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 views0.36
1
0.21
1
1.70
9
0.29
6
0.18
6
0.55
1
0.26
10
0.78
11
0.36
1
0.57
19
0.67
38
0.33
3
0.24
1
0.32
1
0.23
1
0.10
3
0.08
5
0.09
3
0.07
4
0.09
1
0.12
7
R-Stereo Traintwo views0.36
1
0.21
1
1.70
9
0.29
6
0.18
6
0.55
1
0.26
10
0.78
11
0.36
1
0.57
19
0.67
38
0.33
3
0.24
1
0.32
1
0.23
1
0.10
3
0.08
5
0.09
3
0.07
4
0.09
1
0.12
7
DPM-Stereotwo views0.37
3
0.34
5
1.50
2
0.27
3
0.18
6
0.63
24
0.17
1
0.92
48
0.53
4
0.49
11
0.48
2
0.30
2
0.30
4
0.39
3
0.30
4
0.10
3
0.08
5
0.10
9
0.08
8
0.15
7
0.12
7
PMTNettwo views0.40
4
0.29
3
2.32
49
0.22
1
0.14
1
0.68
40
0.27
14
0.59
1
0.70
25
0.36
1
0.36
1
0.29
1
0.24
1
0.63
7
0.28
3
0.11
7
0.07
3
0.08
1
0.06
1
0.17
11
0.08
1
AdaStereotwo views0.44
5
0.38
8
1.79
16
0.34
24
0.20
10
0.57
8
0.28
16
0.92
48
0.69
22
0.52
14
0.55
13
0.61
16
0.33
6
0.48
4
0.39
7
0.17
34
0.08
5
0.17
34
0.11
17
0.18
12
0.12
7
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.
HITNettwo views0.46
6
0.92
52
2.09
33
0.33
15
0.14
1
0.55
1
0.22
3
0.83
23
0.56
7
0.51
12
0.49
4
0.48
7
0.37
9
0.48
4
0.54
29
0.09
1
0.06
1
0.10
9
0.08
8
0.23
28
0.09
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
BEATNet_4xtwo views0.47
7
0.73
30
1.91
22
0.33
15
0.15
4
0.56
4
0.25
9
0.85
28
0.58
8
0.54
17
0.51
8
0.57
9
0.42
19
0.57
6
0.61
36
0.11
7
0.08
5
0.12
16
0.09
12
0.20
18
0.16
19
ccstwo views0.47
7
0.29
3
1.64
6
0.28
4
0.20
10
0.56
4
0.41
43
0.96
65
0.81
57
0.45
4
0.49
4
0.60
14
0.47
31
0.72
10
0.39
7
0.17
34
0.14
36
0.20
51
0.17
42
0.23
28
0.21
32
MLCVtwo views0.48
9
0.89
51
1.85
18
0.29
6
0.14
1
0.59
13
0.23
7
0.72
4
0.59
9
0.66
42
0.66
37
0.55
8
0.36
7
1.03
21
0.49
23
0.09
1
0.06
1
0.09
3
0.07
4
0.19
16
0.11
3
DN-CSS_ROBtwo views0.48
9
1.23
94
1.99
28
0.39
48
0.22
16
0.56
4
0.19
2
0.75
6
0.61
11
0.45
4
0.62
28
0.41
5
0.31
5
0.70
8
0.32
5
0.11
7
0.08
5
0.14
17
0.13
19
0.36
67
0.12
7
DeepPruner_ROBtwo views0.50
11
0.70
27
1.94
24
0.29
6
0.26
27
0.63
24
0.31
29
0.80
14
0.48
3
0.61
29
0.55
13
0.66
26
0.38
10
0.98
17
0.40
9
0.19
47
0.14
36
0.14
17
0.14
23
0.23
28
0.17
24
StereoDRNet-Refinedtwo views0.50
11
0.41
10
1.50
2
0.33
15
0.21
14
0.68
40
0.26
10
0.81
16
0.68
21
0.67
45
0.68
44
0.73
36
0.42
19
1.10
25
0.59
34
0.11
7
0.08
5
0.14
17
