This table lists the benchmark results for the low-res two-view scenario. This benchmark evaluates the Middlebury stereo metrics (for all metrics, smaller is better):

The mask determines whether the metric is evaluated for all pixels with ground truth, or only for pixels which are visible in both images (non-occluded).
The coverage selector allows to limit the table to results for all pixels (dense), or a given minimum fraction of pixels.

Methods with suffix _ROB may participate in the Robust Vision Challenge.

Click one or more dataset result cells or column headers to show visualizations. Most visualizations are only available for training datasets. The visualizations may not work with mobile browsers.




Method Infoalllakes. 1llakes. 1ssand box 1lsand box 1sstora. room 1lstora. room 1sstora. room 2lstora. room 2sstora. room 2 1lstora. room 2 1sstora. room 2 2lstora. room 2 2sstora. room 3lstora. room 3stunnel 1ltunnel 1stunnel 2ltunnel 2stunnel 3ltunnel 3s
sorted bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort by
AdaStereotwo views1.45
1
0.83
11
2.34
3
1.28
20
0.81
8
3.21
4
1.14
20
4.20
40
2.46
26
2.05
12
1.77
10
2.48
15
1.18
3
1.71
1
1.44
1
0.43
24
0.25
11
0.43
24
0.28
10
0.44
13
0.34
8
Xiao Song, Guorun Yang, Xinge Zhu, Hui Zhou, Zhe Wang, Jianping Shi: AdaStereo: A Simple and Efficient Approach for Adaptive Stereo Matching. ArXiv
StereoDRNet-Refinedtwo views1.77
2
0.81
8
2.22
2
1.21
15
0.76
5
3.26
8
0.71
4
3.82
16
2.05
4
2.29
17
2.05
26
2.98
33
1.44
12
7.15
17
2.27
27
0.30
2
0.23
8
0.42
23
0.33
13
0.45
17
0.64
42
Rohan Chabra, Julian Straub, Chris Sweeney, Richard Newcombe, Henry Fuchs: StereoDRNet. CVPR
MLCVtwo views1.87
3
0.91
15
8.59
14
1.10
7
0.57
2
3.59
21
0.53
1
3.70
11
2.20
8
2.80
31
2.25
38
2.00
5
1.42
11
4.13
10
1.86
19
0.30
2
0.20
3
0.32
7
0.21
3
0.33
4
0.36
11
iResNet_ROBtwo views1.89
4
1.02
22
8.60
15
1.39
33
0.98
23
3.72
27
0.66
3
3.84
17
2.65
31
1.88
7
2.05
26
1.98
4
1.27
6
4.52
12
1.67
11
0.30
2
0.20
3
0.28
3
0.18
1
0.31
2
0.40
13
CFNet_RVCtwo views1.90
5
0.95
18
3.56
9
1.07
4
1.54
45
4.09
46
0.90
12
2.85
1
1.82
3
1.78
3
1.70
4
2.75
22
1.48
16
9.33
23
1.72
14
0.33
9
0.30
17
0.61
52
0.36
21
0.52
24
0.39
12
DN-CSS_ROBtwo views1.91
6
1.55
53
11.13
24
1.82
52
0.88
13
3.99
43
0.61
2
4.01
27
1.67
1
1.64
2
1.89
16
1.39
2
1.06
1
3.23
4
1.50
2
0.31
5
0.19
2
0.44
26
0.29
11
0.37
6
0.31
5
LALA_ROBtwo views1.95
7
1.48
50
3.11
7
1.05
3
0.98
23
3.99
43
1.82
39
4.44
51
2.28
14
3.31
42
1.85
14
2.76
25
1.67
25
4.04
9
3.23
38
0.46
29
0.41
33
0.55
45
0.52
45
0.60
38
0.57
32
DLCB_ROBtwo views2.05
8
0.95
18
2.37
4
1.24
18
1.02
27
2.92
2
1.30
25
3.68
9
2.30
16
2.91
34
2.17
34
3.32
36
1.52
18
10.37
29
2.50
29
0.40
20
0.33
20
0.40
19
0.36
21
0.41
10
0.47
19
DeepPruner_ROBtwo views2.05
8
1.09
30
10.50
21
1.01
2
1.20
35
4.39
54
0.91
14
3.14
3
1.72
2
2.60
25
1.54
1
2.78
26
1.27
6
4.52
12
1.72
14
0.51
37
0.40
31
0.37
11
0.34
14
0.48
19
0.52
26
HITNettwo views2.07
10
0.82
10
14.77
32
1.34
26
0.56
1
3.47
15
0.80
7
4.07
32
2.35
18
2.16
16
1.61
2
1.64
3
1.44
12
2.09
2
2.56
30
0.26
1
0.18
1
0.41
22
0.26
9
0.45
17
0.25
1
Vladimir Tankovich, Christian Häne, Sean Fanello, Yinda Zhang, Shahram Izadi, Sofien Bouaziz: HITNet: Hierarchical Iterative Tile Refinement Network for Real-time Stereo Matching.
