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
R-Stereotwo views1.20
1
0.58
4
1.81
2
0.99
5
0.67
7
3.78
46
0.90
18
4.09
59
1.51
1
2.16
25
1.71
17
1.13
3
0.94
2
1.40
2
0.91
1
0.30
2
0.24
13
0.26
4
0.21
4
0.26
1
0.25
2
R-Stereo Traintwo views1.20
1
0.58
4
1.81
2
0.99
5
0.67
7
3.78
46
0.90
18
4.09
59
1.51
1
2.16
25
1.71
17
1.13
3
0.94
2
1.40
2
0.91
1
0.30
2
0.24
13
0.26
4
0.21
4
0.26
1
0.25
2
DPM-Stereotwo views1.24
3
0.69
8
1.77
1
0.65
1
0.61
6
4.29
81
0.58
2
4.44
90
2.53
49
1.98
16
1.33
3
0.91
1
0.86
1
1.25
1
1.25
4
0.36
19
0.23
10
0.32
11
0.23
9
0.32
7
0.29
7
AdaStereotwo views1.45
4
0.83
18
2.34
7
1.28
37
0.81
14
3.21
7
1.14
33
4.20
71
2.46
44
2.05
20
1.77
22
2.48
24
1.18
7
1.71
4
1.44
5
0.43
35
0.25
17
0.43
33
0.28
15
0.44
17
0.34
13
Xiao Song, Guorun Yang, Xinge Zhu, Hui Zhou, Zhe Wang, Jianping Shi: AdaStereo: A Simple and Efficient Approach for Adaptive Stereo Matching. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2021.
ccstwo views1.61
5
0.55
2
2.29
6
1.27
36
0.80
11
3.62
36
0.75
8
4.22
74
2.77
56
1.94
15
1.60
9
2.54
28
1.46
19
3.60
14
1.72
19
0.40
30
0.60
81
0.38
18
0.39
35
0.45
21
0.77
74
StereoDRNet-Refinedtwo views1.77
6
0.81
15
2.22
5
1.21
27
0.76
10
3.26
11
0.71
5
3.82
23
2.05
8
2.29
30
2.05
40
2.98
47
1.44
15
7.15
37
2.27
37
0.30
2
0.23
10
0.42
31
0.33
19
0.45
21
0.64
55
Rohan Chabra, Julian Straub, Chris Sweeney, Richard Newcombe, Henry Fuchs: StereoDRNet. CVPR
MLCVtwo views1.87
7
0.91
21
8.59
19
1.10
14
0.57
4
3.59
34
0.53
1
3.70
19
2.20
16
2.80
55
2.25
55
2.00
11
1.42
14
4.13
18
1.86
26
0.30
2
0.20
3
0.32
11
0.21
4
0.33
9
0.36
15
iResNet_ROBtwo views1.89
8
1.02
31
8.60
20
1.39
58
0.98
32
3.72
44
0.66
4
3.84
24
2.65
52
1.88
12
2.05
40
1.98
10
1.27
9
4.52
20
1.67
15
0.30
2
0.20
3
0.28
7
0.18
2
0.31
5
0.40
18
CFNet_RVCtwo views1.90
9
0.95
25
3.56
13
1.07
9
1.54
66
4.09
71
0.90
18
2.85
2
1.82
7
1.78
8
1.70
13
2.75
35
1.48
20
9.33
52
1.72
19
0.33
12
0.30
22
0.61
73
0.36
26
0.52
30
0.39
17
DN-CSS_ROBtwo views1.91
10
1.55
72
11.13
32
1.82
87
0.88
20
3.99
66
0.61
3
4.01
48
1.67
3
1.64
6
1.89
30
1.39
6
1.06
5
3.23
9
1.50
7
0.31
7
0.19
2
0.44
35
0.29
16
0.37
11
0.31
10
BEATNet_4xtwo views1.92
11
1.36
60
8.75
21
1.20
26
0.56
2
3.46
22
1.03
26
4.00
47
2.43
39
2.25
29
1.70
13
2.10
16
1.59
26
2.57
8
3.05
57
0.37
20
0.22
8
0.47
41
0.29
16
0.60
48
0.40
18
LALA_ROBtwo views1.95
12
1.48
69
3.11
11
1.05
8
0.98
32
3.99
66
1.82
64
4.44
90
2.28
27
3.31
70
1.85
28
2.76
39
1.67
33
4.04
17
3.23
59
0.46
42
0.41
48
0.55
63
0.52
64
0.60
48
0.57
44
DeepPruner_ROBtwo views2.05
13
1.09
40
10.50
29
1.01
7
1.20
45
4.39
83
0.91
22
3.14
4
1.72
5
2.60
43
1.54
7
2.78
40
1.27
9
4.52
20
1.72
19
0.51
53
0.40
45
0.37
16
0.34
20
0.48
25
0.52
34
DLCB_ROBtwo views2.05
13
0.95
25
2.37
8
1.24
30
1.02
36
2.92
5
1.30
39
3.68
17
2.30
29
2.91
58
2.17
49
3.32
52
1.52
24
10.37
67
2.50
41
0.40
30
0.33
25
0.40
26
0.36
26
0.41
14
0.47
25
ETE_ROBtwo views2.07
15
1.40
65
3.01
10
1.09
10
0.92
25
3.94
59
1.53
52
4.25
75
2.23
19
3.31
70
2.06
42
2.35
20
1.52
24
8.74
44
2.07
32
0.44
38
0.36
35
0.51
52
0.48
57
0.54
35
0.62
51
HITNettwo views2.07
15
0.82
17
14.77
50
1.34
48
0.56
2
3.47
24
0.80
9
4.07
56
2.35
32
2.16
25