0.14
23
0.25
39
0.21
32
Rohan Chabra, Julian Straub, Chris Sweeney, Richard Newcombe, Henry Fuchs: StereoDRNet. CVPR
iResNet_ROBtwo views0.50
11
1.15
88
1.73
12
0.31
12
0.21
14
0.58
9
0.22
3
0.95
62
0.82
58
0.58
23
0.62
28
0.57
9
0.39
14
0.84
14
0.40
9
0.11
7
0.08
5
0.09
3
0.06
1
0.21
21
0.13
12
CFNet_RVCtwo views0.51
14
0.76
32
1.79
16
0.28
4
0.30
39
0.63
24
0.28
16
0.66
2
0.55
6
0.45
4
0.57
18
0.74
38
0.41
16
1.46
38
0.42
12
0.14
23
0.10
20
0.28
82
0.13
19
0.18
12
0.14
15
STTStereotwo views0.51
14
0.69
26
2.14
39
0.36
30
0.30
39
0.61
19
0.32
32
0.77
10
0.60
10
0.60
27
0.54
11
0.67
28
0.43
23
0.86
15
0.42
12
0.16
30
0.15
44
0.17
34
0.20
61
0.16
9
0.15
16
iResNettwo views0.51
14
0.76
32
2.59
73
0.41
57
0.17
5
0.64
29
0.27
14
0.82
19
0.70
25
0.57
19
0.64
35
0.43
6
0.36
7
0.70
8
0.52
27
0.10
3
0.07
3
0.09
3
0.08
8
0.21
21
0.13
12
DLCB_ROBtwo views0.54
17
0.55
16
1.49
1
0.36
30
0.25
21
0.65
32
0.36
36
0.80
14
0.70
25
0.72
53
0.69
48
0.83
48
0.46
27
1.53
46
0.60
35
0.13
18
0.12
30
0.14
17
0.14
23
0.14
4
0.13
12
CC-Net-ROBtwo views0.55
18
0.93
57
2.43
61
0.37
35
0.36
54
0.60
15
0.29
22
0.81
16
0.61
11
0.63
35
0.56
16
0.72
33
0.46
27
1.03
21
0.44
17
0.15
25
0.11
22
0.14
17
0.14
23
0.16
9
0.16
19
NLCA_NET_v2_RVCtwo views0.55
18
0.92
52
2.36
54
0.38
44
0.37
59
0.60
15
0.28
16
0.82
19
0.62
14
0.62
32
0.55
13
0.72
33
0.45
25
1.00
19
0.43
16
0.15
25
0.11
22
0.14
17
0.14
23
0.15
7
0.16
19
Zhibo Rao, Mingyi He, Yuchao Dai, Zhidong Zhu, Bo Li, and Renjie He.: NLCA-Net: A non-local context attention network for stereo matching.
DISCOtwo views0.57
20
0.34
5
1.87
20
0.45
69
0.34
48
0.84
66
0.35
35
0.84
25
0.78
49
0.59
25
0.63
31
1.06
80
0.53
38
1.10
25
0.82
56
0.12
14
0.09
15
0.11
15
0.10
15
0.22
24
0.18
27
ccs_robtwo views0.57
20
0.99
70
1.97
27
0.32
13
0.20
10
0.63
24
0.22
3
0.82
19
0.72
33
0.53
16
0.61
25
0.61
16
0.38
10
1.98
70
0.44
17
0.12
14
0.09
15
0.14
17
0.14
23
0.28
50
0.16
19
CFNettwo views0.58
22
0.96
64
1.78
15
0.33
15
0.25
21
0.63
24
0.22
3
0.88
36
0.75
42
0.46
8
0.59
21
0.58
12
0.40
15
2.29
79
0.48
21
0.12
14
0.09
15
0.15
25
0.14
23
0.30
54
0.15
16
NVstereo2Dtwo views0.58
22
0.40
9
2.03
30
0.35
27
0.35
50
0.69
43
0.39
39
0.91
43
0.80
53
0.47
9
0.53
9
0.67
28
0.46
27
1.99
71
0.48
21
0.19
47