ETE_ROBtwo views2.07
10
1.40
47
3.01
6
1.09
5
0.92
18
3.94
37
1.53
32
4.25
41
2.23
9
3.31
42
2.06
28
2.35
12
1.52
18
8.74
20
2.07
22
0.44
26
0.36
26
0.51
39
0.48
42
0.54
28
0.62
37
DISCOtwo views2.25
12
0.73
6
10.06
20
1.76
51
1.12
30
3.47
15
1.25
24
3.24
4
2.26
13
2.65
27
1.96
18
4.53
50
1.92
35
4.17
11
3.61
40
0.32
7
0.25
11
0.31
6
0.29
11
0.66
45
0.41
15
XPNet_ROBtwo views2.27
13
1.11
33
4.10
10
1.36
31
0.97
21
3.94
37
1.63
34
4.07
32
2.14
6
2.82
32
2.00
21
2.32
11
1.73
27
10.81
34
3.39
39
0.51
37
0.47
42
0.48
32
0.40
34
0.51
22
0.56
31
NOSS_ROBtwo views2.27
13
0.73
6
3.00
5
1.71
48
1.00
25
3.90
35
0.97
15
4.29
43
2.81
38
2.94
35
2.19
36
3.88
42
1.19
4
11.77
46
1.62
6
0.59
46
0.54
46
0.56
46
0.52
45
0.56
33
0.54
29
TDLMtwo views2.29
15
1.08
28
4.61
11
1.29
21
1.01
26
4.26
53
2.17
47
3.75
14
2.79
35
2.31
19
1.98
20
2.06
8
1.59
20
12.45
54
1.72
14
0.51
37
0.28
16
0.52
42
0.35
17
0.56
33
0.50
23
CFNettwo views2.34
16
1.10
31
8.96
16
1.34
26
1.24
37
4.09
46
0.73
6
4.04
31
2.77
34
1.62
1
1.87
15
2.27
10
1.45
15
11.41
38
1.82
18
0.31
5
0.24
10
0.43
24
0.36
21
0.43
12
0.33
6
HSMtwo views2.35
17
0.81
8
2.21
1
1.14
10
1.55
46
3.44
12
1.03
16
3.99
26
2.25
11
2.29
17
2.11
31
8.04
57
3.05
46
11.60
42
1.71
12
0.33
9
0.22
7
0.28
3
0.23
6
0.32
3
0.35
10
iResNettwo views2.52
18
0.94
17
21.17
59
1.84
54
0.72
4
3.75
28
0.81
9
4.14
36
2.55
30
2.33
20
2.03
22
1.35
1
1.44
12
3.27
5
2.16
25
0.35
16
0.23
8
0.28
3
0.22
5
0.44
13
0.45
17
NCCL2two views2.53
19
1.35
42
10.90
22
1.15
11
0.97
21
3.53
19
2.56
51
3.65
8
2.23
9
2.76
29
1.80
11
2.41
13
1.66
24
10.35
28
2.07
22
0.44
26
0.40
31
0.57
47
0.53
48
0.59
35
0.65
44
ccs_robtwo views2.54
20
1.14
35
14.21
29
1.35
29
0.96
19
3.98
42
0.71
4
3.86
19
2.88
39
2.04
11
2.07
29
2.05
7
1.29
9
10.60
31
1.71
12
0.32
7
0.25
11
0.38
12
0.35
17
0.39
7
0.33
6
NVstereo2Dtwo views2.54
20
0.90
13
11.10
23
1.32
22
1.47
44
3.58
20
1.38
27
4.13
34
2.45
24
1.90
9
1.71
7
2.86
30
1.76
28
10.86
35
1.66
10
0.60
47
0.43
37
0.53
43
0.41
35
0.78
51
0.94
54
NLCA_NET_v2_RVCtwo views2.54
20
1.04
24
16.73
39
1.53
43
2.00
59
3.96
39
0.87
11
3.87
21
2.31
17
2.54
23
1.67
3
2.85
29
1.63
22
5.55
15
1.62
6
0.46
29
0.33
20
0.38
12
0.38
28
0.53
27
0.62
37
Zhibo Rao, Mingyi He, Yuchao Dai, Zhidong Zhu, Bo Li, and Renjie He.: NLCA-Net: A non-local context attention network for stereo matching.