1.61
10
1.64
8
1.44
15
2.09
5
2.56
43
0.26
1
0.18
1
0.41
30
0.26
14
0.45
21
0.25
2
Vladimir Tankovich, Christian Häne, Yinda Zhang, Adarsh Kowdle, Sean Fanello, Sofien Bouaziz: HITNet: Hierarchical Iterative Tile Refinement Network for Real-time Stereo Matching. CVPR 2021
PMTNettwo views2.07
15
0.52
1
19.44
88
0.73
2
0.45
1
4.46
86
0.89
17
2.78
1
2.25
21
1.63
5
1.52
5
0.94
2
0.94
2
2.47
7
0.97
3
0.32
9
0.20
3
0.23
1
0.16
1
0.31
5
0.22
1
DMCAtwo views2.09
18
1.05
34
6.92
16
1.16
20
1.61
76
2.41
1
0.91
22
3.18
5
1.70
4
3.46
74
1.52
5
2.43
23
1.69
34
9.58
54
1.87
27
0.39
28
0.34
32
0.45
39
0.41
44
0.46
24
0.36
15
DISCOtwo views2.25
19
0.73
12
10.06
26
1.76
85
1.12
40
3.47
24
1.25
38
3.24
6
2.26
24
2.65
45
1.96
32
4.53
82
1.92
45
4.17
19
3.61
61
0.32
9
0.25
17
0.31
10
0.29
16
0.66
60
0.41
21
NOSS_ROBtwo views2.27
20
0.73
12
3.00
9
1.71
82
1.00
34
3.90
55
0.97
24
4.29
79
2.81
63
2.94
60
2.19
51
3.88
65
1.19
8
11.77
88
1.62
10
0.59
64
0.54
71
0.56
64
0.52
64
0.56
41
0.54
38
XPNet_ROBtwo views2.27
20
1.11
43
4.10
14
1.36
54
0.97
30
3.94
59
1.63
57
4.07
56
2.14
12
2.82
56
2.00
35
2.32
19
1.73
36
10.81
72
3.39
60
0.51
53
0.47
63
0.48
45
0.40
42
0.51
28
0.56
41
TDLMtwo views2.29
22
1.08
38
4.61
15
1.29
40
1.01
35
4.26
80
2.17
73
3.75
22
2.79
59
2.31
32
1.98
34
2.06
15
1.59
26
12.45
95
1.72
19
0.51
53
0.28
21
0.52
58
0.35
23
0.56
41
0.50
30
CFNettwo views2.34
23
1.10
41
8.96
22
1.34
48
1.24
47
4.09
71
0.73
7
4.04
52
2.77
56
1.62
4
1.87
29
2.27
18
1.45
18
11.41
77
1.82
25
0.31
7
0.24
13
0.43
33
0.36
26
0.43
16
0.33
11
STTStereotwo views2.34
23
0.95
25
14.93
55
1.43
60
1.54
66
3.86
54
0.86
12
3.65
15
2.44
41
2.15
24
1.55
8
2.49
26
1.59
26
5.14
24
1.44
5
0.38
24
0.39
44
0.47
41
0.51
62
0.52
30
0.51
33
HSMtwo views2.35
25
0.81
15
2.21
4
1.14
18
1.55
69
3.44
19
1.03
26
3.99
46
2.25
21
2.29
30
2.11
46
8.04
100
3.05
72
11.60
81
1.71
17
0.33
12
0.22
8
0.28
7
0.23
9
0.32
7
0.35
14
FADNettwo views2.49
26
1.76
80
13.17
39
1.17
21
0.95
27
3.27
13
0.88
16
3.94
43
2.27
26
1.40
2
1.29
2
2.51
27
2.06
47
9.93
61
1.94
28
0.43
35
0.45
59
0.40
26
0.43
49
1.06
90
0.53
35
FADNet-RVCtwo views2.51
27
2.27
90
14.03
43
1.09
10
0.81
14
3.54
32
1.10
30
4.13
63
2.48
45
1.39
1
1.22
1
1.85
9
1.77
38
9.78
57
2.06
31
0.38
24
0.37
40
0.40
26
0.40
42
0.67
61
0.48
26
iResNettwo views2.52
28
0.94
24
21.17
101
1.84
89
0.72
9
3.75
45
0.81
10
4.14
66
2.55
51
2.33
33
2.03
36
1.35
5
1.44
15
3.27
10
2.16
35
0.35
18
0.23
10
0.28
7
0.22
8
0.44
17
0.45
23
NCCL2two views2.53
29
1.35
59
10.90
30
1.15
19
0.97
30
3.53
30
2.56
78
3.65
15
2.23
19
2.76
51
1.80
25
2.41
22
1.66
32
10.35
66
2.07
32
0.44
38
0.40
45
0.57
65
0.53
67
0.59
45
0.65
58
ccs_robtwo views2.54
30
1.14
45
14.21
46
1.35
51
0.96
28
3.98
65
0.71
5
3.86
27
2.88
66
2.04
19
2.07
43
2.05
14
1.29
12
10.60
69
1.71
17
0.32
9
0.25
17
0.38
18
0.35
23
0.39
12
0.33
11
NLCA_NET_v2_RVCtwo views2.54
30
1.04
33
16.73
70
1.53
70
2.00
96
3.96
62
0.87
15
3.87
29
2.31
30
2.54
39
1.67
12
2.85
43
1.63
30
5.55
28
1.62
10
0.46
42
0.33
25
0.38
18
0.38
33
0.53
34
0.62
51
Zhibo Rao, Mingyi He, Yuchao Dai, Zhidong Zhu, Bo Li, and Renjie He.: NLCA-Net: A non-local context attention network for stereo matching.