0.11
22
0.17
34
0.13
19
0.25
39
0.25
51
iResNetv2_ROBtwo views0.59
24
1.43
103
2.96
86
0.40
54
0.22
16
0.58
9
0.28
16
0.91
43
0.73
38
0.57
19
0.63
31
0.60
14
0.41
16
0.74
11
0.35
6
0.11
7
0.08
5
0.10
9
0.07
4
0.43
83
0.17
24
LALA_ROBtwo views0.60
25
0.76
32
1.69
8
0.36
30
0.29
35
0.79
60
0.55
64
0.94
54
0.77
47
0.81
65
0.67
38
0.94
65
0.49
35
0.99
18
0.73
50
0.21
55
0.15
44
0.20
51
0.18
48
0.23
28
0.21
32
ETE_ROBtwo views0.60
25
0.71
28
1.55
4
0.37
35
0.25
21
0.69
43
0.51
59
1.47
104
0.70
25
0.80
61
0.69
48
0.85
51
0.49
35
1.36
35
0.58
33
0.16
30
0.13
34
0.19
45
0.16
36
0.21
21
0.23
46
XPNet_ROBtwo views0.61
27
0.72
29
1.66
7
0.37
35
0.27
32
0.67
37
0.47
55
0.89
39
0.71
31
0.78
59
0.70
51
0.81
47
0.53
38
1.70
54
0.76
52
0.21
55
0.16
47
0.16
28
0.15
33
0.20
18
0.21
32
FADNet-RVCtwo views0.61
27
1.26
95
2.12
37
0.33
15
0.26
27
0.59
13
0.29
22
0.86
29
0.62
14
0.45
4
0.50
6
0.62
20
0.46
27
1.73
58
0.55
30
0.17
34
0.20
65
0.17
34
0.18
48
0.59
97
0.28
60
RASNettwo views0.61
27
0.68
24
2.03
30
0.35
27
0.46
81
0.65
32
0.28
16
0.78
11
0.70
25
0.48
10
0.64
35
0.93
63
0.61
50
2.21
76
0.73
50
0.12
14
0.09
15
0.09
3
0.10
15
0.14
4
0.11
3
HSMtwo views0.62
30
0.76
32
1.72
11
0.30
10
0.29
35
0.64
29
0.28
16
0.94
54
0.62
14
0.58
23
0.57
18
1.40
95
0.64
57
2.59
90
0.42
12
0.11
7
0.08
5
0.10
9
0.08
8
0.14
4
0.11
3
DRN-Testtwo views0.64
31
0.43
14
2.44
63
0.39
48
0.44
73
0.80
61
0.41
43
1.02
82
0.78
49
0.80
61
0.63
31
0.69
31
0.56
42
1.79
63
0.69
43
0.15
25
0.12
30
0.17
34
0.16
36
0.18
12
0.17
24
StereoDRNettwo views0.64
31
0.88
50
2.35
53
0.41
57
0.45
78
0.73
51
0.47
55
0.94
54
0.74
40
0.75
57
0.54
11
0.74
38
0.47
31
1.70
54
0.72
47
0.15
25
0.14
36
0.15
25
0.14
23
0.20
18
0.19
30
FADNettwo views0.64
31
1.22
93
2.06
32
0.32
13
0.30
39
0.66
35
0.29
22
0.88
36
0.72
33
0.44
2
0.48
2
0.66
26
0.57
44
1.84
67
0.45
20
0.46
100
0.20
65
0.17
34
0.19
53
0.59
97
0.24
48
NOSS_ROBtwo views0.64
31
0.65
22
1.62
5
0.46
72
0.25
21
0.67
37
0.30
26
0.93
51
0.77
47
0.59
25
0.57
18
0.77
41
0.38
10
2.78
97
0.44
17
0.23
70
0.24
79
0.28
82
0.26
81
0.27
47
0.39
85
NCCL2two views0.65
35
0.94
58
1.96
26
0.43
64
0.30
39
0.71
46
0.68
77
0.84
25
0.65
18
0.70
49
0.60
23
0.86
53
0.54
40
1.56
48
0.57
32
0.20
53
0.16
47
0.67
107
0.25
76
