A Multimodal Information Fusion Model for Robot Action Recognition with Time Series

Table 1

Experimental parameter settings.

Parameters	Value
Number of clips D	30
Image frame size	280 ∗ 280
Convolutional layer	5
Pooling layer	3
Inception module	3
Hidden layer dimension	2048
Dropout	0.3
Initial learning rate	0.001
Batch size	32
Weight decay factor	0.00001