Multimode Gesture Recognition Algorithm Based on Convolutional Long Short-Term Memory Network

Table 1

Experimental parameter settings.

Parameter	Options
Initialize the weight coefficient of CNN layer	Kaiming method
LSTM layer weight coefficient initialization	Orthogonal method
The weight coefficient of the full connection layer is initialized	Kaiming method
Optimizer	Adam optimizer
Loss function	Cross entropy
Initial learning rate	0.001
Sample sequence size	24 × 410
Number of training set samples	20088
Number of samples in test set	2232
Number of training wheels	20
Batch size	500
Leaky ReLU divisor	0.1