CVANet_RVCtwo views2.55
23
1.10
31
9.13
17
1.46
36
1.15
32
3.96
39
1.53
32
3.91
24
2.50
27
3.04
37
2.04
24
2.60
20
1.72
26
12.46
55
1.50
2
0.49
33
0.30
17
0.51
39
0.41
35
0.63
40
0.50
23
CC-Net-ROBtwo views2.55
23
1.05
25
16.77
40
1.50
39
1.94
58
3.97
41
0.90
12
3.85
18
2.25
11
2.55
24
1.70
4
2.82
28
1.65
23
5.67
16
1.65
8
0.47
31
0.33
20
0.38
12
0.39
30
0.54
28
0.64
42
DANettwo views2.63
25
1.07
27
8.10
12
1.83
53
1.70
52
3.25
7
1.16
21
4.31
46
2.38
21
3.40
45
2.18
35
4.08
45
1.80
30
9.77
25
3.89
42
0.52
42
0.55
48
0.59
49
0.44
39
0.63
40
0.86
51
iResNetv2_ROBtwo views2.65
26
5.00
67
19.28
51
2.06
59
1.09
29
3.83
33
1.13
19
4.14
36
2.09
5
1.85
5
2.03
22
2.01
6
1.51
17
3.58
6
1.58
4
0.34
12
0.21
5
0.32
7
0.21
3
0.44
13
0.29
3
PASMtwo views2.67
27
2.05
61
9.71
19
1.73
50
1.37
42
2.53
1
1.52
31
3.10
2
2.17
7
2.79
30
1.84
12
2.90
31
1.85
33
11.89
49
2.19
26
0.81
54
0.85
59
0.87
60
1.01
62
1.14
58
1.05
56
PWC_ROBbinarytwo views2.69
28
2.01
60
8.36
13
1.33
24
1.90
57
3.23
5
1.18
22
4.33
48
3.25
47
3.11
38
3.41
47
2.71
21
2.18
39
9.15
22
4.48
44
0.56
44
0.31
19
0.61
52
0.36
21
0.69
47
0.74
47
StereoDRNettwo views2.84
29
1.49
51
14.48
31
1.33
24
1.57
49
3.44
12
2.06
43
3.69
10
2.36
20
3.26
41
1.84
12
2.79
27
1.80
30
10.88
36
3.16
37
0.41
22
0.33
20
0.39
16
0.36
21
0.59
35
0.58
35
PDISCO_ROBtwo views2.86
30
1.53
52
14.27
30
1.98
57
2.19
62
4.52
57
1.35
26
4.65
57
3.19
46
2.09
14
2.16
33
3.53
38
2.45
42
5.32
14
2.12
24
1.30
64
0.78
58
0.72
56
0.59
52
1.24
63
1.28
58
HSM-Net_RVCpermissivetwo views2.87
31
0.57
1
17.46
42
0.92
1
0.90
16
3.70
26
1.20
23
5.64
65
2.51
28
3.22
40
2.43
40
2.54
17
1.59
20
11.41
38
1.65
8
0.34
12
0.25
11
0.26
2
0.23
6
0.30
1
0.29
3
Gengshan Yang, Joshua Manela, Michael Happold, and Deva Ramanan: Hierarchical Deep Stereo Matching on High-resolution Images. CVPR 2019
ccstwo views2.91
32
1.17
37
18.05
44
1.53
43
0.91
17
3.88
34
0.80
7
3.79
15
3.00
42
1.84
4
2.13
32
2.42
14
1.25
5
13.71
60
1.58
4
0.33
9
0.27
15
0.40
19
0.35
17
0.39
7
0.34