NVstereo2Dtwo views2.54
30
0.90
19
11.10
31
1.32
42
1.47
64
3.58
33
1.38
43
4.13
63
2.45
42
1.90
14
1.71
17
2.86
44
1.76
37
10.86
73
1.66
14
0.60
65
0.43
54
0.53
59
0.41
44
0.78
72
0.94
86
CVANet_RVCtwo views2.55
33
1.10
41
9.13
23
1.46
62
1.15
42
3.96
62
1.53
52
3.91
39
2.50
47
3.04
62
2.04
38
2.60
32
1.72
35
12.46
96
1.50
7
0.49
48
0.30
22
0.51
52
0.41
44
0.63
51
0.50
30
CC-Net-ROBtwo views2.55
33
1.05
34
16.77
71
1.50
66
1.94
95
3.97
64
0.90
18
3.85
25
2.25
21
2.55
40
1.70
13
2.82
42
1.65
31
5.67
31
1.65
12
0.47
45
0.33
25
0.38
18
0.39
35
0.54
35
0.64
55
RASNettwo views2.56
35
0.70
9
10.06
26
1.60
75
2.25
102
3.67
40
1.00
25
3.38
9
3.14
75
1.81
9
2.11
46
4.60
84
2.30
53
10.25
65
2.61
44
0.37
20
0.24
13
0.25
3
0.24
13
0.28
3
0.26
5
DANettwo views2.63
36
1.07
37
8.10
17
1.83
88
1.70
82
3.25
10
1.16
34
4.31
84
2.38
36
3.40
73
2.18
50
4.08
71
1.80
40
9.77
56
3.89
64
0.52
58
0.55
73
0.59
68
0.44
50
0.63
51
0.86
80
iResNetv2_ROBtwo views2.65
37
5.00
101
19.28
87
2.06
99
1.09
39
3.83
52
1.13
32
4.14
66
2.09
10
1.85
10
2.03
36
2.01
12
1.51
22
3.58
12
1.58
9
0.34
14
0.21
6
0.32
11
0.21
4
0.44
17
0.29
7
FADNet_RVCtwo views2.65
37
2.28
91
16.51
66
0.98
4
0.92
25
4.09
71
0.83
11
4.26
76
1.73
6
1.46
3
1.78
24
1.60
7
1.50
21
8.88
46
2.39
39
0.47
45
0.38
42
0.44
35
0.58
75
0.89
80
0.97
87
PASMtwo views2.67
39
2.05
88
9.71
25
1.73
84
1.37
57
2.53
2
1.52
51
3.10
3
2.17
15
2.79
54
1.84
26
2.90
45
1.85
43
11.89
91
2.19
36
0.81
89
0.85
98
0.87
99
1.01
101
1.14
94
1.05
91
PWC_ROBbinarytwo views2.69
40
2.01
85
8.36
18
1.33
44
1.90
93
3.23
8
1.18
35
4.33
86
3.25
82
3.11
65
3.41
84
2.71
33
2.18
51
9.15
50
4.48
67
0.56
62
0.31
24
0.61
73
0.36
26
0.69
64
0.74
68
RPtwo views2.79
41
1.20
48
10.36
28
1.54
71
1.93
94
3.44
19
1.97
67
4.05
54
2.34
31
2.66
46
2.79
67
3.98
68
2.58
62
10.21
64
2.61
44
0.95
95
0.47
63
0.71
86
0.57
74
0.71
68
0.68
63
SHDtwo views2.82
42
2.04
87
13.80
41
1.69
81
1.79
90
2.73
4
1.07
29
3.85
25
3.18
77
4.63
90
2.74
65
3.38
53
2.77
67
5.48
27
3.66
62
0.62
69
0.47
63
0.51
52
0.55
69
0.65
56
0.74
68
FADNet-RVC-Resampletwo views2.83
43
2.02
86
21.07
98
1.12
15
0.91
24
3.53
30
1.37
42
4.08
58
2.09
10
1.68
7
1.33
3
2.03
13
1.51
22
9.54
53
1.69
16
0.38
24
0.37
40
0.42
31
0.44
50
0.56
41
0.56
41
StereoDRNettwo views2.84
44
1.49
70
14.48
49
1.33
44
1.57
74
3.44
19
2.06
69
3.69
18
2.36
34
3.26
69
1.84
26
2.79
41
1.80
40
10.88
74
3.16
58
0.41
33
0.33
25
0.39
23
0.36
26
0.59
45
0.58
47
PDISCO_ROBtwo views2.86
45
1.53
71
14.27
48
1.98
96
2.19
99
4.52
87
1.35
41
4.65
97
3.19
78
2.09
22
2.16
48
3.53
58
2.45
60
5.32
26
2.12
34
1.30
106
0.78
95
0.72
89
0.59
77
1.24
100
1.28
96
AF-Nettwo views2.87
46
1.30
57
16.24
61
1.33
44
1.75
86
3.26
11
1.59
56
4.09
59
2.43
39
2.69
47
2.94
71
4.51
81
3.20
76
5.84
33
2.75
49
0.76
84
0.36
35
0.72
89
0.49
58
0.70
65
0.56
41
HSM-Net_RVCpermissivetwo views2.87
46
0.57
3
17.46
74
0.92
3
0.90
23
3.70
41
1.20
37
5.64
107
2.51
48
3.22
68
2.43
57
2.54
28
1.59
26
11.41
77
