Table 3 Experimental results comparison of six methods on the self-constructed dataset.

From: Feature-guided multilayer encoding–decoding network for segmentation for 3D intraoral scan data

Method

OA (%)

mIoU (%)

Params (M)

FLOPs (G)

Test(s/Dental)

PointNet13

93.2

82.6

3.53

11.57

0.32

MeshSegNet12

85.3

64.8

1.79

117.59

4.31

PointMLP14

94.3

84.5

16.74

30.61

0.46

MBeSegNet15

76.1

49.6

0.92

30.88

0.44

PointeNet16

94.4

85.2

14.76

29.64

0.43

Our method

96.6

90.8

1.54

6.53

0.30

Back to article page