8
MFMNet_retwo views2.91
32
1.91
58
18.25
46
3.06
64
1.89
56
3.13
3
1.67
36
3.52
5
2.97
41
3.14
39
2.87
42
2.26
9
2.41
40
2.42
3
2.00
21
1.10
61
1.05
61
0.94
61
0.90
59
1.28
64
1.46
61
GANettwo views2.93
34
1.05
25
11.29
25
1.64
47
0.96
19
3.91
36
2.52
50
3.72
12
2.41
22
2.95
36
3.00
44
3.21
35
2.03
36
12.42
53
4.72
46
0.51
37
0.41
33
0.46
29
0.39
30
0.60
38
0.50
23
PSMNet_ROBtwo views2.95
35
1.38
44
18.46
47
1.19
13
1.20
35
3.80
31
1.65
35
3.54
6
2.35
18
1.98
10
1.71
7
2.75
22
1.79
29
11.43
40
2.78
32
0.42
23
0.36
26
0.50
37
0.60
53
0.59
35
0.49
21
PWCDC_ROBbinarytwo views2.95
35
3.26
63
14.06
28
1.59
45
1.56
47
4.19
51
0.86
10
4.30
45
4.08
55
2.06
13
7.37
51
2.75
22
2.42
41
3.59
7
2.67
31
1.34
66
0.36
26
0.39
16
0.34
14
1.13
57
0.62
37
SGM-Foresttwo views2.96
37
0.62
3
3.49
8
1.09
5
0.80
6
4.06
45
3.30
54
4.37
49
3.09
44
4.18
51
3.27
45
4.67
51
3.86
49
11.69
45
7.91
54
0.51
37
0.46
41
0.49
35
0.39
30
0.48
19
0.49
21
Johannes L. Schönberger, Sudipta Sinha, Marc Pollefeys: Learning to Fuse Proposals from Multiple Scanline Optimizations in Semi-Global Matching. ECCV 2018
RYNettwo views2.97
38
0.93
16
15.09
33
1.12
8
1.38
43
3.66
25
1.42
28
4.03
29
2.29
15
2.10
15
1.70
4
3.79
41
2.16
38
13.64
59
3.02
36
0.37
17
0.41
33
0.48
32
0.34
14
0.81
52
0.75
48
AANet_RVCtwo views3.03
39
1.72
56
12.78
27
1.17
12
1.12
30
3.79
30
1.10
18
3.86
19
3.13
45
2.62
26
4.26
50
3.96
43
1.41
10
11.43
40
5.23
49
0.94
57
0.47
42
0.34
10
0.23
6
0.40
9
0.55
30
DRN-Testtwo views3.03
39
0.99
21
18.18
45
1.35
29
1.60
50
3.59
21
1.48
29
4.13
34
2.45
24
3.47
46
1.92
17
2.58
19
2.53
43
10.88
36
2.80
33
0.39
19
0.36
26
0.44
26
0.41
35
0.55
30
0.57
32
PA-Nettwo views3.09
41
1.24
40
18.84
48
1.38
32
1.19
34
3.42
11
1.75
37
3.89
22
3.27
50
1.88
7
2.10
30
3.03
34
1.88
34
11.60
42
2.88
34
0.40
20
0.54
46
0.44
26
0.62
54
0.52
24
1.02
55
Zhibo Rao, Mingyi He, Yuchao Dai, Zhelun Shen: Patch Attention Network with Generative Adversarial Model for Semi-Supervised Binocular Disparity Prediction.
