Fault Recognition Method Based on Attention Mechanism and the 3D-UNet

Table 2

The number of parameters and FLOPs of the compared segmentation model.

Model	Parameters	GFLOPs
3D-UNet-SE [34]	1634865	2.72e + 02 G
FCN4 [37]	1503337	21.7 G
FCN8 [37]	1615689	21.6 G
AttUNet [38]	1756008	3.76e + 02 G