Table 3 Evaluation and comparison experiments of 3D semantic occupancy prediction.

MonoScene¹⁹

CVPR2022

C

384 × 1280

4.53

9.28

4.65

0.71

0.94

0.00

28.81

4.53

8.93

0.93

2.64

22.13

5.92

TPVFormer²⁰

CVPR2023

C

384 × 1280

4.31

10.11

2.06

0.15

0.12

0.00

28.51

3.37

8.18

0.82

2.77

22.11

5.49

Occformer²¹

ICCV2023

C

384 × 1280

9.9

17.14

6.19

2.82

0.38

0.00

29.2

6.83

9.51

5.77

5.41

22.45

8.17

CG-Former³⁹

NeurIPS2024

C

384 × 1280

14.62

16.89

8.34

1.5

1.34

0.00

57.00

10.57

31.65

2.42

11.4

49.47

14.14

CO-Occ(L)²²

RAL2024

L

384 × 1280

26.28

39.81

16.74

8.78

10.01

0.00

34.43

9.74

29.82

1.94

19.58

34.19

17.92

CO-Occ(M)²²

RAL2024

C+L

384 × 1280

26.15

36.4

16.7

15.78

7.36

2.11

34.96

8.52

28.59

5.35

18.14

34.1

18.19

L2COcc²³

arXiv2025

C

384 × 1280

19.45

18.66

10.67

3.03

4.07

0.00

57.00

10.28

31.97

3.49

13.19

49.66

15.65