Anonymous Stereotwo views3.09
41
3.14
62
16.69
38
1.52
42
1.25
39
3.33
9
2.95
52
4.02
28
2.42
23
2.35
21
2.04
24
2.57
18
1.27
6
11.88
48
2.92
35
0.49
33
0.49
45
0.57
47
0.56
49
0.70
48
0.65
44
GANetREF_RVCpermissivetwo views3.20
43
1.93
59
18.84
48
1.51
41
0.80
6
4.12
49
1.82
39
4.16
38
2.91
40
1.87
6
1.74
9
2.48
15
1.84
32
13.18
56
1.74
17
0.80
53
0.57
50
0.79
58
0.74
56
1.20
62
0.86
51
Zhang, Feihu and Prisacariu, Victor and Yang, Ruigang and Torr, Philip HS: GA-Net: Guided Aggregation Net for End- to-end Stereo Matching. CVPR 2019
FBW_ROBtwo views3.24
44
1.02
22
9.21
18
1.49
38
1.07
28
3.78
29
1.48
29
4.31
46
3.08
43
2.46
22
2.32
39
4.20
48
2.08
37
15.19
63
7.16
53
0.84
55
0.62
55
1.33
64
0.92
60
0.99
54
1.31
59
CBMVpermissivetwo views3.32
45
0.96
20
21.16
58
1.32
22
0.83
10
3.81
32
3.91
60
4.47
52
2.53
29
3.34
44
2.24
37
4.45
49
2.58
44
10.07
27
1.96
20
0.45
28
0.43
37
0.48
32
0.38
28
0.52
24
0.53
27
Konstantinos Batsos, Changjiang Cai, Philippos Mordohai: CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation. Computer Vision and Pattern Recognition (CVPR) 2018
DPSNettwo views3.43
46
1.37
43
12.19
26
1.39
33
1.24
37
4.15
50
2.06
43
4.51
53
3.34
51
2.69
28
1.96
18
3.52
37
5.56
55
10.68
32
9.68
58
0.93
56
0.73
56
0.40
19
0.37
27
1.06
56
0.86
51
CBMV_ROBtwo views3.62
47
0.70
4
21.42
62
1.21
15
0.89
15
4.43
56
1.76
38
4.39
50
2.79
35
5.88
55
3.38
46
4.18
47
5.12
53
10.75
33
2.32
28
0.56
44
0.55
48
0.53
43
0.51
44
0.55
30
0.46
18
SPS-STEREOcopylefttwo views3.97
48
1.44
49
20.85
56
1.72
49
1.71
53
3.24
6
2.30
48
3.55
7
2.68
33
4.79
53
3.66
48
5.44
53
2.97
45
9.11
21
9.35
57
1.15
62
1.10
62
1.06
62
1.02
63
1.16
60
1.18
57
K. Yamaguchi, D. McAllester, R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation. ECCV 2014
NaN_ROBtwo views4.06
49
1.13
34
16.64
36
1.12
8
1.25
39
3.50
18
3.05
53
4.29
43
3.26
48
3.64
47
2.56
41
2.97
32
4.79
52
13.33
57
16.91
63
0.37
17
0.41
33
0.38
12
0.39
30
0.48
19
0.81
49
SGM_RVCbinarytwo views4.57
50
0.72
5
17.46
42
1.84
54
0.58
3
4.23
52
3.45
55
4.66
58
3.66
53
7.99
58
8.82
58
7.56
56
8.74
61
12.30
51
6.84
51
0.43
24
0.38
30
0.39
16
0.35
17
0.51
22
0.48
20
Heiko Hirschmueller: Stereo processing by semiglobal matching and mutual information. TPAMI 2008, Volume 30(2), pp. 328-341
DispFullNettwo views4.64
51
3.48
64
23.95
66
3.99
65
3.03
66
3.35
10
1.06
17
4.18
39
2.80
37
6.33
56
2.94
43
4.04
44
7.81
59
3.92
8
5.16
48
2.75
67
1.61
68
3.10
68
2.68
67
3.32
67
3.31
67
CSANtwo views4.71
52
1.39
46
20.23
52
1.19
13
0.87
12
3.65
24
3.50
57
4.57
55
3.87
54
3.89
50
7.82
53
3.54
40
6.16
57
23.47
71
6.82
50
0.52
42
0.45
40
0.49
35