1.65
12
0.34
14
0.25
17
0.26
4
0.23
9
0.30
4
0.29
7
Gengshan Yang, Joshua Manela, Michael Happold, and Deva Ramanan: Hierarchical Deep Stereo Matching on High-resolution Images. CVPR 2019
Nwc_Nettwo views2.88
48
1.16
46
18.30
80
1.35
51
1.68
80
3.52
29
1.18
35
3.90
37
2.06
9
2.52
38
2.82
69
3.82
63
2.34
56
6.78
36
2.55
42
0.86
92
0.40
45
0.59
68
0.67
81
0.57
44
0.55
39
MFMNet_retwo views2.91
49
1.91
81
18.25
79
3.06
106
1.89
92
3.13
6
1.67
60
3.52
11
2.97
69
3.14
66
2.87
70
2.26
17
2.41
57
2.42
6
2.00
30
1.10
99
1.05
100
0.94
101
0.90
97
1.28
101
1.46
101
GANettwo views2.93
50
1.05
34
11.29
33
1.64
78
0.96
28
3.91
56
2.52
77
3.72
20
2.41
37
2.95
61
3.00
73
3.21
51
2.03
46
12.42
94
4.72
70
0.51
53
0.41
48
0.46
40
0.39
35
0.60
48
0.50
30
PSMNet_ROBtwo views2.95
51
1.38
62
18.46
83
1.19
24
1.20
45
3.80
50
1.65
59
3.54
12
2.35
32
1.98
16
1.71
17
2.75
35
1.79
39
11.43
79
2.78
50
0.42
34
0.36
35
0.50
50
0.60
78
0.59
45
0.49
28
RTSCtwo views2.95
51
2.51
93
15.49
60
1.54
71
1.30
53
3.51
28
0.86
12
3.89
32
2.48
45
3.64
76
3.12
79
3.04
49
2.30
53
5.70
32
6.67
78
0.55
61
0.35
34
0.37
16
0.39
35
0.70
65
0.60
49
PWCDC_ROBbinarytwo views2.95
51
3.26
95
14.06
44
1.59
74
1.56
72
4.19
77
0.86
12
4.30
83
4.08
94
2.06
21
7.37
91
2.75
35
2.42
58
3.59
13
2.67
47
1.34
108
0.36
35
0.39
23
0.34
20
1.13
93
0.62
51
SGM-Foresttwo views2.96
54
0.62
7
3.49
12
1.09
10
0.80
11
4.06
68
3.30
89
4.37
87
3.09
73
4.18
85
3.27
81
4.67
85
3.86
87
11.69
86
7.91
87
0.51
53
0.46
61
0.49
48
0.39
35
0.48
25
0.49
28
Johannes L. Schönberger, Sudipta Sinha, Marc Pollefeys: Learning to Fuse Proposals from Multiple Scanline Optimizations in Semi-Global Matching. ECCV 2018
XQCtwo views2.96
54
2.42
92
14.85
51
1.36
54
1.32
55
3.29
14
1.53
52
4.28
78
2.78
58
3.21
67
2.46
58
3.53
58
2.92
70
6.58
34
4.67
69
0.70
81
0.41
48
0.59
68
0.56
71
0.96
87
0.83
79
RYNettwo views2.97
56
0.93
22
15.09
56
1.12
15
1.38
59
3.66
39
1.42
48
4.03
50
2.29
28
2.10
23
1.70
13
3.79
62
2.16
50
13.64
101
3.02
56
0.37
20
0.41
48
0.48
45
0.34
20
0.81
75
0.75
70
DRN-Testtwo views3.03
57
0.99
29
18.18
78
1.35
51
1.60
75
3.59
34
1.48
49
4.13
63
2.45
42
3.47
75
1.92
31
2.58
31
2.53
61
10.88
74
2.80
51
0.39
28
0.36
35
0.44
35
0.41
44
0.55
38
0.57
44
AANet_RVCtwo views3.03
57
1.72
77
12.78
37
1.17
21
1.12
40
3.79
49
1.10
30
3.86
27
3.13
74
2.62
44
4.26
89
3.96
67
1.41
13
11.43
79
5.23
76
0.94
94
0.47
63
0.34
15
0.23
9
0.40
13
0.55
39
Anonymous Stereotwo views3.09
59
3.14
94
16.69
69
1.52
69
1.25
49
3.33
15
2.95
82
4.02
49
2.42
38
2.35
34
2.04
38
2.57
30
1.27
9
11.88
90
2.92
53
0.49
48
0.49
69
0.57
65
0.56
71
0.70
65
0.65
58
PA-Nettwo views3.09
59
1.24
54
18.84
84
1.38
57
1.19
44
3.42
18
1.75
62
3.89
32
3.27
85
1.88
12
2.10
44
3.03
48
1.88
44
11.60
81
2.88
52
0.40
30
0.54
71
0.44
35
0.62
79
0.52
30
1.02
89
Zhibo Rao, Mingyi He, Yuchao Dai, Zhelun Shen: Patch Attention Network with Generative Adversarial Model for Semi-Supervised Binocular Disparity Prediction.