0.22
24
0.24
48
CVANet_RVCtwo views0.65
35
0.80
42
2.13
38
0.36
30
0.26
27
0.62
20
0.42
46
0.84
25
0.72
33
0.69
47
0.68
44
0.68
30
0.52
37
2.64
91
0.42
12
0.19
47
0.12
30
0.22
60
0.17
42
0.37
72
0.15
16
PSMNet_ROBtwo views0.66
37
1.01
75
2.33
51
0.38
44
0.31
44
0.72
49
0.55
64
0.86
29
0.69
22
0.54
17
0.60
23
0.75
40
0.47
31
2.00
72
0.82
56
0.17
34
0.13
34
0.19
45
0.19
53
0.24
35
0.27
57
RPtwo views0.67
38
0.82
44
1.94
24
0.43
64
0.44
73
0.65
32
0.45
49
0.89
39
0.82
58
0.66
42
0.92
67
0.99
71
0.76
68
1.55
47
0.66
40
0.25
77
0.18
55
0.23
65
0.20
61
0.31
59
0.22
39
FADNet-RVC-Resampletwo views0.67
38
1.16
91
3.16
94
0.37
35
0.29
35
0.62
20
0.38
37
0.86
29
0.63
17
0.51
12
0.50
6
0.57
9
0.45
25
1.57
49
0.49
23
0.24
74
0.32
93
0.23
65
0.19
53
0.27
47
0.62
97
TDLMtwo views0.67
38
0.92
52
1.76
13
0.37
35
0.24
19
0.66
35
0.56
66
0.82
19
0.74
40
1.16
86
0.62
28
0.61
16
0.48
34
2.70
94
0.50
26
0.20
53
0.11
22
0.35
92
0.14
23
0.32
62
0.16
19
HSM-Net_RVCpermissivetwo views0.67
38
0.34
5
2.23
46
0.26
2
0.19
9
0.67
37
0.29
22
1.26
98
0.69
22
0.72
53
0.70
51
0.78
44
0.42
19
3.77
106
0.41
11
0.13
18
0.09
15
0.10
9
0.09
12
0.12
3
0.11
3
Gengshan Yang, Joshua Manela, Michael Happold, and Deva Ramanan: Hierarchical Deep Stereo Matching on High-resolution Images. CVPR 2019
AANet_RVCtwo views0.68
42
1.00
72
2.09
33
0.33
15
0.25
21
0.58
9
0.32
32
0.81
16
0.80
53
0.61
29
0.94
68
0.80
46
0.41
16
2.46
87
1.02
66
0.36
96
0.14
36
0.10
9
0.09
12
0.23
28
0.22
39
Nwc_Nettwo views0.68
42
0.83
45
2.53
69
0.41
57
0.39
61
0.77
56
0.38
37
1.01
80
0.70
25
0.60
27
0.99
73
0.98
69
0.68
58
1.41
36
0.63
39
0.23
70
0.15
44
0.23
65
0.25
76
0.23
28
0.18
27
stereogantwo views0.69
44
0.51
15
2.44
63
0.39
48
0.39
61
1.01
77
0.41
43
0.94
54
0.92
72
0.73
56
0.85
63
1.15
88
0.69
59
1.57
49
0.53
28
0.19
47
0.18
55
0.20
51
0.16
36
0.36
67
0.24
48
AF-Nettwo views0.69
44
0.92
52
2.21
42
0.41
57
0.40
67
0.68
40
0.40
41
1.02
82
0.86
64
0.69
47
1.03
77
1.04
77
0.84
70
1.19
29
0.78
54
0.22
67
0.14
36
0.25
77
0.17
42
0.30
54
0.18
27
Anonymous Stereotwo views0.70
46
1.08
80
2.47
66
0.42
62
0.36
54
0.60
15
0.78
85
0.83
23
0.75
42
0.65
41
0.63
31
0.62
20
0.38
10
2.43
85
0.78
54
0.19
47
0.18
55
0.19
45
0.20
61
0.30
54
0.23
46
RYNettwo views0.70
46
