Table 5.

Performance comparison to the related literature.

Study	Avg Error ± SD	Counting Method	No. of Subjects
Farooq and Sazonov [3]		10.40% ± 7.03%	Peak detection in manually annotated segments	30
15.01%± 11.06%	Counting in ANN classified epochs	30
Farooq and Sazonov [22]		8.09% ± 7.16%	Piezoelectric strain sensor	5
8.26% ± 7.51%	Piezoelectric strain sensor	5
Farooq and Sazonov [42]	9.66% ± 6.28%	Linear regression of piezoelectric sensor signal	10
Bedri et al. [24]	F1-score = 90.9%	Acoustic sensor	10
Cadavid et al. [27]	Avg agreement = 93%	SVM classification of AMM spectral features	37
Taniguchi et al. [43]	Precision = 0.958	Earphone sensor	6
Wang et al. [44]	12.2%	Triaxial accelerometer on the temporalis	4
Hossain et al. [26]	Mean accuracy 88.9% ±7.4%	Deep learning and affine optical flow	28
This paper	5.42% ± 4.61 (slow) 7.47% ± 6.85 (normal) 9.84% ± 9.55 (fast)	Image processing of chewing videos	100