stereogantwo views3.09
59
0.93
22
16.43
64
1.37
56
1.66
79
3.70
41
1.67
60
3.87
29
3.15
76
3.04
62
2.70
64
5.04
91
2.79
68
9.96
62
1.78
24
0.63
70
0.49
69
0.69
82
0.47
54
0.83
78
0.68
63
RGCtwo views3.10
62
1.45
68
14.90
53
1.46
62
1.79
90
3.33
15
1.39
44
4.04
52
2.26
24
2.77
52
3.26
80
4.00
69
2.71
66
11.63
84
2.62
46
0.82
90
0.46
61
0.79
94
0.79
93
0.71
68
0.78
75
PVDtwo views3.12
63
1.97
84
12.80
38
1.79
86
1.63
77
3.83
52
1.34
40
3.48
10
3.88
93
5.77
94
3.56
85
3.45
54
3.29
78
5.58
30
5.30
77
0.71
82
0.69
92
0.74
92
0.77
90
0.79
74
1.13
93
NCC-stereotwo views3.16
64
1.28
55
18.31
81
1.25
31
1.75
86
4.06
68
1.40
46
3.89
32
2.22
17
2.72
49
2.55
59
3.49
55
2.65
64
9.91
59
2.99
54
1.29
104
0.65
87
0.64
78
0.86
95
0.65
56
0.71
65
Abc-Nettwo views3.16
64
1.28
55
18.31
81
1.25
31
1.75
86
4.06
68
1.40
46
3.89
32
2.22
17
2.72
49
2.55
59
3.49
55
2.65
64
9.91
59
2.99
54
1.29
104
0.65
87
0.64
78
0.86
95
0.65
56
0.71
65
GANetREF_RVCpermissivetwo views3.20
66
1.93
82
18.84
84
1.51
68
0.80
11
4.12
75
1.82
64
4.16
68
2.91
68
1.87
11
1.74
21
2.48
24
1.84
42
13.18
98
1.74
23
0.80
87
0.57
75
0.79
94
0.74
87
1.20
99
0.86
80
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
67
1.02
31
9.21
24
1.49
65
1.07
37
3.78
46
1.48
49
4.31
84
3.08
72
2.46
35
2.32
56
4.20
75
2.08
48
15.19
106
7.16
86
0.84
91
0.62
84
1.33
107
0.92
98
0.99
88
1.31
97
ADCReftwo views3.30
68
1.23
53
14.90
53
1.17
21
1.41
62
3.95
61
1.64
58
3.88
31
2.36
34
3.05
64
2.21
53
2.40
21
2.81
69
5.04
22
16.63
100
0.46
42
0.34
32
0.63
75
0.62
79
0.54
35
0.67
61
CBMVpermissivetwo views3.32
69
0.96
28
21.16
100
1.32
42
0.83
17
3.81
51
3.91
99
4.47
92
2.53
49
3.34
72
2.24
54
4.45
79
2.58
62
10.07
63
1.96
29
0.45
41
0.43
54
0.48
45
0.38
33
0.52
30
0.53
35
Konstantinos Batsos, Changjiang Cai, Philippos Mordohai: CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation. Computer Vision and Pattern Recognition (CVPR) 2018
edge stereotwo views3.37
70
1.01
30
21.99
106
1.28
37
1.37
57
2.67
3
1.39
44
3.98
45
2.87
65
2.93
59
3.27
81
4.35
78
2.43
59
11.62
83
2.39
39
0.60
65
0.63
85
0.79
94
0.55
69
0.68
62
0.67
61
DPSNettwo views3.43
71
1.37
61
12.19
34
1.39
58
1.24
47
4.15
76
2.06
69
4.51
93
3.34
86
2.69
47
1.96
32
3.52
57
5.56
95
10.68
70
9.68
91
0.93
93
0.73
93
0.40
26
0.37
32
1.06
90
0.86
80
SAMSARAtwo views3.55
72
2.16
89
14.18
45
1.86
91
1.49
65
4.62
88
2.65
79
3.58
14
2.80
61
7.05
100
3.05
76
5.16
92
4.07
88
9.18
51
4.80
71
0.64
72
0.67
90
0.59
68
0.68
83
0.89
80
0.91
85
CBMV_ROBtwo views3.62
73
0.70
9
21.42
104
1.21
27
0.89
22
4.43
85
1.76
63
4.39
88
2.79
59
5.88
95
3.38
83
4.18
74
5.12
93
10.75
71
2.32
38
0.56
62
0.55
73
0.53
59
0.51
62
0.55
38
0.46
24
DeepPrunerFtwo views3.70
74
1.72
77
14.87
52
1.33
44
1.74
85
3.70
41
1.55
55
4.20
71
14.00
109
2.02
18
1.77
22
2.74
34
2.30
53
10.88
74
6.88
83
0.68
79
0.60
81
0.78
93
0.78
92
0.81
75
0.71
65
ADCP+two views3.74
75
1.96
83
16.90
72
1.09
10
1.55
69
4.35
82
3.65
96
3.93
40
2.98
70
2.22
28
1.66
11
3.95
66
4.83
92
3.41
11
18.87
106
0.38
24
0.41
48
0.63
75
0.53
67
0.64
55
0.81
76
ADCPNettwo views3.89
76
1.63
74
20.25
92
1.28
37
1.27
52
6.43
101
2.65
79
3.93
40
2.70
55
2.78
53
2.20
52
4.28