0.56
17
2.25
47
0.35
27
0.41
71
0.86
68
0.43
47
0.94
54
0.82
58
0.62
32
0.53
9
0.91
60
0.60
48
2.70
94
1.00
63
0.14
23
0.11
22
0.16
28
0.14
23
0.25
39
0.27
57
PWC_ROBbinarytwo views0.70
46
0.94
58
1.88
21
0.39
48
0.39
61
0.62
20
0.30
26
0.98
71
0.94
73
0.91
72
1.29
86
0.77
41
0.62
55
1.62
52
1.03
68
0.18
42
0.11
22
0.18
42
0.13
19
0.51
94
0.22
39
CBMVpermissivetwo views0.71
49
0.76
32
2.64
75
0.33
15
0.20
10
0.78
57
0.79
86
0.98
71
0.78
49
0.85
69
0.71
53
0.86
53
1.10
86
1.46
38
0.62
37
0.18
42
0.20
65
0.21
57
0.19
53
0.25
39
0.33
76
Konstantinos Batsos, Changjiang Cai, Philippos Mordohai: CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation. Computer Vision and Pattern Recognition (CVPR) 2018
RGCtwo views0.71
49
0.97
67
2.18
41
0.44
66
0.44
73
0.73
51
0.40
41
0.94
54
0.71
31
0.71
51
1.04
78
1.01
73
0.71
61
1.74
60
0.70
44
0.23
70
0.18
55
0.28
82
0.26
81
0.29
52
0.27
57
RTSCtwo views0.72
51
1.00
72
2.22
44
0.40
54
0.29
35
0.82
64
0.31
29
0.96
65
1.10
87
0.92
73
0.81
57
0.78
44
0.58
45
1.23
31
1.55
79
0.18
42
0.17
53
0.14
17
0.17
42
0.43
83
0.30
64
SHDtwo views0.72
51
0.77
37
2.36
54
0.45
69
0.40
67
0.71
46
0.32
32
1.02
82
1.25
92
1.07
79
0.84
60
0.90
59
0.73
64
1.09
24
0.98
61
0.21
55
0.18
55
0.20
51
0.21
66
0.32
62
0.31
67
PWCDC_ROBbinarytwo views0.72
51
1.06
78
2.09
33
0.47
74
0.36
54
0.78
57
0.26
10
0.98
71
1.34
95
0.61
29
2.07
95
0.64
23
0.61
50
0.83
12
0.71
45
0.31
91
0.14
36
0.15
25
0.12
18
0.60
100
0.21
32
PASMtwo views0.72
51
0.68
24
1.91
22
0.47
74
0.44
73
0.60
15
0.51
59
0.76
7
0.80
53
0.77
58
0.67
38
0.87
55
0.61
50
2.40
83
0.67
41
0.32
92
0.35
94
0.36
94
0.41
98
0.45
87
0.39
85
XQCtwo views0.73
55
0.98
69
2.39
57
0.46
72
0.35
50
0.81
62
0.46
50
1.00
78
1.00
78
0.80
61
0.69
48
0.88
57
0.74
67
1.32
33
1.20
72
0.24
74
0.16
47
0.23
65
0.20
61
0.41
82
0.33
76
GANetREF_RVCpermissivetwo views0.75
56
1.15
88
2.46
65
0.47
74
0.26
27
0.83
65
0.61
71
0.97
70
0.89
67
0.62
32
0.61
25
0.72
33
0.59
46
2.42
84
0.55
30
0.32
92
0.25
85
0.36
94
0.24
70
0.39
79
0.31
67
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
GANettwo views0.75
56
0.84
47
2.01
29
0.40
54
0.25
21
0.74
53
0.66
75
0.87
33
0.73
38
0.71
51
1.07
80
0.84
49
0.59
46
2.71
96
1.15
70
0.17
34
0.16
47
0.44
101
0.16
36
