Classification of the Korean Sign Language Alphabet Using an Accelerometer with a Support Vector Machine

Table 3

Comparison studies on computational cost.

Author, year	Recognition targets	Raw signal dimension (sensor types)	Number of features	Training time (s)	Recognition time (ms/sample)	Classification accuracy (%) (random chance)
Paudyal et al., 2017 [11]	Letters	17 (8D sEMG, 9D IMU)	327	N/A	67.0 (PC), 600.0–2000.0 (smartphone)	95.36 (3.85)
Mummadi et al., 2018 [9]	Letters	45 (9D IMU 5)	10	105	140.0	92.95 (4.55)
Suri and Gupta, 2019 [8]	Sentences	6 (3D ACC, 3D gyro)	N/A	24000	0.5	94.00 (5.00)
This study	Letters	3 (3D ACC)	7	1.6	1.8	92.24 (3.23)