76
6.16
98
5.56
29
10.27
94
0.65
74
1.18
103
0.71
86
1.70
108
0.84
79
1.37
99
RTStwo views3.90
77
7.04
105
17.94
76
1.92
93
2.19
99
5.00
93
2.96
85
4.29
79
3.19
78
4.05
82
3.08
77
4.94
88
3.16
74
7.64
39
6.72
79
0.66
75
0.44
56
0.51
52
0.47
54
0.95
83
0.76
71
RTSAtwo views3.90
77
7.04
105
17.94
76
1.92
93
2.19
99
5.00
93
2.96
85
4.29
79
3.19
78
4.05
82
3.08
77
4.94
88
3.16
74
7.64
39
6.72
79
0.66
75
0.44
56
0.51
52
0.47
54
0.95
83
0.76
71
MFN_U_SF_DS_RVCtwo views3.94
79
10.88
113
12.30
35
1.26
33
1.55
69
5.19
95
3.33
90
4.42
89
2.82
64
2.58
41
2.62
62
5.00
90
2.11
49
14.03
104
2.73
48
1.28
103
2.34
110
0.67
81
1.83
109
0.81
75
1.03
90
ADCMidtwo views3.96
80
3.27
96
15.48
59
1.26
33
1.40
60
4.66
89
2.28
74
4.06
55
3.54
88
3.94
81
3.02
74
3.85
64
3.67
84
4.03
16
18.65
104
0.68
79
0.61
83
1.09
105
1.18
104
1.38
103
1.08
92
SPS-STEREOcopylefttwo views3.97
81
1.44
67
20.85
96
1.72
83
1.71
83
3.24
9
2.30
75
3.55
13
2.68
54
4.79
91
3.66
86
5.44
93
2.97
71
9.11
49
9.35
90
1.15
100
1.10
101
1.06
104
1.02
102
1.16
96
1.18
94
K. Yamaguchi, D. McAllester, R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation. ECCV 2014
MADNet+two views3.97
81
9.54
110
13.83
42
1.87
92
1.69
81
4.19
77
2.87
81
4.80
101
4.10
95
2.51
37
2.67
63
4.55
83
3.41
80
8.72
43
6.92
84
1.25
102
1.27
106
1.04
102
0.93
99
1.39
104
1.85
107
NaN_ROBtwo views4.06
83
1.13
44
16.64
67
1.12
15
1.25
49
3.50
27
3.05
87
4.29
79
3.26
83
3.64
76
2.56
61
2.97
46
4.79
91
13.33
99
16.91
101
0.37
20
0.41
48
0.38
18
0.39
35
0.48
25
0.81
76
STTStereo_v2two views4.11
84
1.21
49
19.47
89
2.31
102
2.94
106
4.74
90
2.95
82
3.33
7
2.14
12
4.44
87
9.43
100
6.93
96
3.84
85
9.01
47
4.92
73
0.79
85
0.80
96
0.70
84
0.50
59
0.95
83
0.87
83
G-Nettwo views4.11
84
1.21
49
19.47
89
2.31
102
2.94
106
4.74
90
2.95
82
3.33
7
2.14
12
4.44
87
9.43
100
6.93
96
3.84
85
9.01
47
4.92
73
0.79
85
0.80
96
0.70
84
0.50
59
0.95
83
0.87
83
AnyNet_C32two views4.17
86
5.75
103
12.62
36
1.60
75
1.54
66
5.51
96
4.09
100
3.90
37
3.23
81
4.22
86
3.03
75
3.10
50
3.41
80
8.01
41
18.85
105
0.53
60
0.58
77
0.53
59
0.74
87
1.19
98
0.97
87
ADCStwo views4.22
87
5.33
102
16.32
62
1.64
78
1.31
54
3.91
56
3.64
95
4.21
73
2.89
67
5.11
92
2.81
68
4.46
80
3.06
73
5.21
25
19.50
107
0.63
70
0.64
86
0.69
82
0.69
85
1.12
92
1.26
95
ADCLtwo views4.35
88
1.63
74
14.21
46
1.29
40
1.26
51
5.56
97
11.16
106
3.97
44
3.37
87
6.07
96
4.67
90
4.30
77
5.72
96
5.08
23
14.51
97
0.50
51
0.48
67
0.89
100
0.77
90
0.78
72
0.76
71
AnyNet_C01two views4.37
89
7.80
107
16.40
63
1.98
96
1.40
60
6.37
100
6.47
102
4.09
59
5.03
100
4.06
84
3.78
87
4.16
73
3.31
79
6.62
35
10.00
92
0.61
67
0.67
90
0.63
75
0.67
81
2.01
106
1.31
97
MSC_U_SF_DS_RVCtwo views4.48
90
10.62
112
15.18
57
1.55
73
1.45
63
6.16
99
3.54
94
5.41
106
3.68
91
2.59
42
2.74
65
4.81
86
2.20
52
13.79
102
4.40
66
2.04
109
3.36
111
1.05
103
1.20
105
2.09
107
1.72
106
SGM_RVCbinarytwo views4.57
91
0.72
11
17.46
74
1.84
89
0.58
5
4.23
79
3.45
91
4.66
98
3.66
90
7.99
101
8.82
99
7.56
99
8.74
104
12.30
93
6.84
82
0.43
35
0.38
42
0.39
23
0.35
23
0.51
28
0.48
26
Heiko Hirschmueller: Stereo processing by semiglobal matching and mutual information. TPAMI 2008, Volume 30(2), pp. 328-341