0.31
59
0.20
31
PDISCO_ROBtwo views0.75
56
1.06
78
2.11
36
0.64
97
0.59
96
1.13
84
0.43
47
1.14
92
1.05
82
0.64
38
0.71
53
0.96
67
0.61
50
1.21
30
0.76
52
0.37
97
0.18
55
0.27
80
0.24
70
0.56
96
0.32
71
CBMV_ROBtwo views0.76
59
0.42
11
2.68
77
0.33
15
0.22
16
0.74
53
0.47
55
0.96
65
0.76
45
1.04
78
0.96
71
0.88
57
0.93
76
2.06
73
1.19
71
0.22
67
0.24
79
0.27
80
0.25
76
0.25
39
0.28
60
NCC-stereotwo views0.79
60
0.94
58
2.42
58
0.59
94
0.45
78
1.90
100
0.46
50
0.99
76
0.72
33
0.64
38
0.84
60
0.91
60
0.73
64
1.80
64
1.00
63
0.27
86
0.19
61
0.23
65
0.31
91
0.24
35
0.22
39
Abc-Nettwo views0.79
60
0.94
58
2.42
58
0.59
94
0.45
78
1.90
100
0.46
50
0.99
76
0.72
33
0.64
38
0.84
60
0.91
60
0.73
64
1.80
64
1.00
63
0.27
86
0.19
61
0.23
65
0.31
91
0.24
35
0.22
39
DMCA-RVCcopylefttwo views0.80
62
1.01
75
2.60
74
0.33
15
0.79
103
0.58
9
0.30
26
0.74
5
0.53
4
0.63
35
0.67
38
0.73
36
3.68
110
1.50
42
0.62
37
0.19
47
0.17
53
0.21
57
0.20
61
0.24
35
0.21
32
ADCReftwo views0.80
62
0.96
64
2.29
48
0.37
35
0.35
50
0.96
72
0.51
59
0.87
33
0.75
42
0.88
70
0.83
59
0.61
16
1.36
94
1.02
20
3.27
102
0.13
18
0.11
22
0.17
34
0.19
53
0.25
39
0.21
32
DANettwo views0.80
62
0.65
22
2.21
42
1.19
109
0.48
83
0.74
53
1.25
100
0.89
39
1.53
100
0.89
71
0.71
53
0.96
67
0.56
42
1.83
66
0.96
60
0.18
42
0.16
47
0.19
45
0.16
36
0.26
45
0.29
62
FADNet_RVCtwo views0.81
65
1.14
85
2.42
58
0.30
10
0.24
19
0.71
46
0.24
8
0.95
62
0.61
11
0.44
2
0.56
16
0.59
13
0.42
19
1.52
45
0.68
42
0.94
109
0.96
111
0.28
82
1.00
111
0.90
107
1.36
109
edge stereotwo views0.81
65
0.73
30
3.30
98
0.37
35
1.21
107
0.62
20
0.39
39
0.93
51
0.79
52
0.80
61
0.88
66
1.14
87
0.71
61
2.18
75
0.71
45
0.21
55
0.20
65
0.26
79
0.18
48
0.38
76
0.26
55
FBW_ROBtwo views0.83
67
0.77
37
1.77
14
0.44
66
0.30
39
0.87
70
0.46
50
1.10
90
0.91
69
0.81
65
0.75
56
1.01
73
0.61
50
3.66
105
1.23
73
0.26
81
0.23
76
0.36
94
0.25
76
0.27
47
0.62
97
PA-Nettwo views0.84
68
1.00
72
3.49
100
0.63
96
0.40
67
0.70
45
0.52
62
0.86
29
0.84
62
0.52
14
0.59
21
0.70
32
0.54
40
2.35
80
2.23
93
0.17
34
0.27
88
0.19
45
0.28
87
0.22
24
0.36
81
Zhibo Rao, Mingyi He, Yuchao Dai, Zhelun Shen: Patch Attention Network with Generative Adversarial Model for Semi-Supervised Binocular Disparity Prediction.
