An OCaNet Model Based on Octave Convolution and Attention Mechanism for Iris Recognition

Table 1

Related parameters of different layers.

Network	Layer name	Layer type	Related parameters
—	Conv1	Conv2D	7 × 7, 64, stride 2
—	Pool1	MaxPool2d	3 × 3, 64, stride 2
OCT module	OCT block 1	Convolution
	OCT block 2	Convolution
	OCT block 3	Convolution
	OCT block 4	Convolution
	Pool3	AdaptiveAvgPool2d	1, 1
Attention module	EA_1	Attention	× 4
	Attention_2	Attention	× 4
	Attention_3	Attention	× 4
	Attention_4	Attention	× 4
	Pool2	AdaptiveAvgPool2d	1, 1
—	FC	Inner product	—