DispFullNettwo views4.64
92
3.48
97
23.95
109
3.99
107
3.03
108
3.35
17
1.06
28
4.18
69
2.80
61
6.33
98
2.94
71
4.04
70
7.81
101
3.92
15
5.16
75
2.75
110
1.61
109
3.10
111
2.68
110
3.32
108
3.31
110
CSANtwo views4.71
93
1.39
64
20.23
91
1.19
24
0.87
19
3.65
38
3.50
93
4.57
95
3.87
92
3.89
80
7.82
94
3.54
61
6.16
98
23.47
114
6.82
81
0.52
58
0.45
59
0.49
48
0.52
64
0.63
51
0.66
60
SuperBtwo views4.79
94
4.41
100
21.08
99
1.45
61
1.08
38
3.91
56
3.08
88
4.18
69
3.01
71
2.50
36
2.10
44
2.75
35
3.49
83
8.80
45
19.60
108
0.44
38
0.33
25
0.53
59
0.39
35
12.01
114
0.64
55
NVStereoNet_ROBtwo views4.91
95
1.40
65
16.46
65
1.47
64
1.72
84
3.46
22
1.96
66
3.89
32
3.61
89
2.90
57
24.61
115
4.81
86
4.49
90
13.57
100
4.87
72
1.33
107
1.29
107
1.70
109
1.68
107
1.41
105
1.66
103
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
96
0.90
19
21.33
102
1.50
66
1.33
56
4.09
71
3.49
92
4.03
50
4.59
97
6.57
99
10.95
102
12.92
105
8.41
103
9.86
58
8.78
89
0.75
83
0.58
77
0.51
52
0.50
59
0.68
62
0.57
44
MeshStereopermissivetwo views5.35
97
1.32
58
21.34
103
1.34
48
1.15
42
4.40
84
3.90
98
5.18
102
4.55
96
10.14
107
11.82
104
11.47
103
5.84
97
13.84
103
7.03
85
0.61
67
0.58
77
0.60
72
0.58
75
0.72
70
0.59
48
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
98
1.22
52
17.40
73
1.60
75
2.41
103
3.48
26
2.12
72
4.52
94
3.26
83
3.67
78
13.15
105
22.10
110
3.46
82
11.92
92
17.16
102
0.34
14
0.21
6
0.23
1
0.19
3
0.36
10
0.28
6
MDST_ROBtwo views5.53
99
0.59
6
23.11
107
2.55
104
2.69
104
21.20
111
1.98
68
4.77
100
2.65
52
9.36
104
4.25
88
4.14
72
4.45
89
22.57
110
3.68
63
0.48
47
0.33
25
0.47
41
0.44
50
0.44
17
0.44
22
MSMD_ROBtwo views5.61
100
1.08
38
16.68
68
1.21
27
0.83
17
10.14
106
2.07
71
3.93
40
6.22
103
10.37
108
8.03
96
22.22
111
9.45
105
11.66
85
4.52
68
0.66
75
0.58
77
0.66
80
0.68
83
0.65
56
0.63
54
SANettwo views5.65
101
1.72
77
20.94
97
1.26
33
0.88
20
7.20
103
3.81
97
4.60
96
11.23
106
4.56
89
8.78
98
7.49
98
6.95
100
11.85
89
18.11
103
0.49
48
0.44
56
0.50
50
0.42
48
0.90
82
0.82
78
ELAS_RVCcopylefttwo views5.72
102
1.64
76
20.40
94
2.09
100
2.01
97
4.95
92
11.47
107
4.66
98
6.31
104
8.09
102
8.01
95
8.06
102
11.65
107
8.36
42
10.22
93
0.99
98
1.17
102
0.81
97
0.76
89
1.14
94
1.58
102
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
PWCKtwo views5.85
103
9.95
111
15.22
58
2.90
105
1.56
72
5.66
98
7.69
103
5.27
104
4.74
98
3.88
79
7.51
92
8.04
100
5.28
94
15.18
105
8.40
88
3.08
111
1.51
108
2.24
110
1.37
106
5.55
110
1.96
108
ELAScopylefttwo views5.99
104
1.62
73
20.29
93
1.96
95
2.01
97
15.09
110
8.38
105
5.23
103
5.54
101
9.02
103
8.61
97
6.74
95
9.75
106
7.48
38
11.40
95
0.96
96
1.18
103
0.72
89
0.93
99
1.30
102
1.67
104
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
FC-DCNNcopylefttwo views6.64
105
0.77
14
19.26
86
1.68
80
1.64
78
8.27
105
2.35
76
6.35
108
9.23
105
10.85
109
11.27
103
19.89
108
13.33
108
12.51
97
11.79
96
0.66
75
0.57
75
0.58
67
0.56
71
0.63
51
0.61
50
MADNet++two views7.57
106
5.80
104
13.28
40
6.38
110
5.26
110
7.18
102
7.77
104
5.32
105
5.62
102
6.11
97
7.78
93
6.02
94
8.01
102
11.74
87
16.36
99
9.36
114
4.95
112