DeepPrunerFtwo views0.86
69
0.86
49
2.91
85
0.45
69
0.50
84
0.64
29
0.48
58
1.14
92
2.64
109
0.63
35
0.61
25
0.65
25
0.62
55
2.21
76
1.31
75
0.23
70
0.19
61
0.30
88
0.28
87
0.36
67
0.25
51
DPSNettwo views0.86
69
0.60
19
2.34
52
0.58
91
0.36
54
1.07
81
0.75
83
1.30
99
1.05
82
0.70
49
0.68
44
0.94
65
1.29
91
2.08
74
1.82
85
0.26
81
0.23
76
0.16
28
0.17
42
0.48
88
0.40
88
PVDtwo views0.90
71
0.85
48
2.15
40
0.53
82
0.50
84
0.99
74
0.56
66
1.15
94
1.64
102
1.41
98
1.41
90
1.11
85
1.13
88
1.35
34
1.31
75
0.26
81
0.28
89
0.28
82
0.26
81
0.37
72
0.49
93
SAMSARAtwo views0.91
72
0.97
67
2.22
44
0.80
102
0.58
95
1.32
88
1.31
101
1.01
80
0.94
73
1.36
94
0.87
64
1.38
94
0.90
74
1.72
57
1.02
66
0.24
74
0.30
90
0.22
60
0.25
76
0.36
67
0.35
79
MFMNet_retwo views0.92
73
1.28
96
2.53
69
0.82
104
0.76
102
0.90
71
0.75
83
0.98
71
1.04
81
1.03
76
1.05
79
0.84
49
0.94
78
0.83
12
0.72
47
0.68
106
0.60
103
0.63
105
0.60
103
0.77
105
0.74
101
ADCP+two views0.94
74
1.12
82
3.14
93
0.34
24
0.39
61
1.06
79
0.79
86
0.89
39
0.80
53
0.66
42
0.67
38
0.85
51
1.55
100
0.93
16
4.55
109
0.13
18
0.11
22
0.16
28
0.15
33
0.26
45
0.25
51
ADCMidtwo views0.96
75
1.32
99
2.74
78
0.38
44
0.35
50
0.99
74
0.58
69
0.94
54
0.86
64
1.23
91
1.01
74
1.01
73
1.00
81
1.05
23
3.97
106
0.21
55
0.19
61
0.42
99
0.32
93
0.40
81
0.29
62
G-Nettwo views0.99
76
0.64
20
2.76
79
0.57
89
0.67
98
2.03
104
0.72
81
0.76
7
0.67
19
1.14
84
2.56
101
1.30
92
1.02
82
1.48
40
1.74
82
0.25
77
0.24
79
0.24
72
0.26
81
0.37
72
0.31
67
ADCLtwo views0.99
76
0.79
40
2.43
61
0.39
48
0.32
45
1.50
92
1.90
106
0.92
48
0.97
77
1.26
93
1.27
85
1.00
72
1.42
97
1.14
27
3.31
103
0.17
34
0.14
36
0.22
60
0.21
66
0.30
54
0.22
39
STTStereo_v2two views0.99
76
0.64
20
2.76
79
0.57
89
0.67
98
2.03
104
0.72
81
0.76
7
0.67
19
1.14
84
2.56
101
1.30
92
1.02
82
1.48
40
1.74
82
0.25
77
0.24
79
0.24
72
0.26
81
0.37
72
0.31
67
ADCPNettwo views0.99
76
0.99
70
3.93
105
0.37
35
0.68
100
1.36
89
0.62
72
0.87
33
0.76
45
0.78
59
0.81
57
1.08
81
1.18
89
1.14
27
2.49
97
0.21
55
0.69
106
0.21
57
0.95
109
0.29
52
0.37
83
AnyNet_C32two views1.00
80
1.33
100
2.53
69
0.49
78
0.54
91
1.15
86
1.14
98
0.91
43
0.89
67
1.11
81
0.95
70
0.87
55
0.93
76
1.50
42
4.07
107
0.21
55
0.20
65
0.25
77
0.24
70
0.38
76
0.33
76
MFN_U_SF_DS_RVCtwo views1.01
81
1.67
106
2.32
49
0.47
74
0.39
61
1.47
90
0.95
93
0.95
62
0.91
69
0.84
67
0.94
68
1.05
79
0.60
48
2.95
101
0.72
47
0.41
98
0.72
109
0.32
90
0.55
100
0.74
102
1.22
108
SGM-Foresttwo views1.04
82
0.42
11
3.59
102
0.41
57
0.32
45
1.05
78
1.32
102
0.94
54
0.83
61
1.16
86
2.23
98
1.03
76
1.39
95
2.49
88
2.35
94
0.21
55
0.21
70
0.20
51
0.19
53
0.23
28
0.32
71
Johannes L. Schönberger, Sudipta Sinha, Marc Pollefeys: Learning to Fuse Proposals from Multiple Scanline Optimizations in Semi-Global Matching. ECCV 2018
