TABLE 1.

Three convolution layers’ parameters of three branches.

Conv layer	SRF	MRF	MRF
Conv1	Size: 3 × 3 × 5	Size: 3 × 3 × 5	Size: 3 × 3 × 5
	Strides: 2, 2, 4	Strides: 2, 2, 4	Strides: 2, 2, 4
	Filters: 16	Filters: 16	Filters: 16
Conv2	Size: 2 × 2 × 1	Size: 2 × 2 × 3	Size: 2 × 2 × 5
	Strides: 2, 2, 1	Strides: 2, 2, 2	Strides: 2, 2, 4
	Filters: 32	Filters: 32	Filters: 32
Conv3	Size: 2 × 2 × 1	Size: 2 × 2 × 3	Size: 2 × 2 × 5
	Strides: 2, 2, 1	Strides: 2, 2, 2	Strides: 2, 2, 4
	Filters: 64	Filters: 64	Filters: 64

“Size” means the height, width, and depth of the 3D convolution window. “Strides” means the strides of the convolution window along each dimension. “Filters” means the number of output filters in each convolution layer.