0.52
45
0.63
40
0.66
46
NVStereoNet_ROBtwo views4.91
53
1.40
47
16.46
35
1.47
37
1.72
54
3.46
14
1.96
41
3.89
22
3.61
52
2.90
33
24.61
72
4.81
52
4.49
51
13.57
58
4.87
47
1.33
65
1.29
66
1.70
66
1.68
66
1.41
66
1.66
63
Nikolai Smolyanskiy, Alexey Kamenev, Stan Birchfield: On the Importance of Stereo for Accurate Depth Estimation: An Efficient Semi-Supervised Deep Neural Network Approach. Arxiv
WCMA_ROBtwo views5.12
54
0.90
13
21.33
60
1.50
39
1.33
41
4.09
46
3.49
56
4.03
29
4.59
57
6.57
57
10.95
59
12.92
62
8.41
60
9.86
26
8.78
56
0.75
52
0.58
52
0.51
39
0.50
43
0.68
46
0.57
32
MeshStereopermissivetwo views5.35
55
1.32
41
21.34
61
1.34
26
1.15
32
4.40
55
3.90
59
5.18
61
4.55
56
10.14
64
11.82
60
11.47
60
5.84
56
13.84
61
7.03
52
0.61
48
0.58
52
0.60
51
0.58
51
0.72
49
0.59
36
C. Zhang, Z. Li, Y. Cheng, R. Cai, H. Chao, Y. Rui: MeshStereo: A Global Stereo Model with Mesh Alignment Regularization for View Interpolation. ICCV 2015
pmcnntwo views5.45
56
1.22
39
17.40
41
1.60
46
2.41
63
3.48
17
2.12
46
4.52
54
3.26
48
3.67
48
13.15
62
22.10
66
3.46
48
11.92
50
17.16
64
0.34
12
0.21
5
0.23
1
0.19
2
0.36
5
0.28
2
MDST_ROBtwo views5.53
57
0.59
2
23.11
64
2.55
62
2.69
64
21.20
68
1.98
42
4.77
60
2.65
31
9.36
61
4.25
49
4.14
46
4.45
50
22.57
67
3.68
41
0.48
32
0.33
20
0.47
30
0.44
39
0.44
13
0.44
16
MSMD_ROBtwo views5.61
58
1.08
28
16.68
37
1.21
15
0.83
10
10.14
63
2.07
45
3.93
25
6.22
61
10.37
65
8.03
55
22.22
67
9.45
62
11.66
44
4.52
45
0.66
50
0.58
52
0.66
54
0.68
55
0.65
44
0.63
41
SANettwo views5.65
59
1.72
56
20.94
57
1.26
19
0.88
13
7.20
60
3.81
58
4.60
56
11.23
64
4.56
52
8.78
57
7.49
55
6.95
58
11.85
47
18.11
65
0.49
33
0.44
39
0.50
37
0.42
38
0.90
53
0.82
50
ELAS_RVCcopylefttwo views5.72
60
1.64
55
20.40
54
2.09
60
2.01
60
4.95
58
11.47
64
4.66
58
6.31
62
8.09
59
8.01
54
8.06
59
11.65
65
8.36
19
10.22
60
0.99
60
1.17
63
0.81
59
0.76
57
1.14
58
1.58
62
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
PWCKtwo views5.85
61
9.95
70
15.22
34
2.90
63
1.56
47
5.66
59
7.69
62
5.27
63
4.74
58
3.88
49
7.51
52
8.04
57
5.28
54
15.18
62
8.40
55
3.08
68
1.51
67
2.24
67
1.37
65
5.55
68
1.96
66
ELAScopylefttwo views5.99
62
1.62
54
20.29
53
1.96
56
2.01
60
15.09
67
8.38
63
5.23
62
5.54
60
9.02
60
8.61
56
6.74
54
9.75
63
7.48
18
11.40
61
0.96
58
1.18
64
0.72
56
0.93
61
1.30
65
1.67
64
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
FC-DCNNcopylefttwo views6.70
63
0.85
12
19.25
50
1.39
33
1.62
51
9.01
62
2.44
49
5.59
64
9.41
63
11.88
67
12.83
61
22.42
68
11.52
64
12.39
52
9.71
59
0.66
50
0.57
50
0.59
49
0.56
49
0.64
43
0.62
37
SGM+DAISYtwo views8.36
64
3.82
66
24.83