8.57
114
5.66
112
4.83
109
5.34
111
SGM+DAISYtwo views8.36
107
3.82
99
24.83
110
2.02
98
1.75
86
14.80
109
17.22
109
4.27
77
11.79
108
13.88
111
14.38
108
11.74
104
13.83
109
10.56
68
15.12
98
1.21
101
1.19
105
1.14
106
1.13
103
1.17
97
1.37
99
MANEtwo views11.39
108
1.38
62
23.16
108
4.82
109
2.79
105
22.02
112
14.51
108
25.81
113
26.28
114
10.93
110
13.17
106
19.90
109
18.03
110
15.69
107
20.41
109
0.96
96
0.74
94
1.42
108
3.24
111
0.74
71
1.71
105
SGM-ForestMtwo views11.55
109
1.16
46
21.62
105
2.14
101
0.81
14
22.25
113
19.08
110
16.50
112
11.34
107
15.94
114
14.44
109
27.12
114
22.61
115
24.76
115
28.39
114
0.50
51
0.48
67
0.47
41
0.44
50
0.55
38
0.53
35
LSMtwo views12.71
110
3.52
98
20.79
95
55.63
118
84.76
118
3.63
37
6.30
101
3.73
21
4.93
99
5.55
93
13.92
107
3.53
58
3.25
77
9.72
55
4.12
65
0.64
72
0.85
98
0.71
86
0.81
94
1.03
89
26.84
116
DGTPSM_ROBtwo views15.31
111
8.85
108
28.14
113
9.90
112
18.89
111
12.44
107
33.21
113
13.48
110
17.86
112
9.50
105
22.89
113
13.78
106
19.67
111
21.26
108
23.68
110
5.08
112
8.84
113
5.62
112
11.50
113
6.72
111
14.87
112
DPSMNet_ROBtwo views15.52
112
8.89
109
31.44
114
10.07
113
18.91
112
12.44
107
33.22
114
13.51
111
17.94
113
9.58
106
22.91
114
13.79
107
19.67
111
21.43
109
23.73
111
5.17
113
8.84
113
5.63
113
11.52
116
6.79
112
14.90
113
LE_ROBtwo views18.58
113
1.21
49
48.38
116
8.60
111
23.00
115
7.32
104
19.74
111
9.94
109
38.69
116
43.53
116
21.91
112
28.40
115
40.92
117
23.18
111
54.71
117
0.34
14
0.33
25
0.32
11
0.36
26
0.42
15
0.40
18
DPSMtwo views20.78
114
20.03
114
26.57
111
23.62
114
22.46
113
31.10
114
39.97
116
33.52
115
17.41
110
15.25
112
15.80
110
24.06
112
20.71
113
23.30
112
27.62
112
10.78
115
9.41
115
11.69
115
11.50
113
14.49
115
16.36
114
DPSM_ROBtwo views20.78
114
20.03
114
26.57
111
23.62
114
22.46
113
31.10
114
39.97
116
33.52
115
17.41
110
15.25
112
15.80
110
24.06
112
20.71
113
23.30
112
27.62
112
10.78
115
9.41
115
11.69
115
11.50
113
14.49
115
16.36
114
AVERAGE_ROBtwo views44.46
116
45.91
117
46.59
115
40.59
117
38.66
116
32.99
116
30.52
112
46.19
117
43.26
117
49.59
117
48.97
117
44.62
116
39.89
116
45.04
116
43.94
115
48.74
118
50.69
118
50.80
118
49.79
118
45.54
118
46.97
117
MEDIAN_ROBtwo views47.37
117
49.08
118
49.92
117
40.32
116
39.97
117
34.95
117
33.43
115
49.17
118
46.39
118
53.06
118
52.65
118
47.65
117
42.23
119
48.35
118
47.29
116
52.10
119
54.04
119
54.28
119
53.48
119
48.76
119
50.33
118
LSM0two views49.58
118
40.04
116
53.62
118
97.15
119
128.68
120
62.75
118
79.89
119
67.63
119
35.11
115
33.02
115
32.61
116
48.43
118
41.60
118
46.60
117
55.43
118
21.63
117
18.99
117
23.62
117
23.13
117
29.47
117
52.26
119
BEATNet-Init1two views49.72
119
64.32
119
97.77
119
4.68
108
4.28
109
84.75
119
60.42
118
73.38
120
68.59
119
88.29
119
88.51
120
91.94
119
97.99
120
85.26
120
70.96
119
0.80
87
0.65
87
0.85
98
0.70
86
7.96
113
2.33
109
DPSimNet_ROBtwo views102.30
120
119.72
120
143.59
120
109.42
120
94.10
119
127.22
120
127.15
120
26.62
114
102.36
120
149.58
120
83.99
119
139.34
120
137.54
121
64.62
119
82.36
120
105.18
120
66.15
120
120.15
120
69.61
120
99.58
120
77.72
120
MSMDNettwo views1.11
6