leg, forward buttocks, bending down the upper body, correct posture	Decision Tree	N/A
Zemp et al. [40]	16	10 on the seat, 2 on the armrests and 4 on the backrest	Upright, reclined, forward inclined, laterally right/left, crossed legs, left over right/ right over left	SVM, Multinomial Regression, Boosting, Neural Networks and Random Forest	90.9%
Barba et al. [41]	16	8 on the seat and 8 on the backrest	Standard, lying, forward, normal position, sitting on the edge, legs crossed, sitting on one/two foot etc.	N/A	N/A
Fu et al. [42]	8	4 on the seat and 4 on the backrest	N/A	Decision Tree	N/A
Kumar et al. [43]	4	4 on the backrest	N/A	Extremely Randomized Trees	86%
Ma et al. [44]	3	2 on the seat and 1 on the backrest	Upright sitting, lean left, right, forward, backward	Decision Tree	99.5%
Darma [45]	6	6 on the seat	N/A	N/A	N/A
Sensimat [46]	6	6 on the seat	N/A	N/A	N/A