A Noble Classification Framework for Data Glove Classification of a Large Number of Hand Movements

Table 4

Comparison of the number of movements and classification accuracy of the proposed method with existing algorithms.

Author	Classification algorithm	Sensing material	Number of movements	Accuracy of classification (%)
Nassour et al. [1]	A linear regression	Electrolyte solution resistance	15	89.4
Chen et al. [2]	Support vector machine	Flexion sensors and force sensors	16	89.4
Pan et al. [3]	Extremely randomized trees	Capacitive pressure sensors	10	99.7
Maitre et al. [4]	Random forest	The force and bending sensors	8	95
Lee and Bae [5]	LSTM	Eutectic gallium indium sensors	11	100
Ayodele et al. [6]	CNN	Weft knit sensors	6	88.27
Chauhan et al. [7]	Gaussian process	Rotary potentiometers	5	75
Fatimah et al. [8]	Ensemble subspace discriminant	Surface electromyogram	17	93.53
Maitre et al. [9]	CNN-LSTM	The force and bending sensors	13	93
Huang et al. [10]	Dynamic time warping	RGO-coated fiber sensors	9	98.3
Lun et al. [11]	GCNs-Net	Electroencephalogram	4	96.24
Hou et al. [12]	BiLSTM-GCN	Electroencephalogram	4	98.81
Proposed work	DBDF	Flexion sensors (CyberGlove II)	52	93.15