67
2.02
58
1.75
55
14.80
66
17.22
66
4.27
42
11.79
66
13.88
68
14.38
65
11.74
61
13.83
66
10.56
30
15.12
62
1.21
63
1.19
65
1.14
63
1.13
64
1.17
61
1.37
60
MANEtwo views11.39
65
1.38
44
23.16
65
4.82
66
2.79
65
22.02
69
14.51
65
25.81
70
26.28
71
10.93
66
13.17
63
19.90
65
18.03
67
15.69
64
20.41
66
0.96
58
0.74
57
1.42
65
3.24
68
0.74
50
1.71
65
SGM-ForestMtwo views11.55
66
1.16
36
21.62
63
2.14
61
0.81
8
22.25
70
19.08
67
16.50
69
11.34
65
15.94
71
14.44
66
27.12
71
22.61
72
24.76
72
28.39
71
0.50
36
0.48
44
0.47
30
0.44
39
0.55
30
0.53
27
LSMtwo views12.71
67
3.52
65
20.79
55
55.63
74
84.76
74
3.63
23
6.30
61
3.73
13
4.93
59
5.55
54
13.92
64
3.53
38
3.25
47
9.72
24
4.12
43
0.64
49
0.85
59
0.71
55
0.81
58
1.03
55
26.84
72
DGTPSM_ROBtwo views15.31
68
8.85
68
28.14
70
9.90
68
18.89
67
12.44
64
33.21
70
13.48
67
17.86
69
9.50
62
22.89
70
13.78
63
19.67
68
21.26
65
23.68
67
5.08
69
8.84
69
5.62
69
11.50
69
6.72
69
14.87
68
DPSMNet_ROBtwo views15.52
69
8.89
69
31.44
71
10.07
69
18.91
68
12.44
64
33.22
71
13.51
68
17.94
70
9.58
63
22.91
71
13.79
64
19.67
68
21.43
66
23.73
68
5.17
70
8.84
69
5.63
70
11.52
72
6.79
70
14.90
69
LE_ROBtwo views18.58
70
1.21
38
48.38
73
8.60
67
23.00
71
7.32
61
19.74
68
9.94
66
38.69
73
43.53
73
21.91
69
28.40
72
40.92
74
23.18
68
54.71
74
0.34
12
0.33
20
0.32
7
0.36
21
0.42
11
0.40
13
DPSMtwo views20.78
71
20.03
71
26.57
68
23.62
70
22.46
69
31.10
71
39.97
73
33.52
72
17.41
67
15.25
69
15.80
67
24.06
69
20.71
70
23.30
69
27.62
69
10.78
71
9.41
71
11.69
71
11.50
69
14.49
71
16.36
70
DPSM_ROBtwo views20.78
71
20.03
71
26.57
68
23.62
70
22.46
69
31.10
71
39.97
73
33.52
72
17.41
67
15.25
69
15.80
67
24.06
69
20.71
70
23.30
69
27.62
69
10.78
71
9.41
71
11.69
71
11.50
69
14.49
71
16.36
70
AVERAGE_ROBtwo views44.46
73
45.91
74
46.59
72
40.59
73
38.66
72
32.99
73
30.52
69
46.19
74
43.26
74
49.59
74
48.97
74
44.62
73
39.89
73
45.04
73
43.94
72
48.74
74
50.69
74
50.80
74
49.79
74
45.54
74
46.97
73
MEDIAN_ROBtwo views47.37
74
49.08
75
49.92
74
40.32
72
39.97
73
34.95
74
33.43
72
49.17
75
46.39
75
53.06
75
52.65
75
47.65
74
42.23
76
48.35
75
47.29
73
52.10
75
54.04
75
54.28
75
53.48
75
48.76
75
50.33
74
LSM0two views49.58
75
40.04
73
53.62
75
97.15
75
128.68
76
62.75
75
79.89
75
67.63
76
35.11
72
33.02
72
32.61
73
48.43
75
41.60
75
46.60
74
55.43
75
21.63
73
18.99
73
23.62
73
23.13
73
29.47
73
52.26
75
DPSimNet_ROBtwo views102.30
76
119.72
76
143.59
76
109.42
76
94.10
75
127.22
76
127.15
76
26.62
71
102.36
76
149.58
76
83.99
76
139.34
76
137.54
77
64.62
76
82.36
76
105.18
76
66.15
76
120.15
76
69.61
76
99.58
76
77.72
76
MSMDNettwo views1.11
2