ADCStwo views1.05
83
1.36
102
2.98
90
0.42
62
0.36
54
0.96
72
0.81
90
1.06
89
1.12
88
1.25
92
0.97
72
1.08
81
0.90
74
1.25
32
4.86
111
0.21
55
0.21
70
0.22
60
0.23
69
0.36
67
0.38
84
SPS-STEREOcopylefttwo views1.07
84
1.13
84
2.88
83
0.65
98
0.69
101
0.99
74
0.68
77
1.05
88
0.87
66
1.36
94
1.23
84
1.46
97
0.94
78
1.87
68
1.72
81
0.66
105
0.61
104
0.58
104
0.57
101
0.65
101
0.77
104
K. Yamaguchi, D. McAllester, R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation. ECCV 2014
DMCAtwo views1.12
85
0.81
43
1.85
18
0.37
35
2.02
109
0.56
4
0.52
62
0.67
3
1.95
106
0.72
53
2.74
105
0.64
23
5.78
113
1.43
37
0.49
23
0.17
34
0.50
101
0.23
65
0.19
53
0.19
16
0.52
94
WCMA_ROBtwo views1.12
85
0.83
45
2.66
76
0.58
91
0.34
48
1.17
87
0.71
80
0.93
51
0.94
73
1.74
101
2.63
104
3.09
107
1.71
101
1.70
54
1.84
86
0.26
81
0.21
70
0.20
51
0.21
66
0.28
50
0.43
90
AnyNet_C01two views1.14
87
1.49
105
4.23
106
0.54
85
0.53
90
1.51
93
1.37
103
1.04
87
1.07
84
1.17
89
1.33
89
1.15
88
1.08
85
1.61
51
2.74
99
0.25
77
0.22
73
0.28
82
0.24
70
0.48
88
0.40
88
NaN_ROBtwo views1.14
87
0.94
58
2.96
86
0.53
82
0.56
94
1.13
84
0.99
94
0.98
71
1.00
78
1.09
80
0.87
64
0.77
41
1.06
84
5.90
111
2.72
98
0.16
30
0.23
76
0.16
28
0.24
70
0.22
24
0.36
81
SuperBtwo views1.15
89
1.14
85
5.43
109
0.36
30
0.26
27
0.81
62
0.89
92
0.88
36
1.17
89
0.67
45
0.68
44
0.63
22
1.39
95
1.74
60
3.37
105
0.15
25
0.12
30
0.17
34
0.16
36
2.80
113
0.25
51
NVStereoNet_ROBtwo views1.16
90
0.95
63
2.54
72
0.58
91
0.55
93
0.78
57
0.63
74
1.00
78
1.08
85
1.03
76
4.07
111
1.48
98
1.22
90
2.88
100
1.27
74
0.50
101
0.46
98
0.54
103
0.62
104
0.50
92
0.63
99
Nikolai Smolyanskiy, Alexey Kamenev, Stan Birchfield: On the Importance of Stereo for Accurate Depth Estimation: An Efficient Semi-Supervised Deep Neural Network Approach. Arxiv
MDST_ROBtwo views1.17
91
0.42
11
3.87
104
0.54
85
0.44
73
4.11
111
0.58
69
1.31
100
0.84
62
2.12
103
1.12
82
0.98
69
0.71
61
4.24
109
0.99
62
0.16
30
0.14
36
0.19
45
0.17
42
0.18
12
0.35
79
SANettwo views1.20
92
1.21
92
3.33
99
0.38
44
0.27
32
1.73
99
0.85
91
1.10
90
2.49
108
1.16
86
1.83
93
1.72
99
1.50
99
2.37
81
2.84
100
0.18
42
0.16
47
0.16
28
0.15
33
0.32
62
0.32
71
RTSAtwo views1.21
93
1.31
97
6.39
111
0.52
80
0.51
88
1.61
95
0.79
86
1.03
85
1.37
96
1.13
82
1.30
87
1.10
83
0.85
71
2.65
92
2.08
89
0.21
55
0.24
79
0.18
42
0.18
48
0.43
83
0.32
71
MeshStereopermissivetwo views1.21
93
1.15
88
3.01
91
0.34
24
0.28
34
1.10
82
0.69
79
1.35
101
1.01
80
2.71
106
2.41
100
2.96
106
1.45
98
2.80
98
1.40
78
0.22
67
0.22
73
0.22
60
0.19
53
0.50
92
0.26
55
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
RTStwo views1.21
93
1.31
97
6.39
111
0.52
80
0.51
88
1.61
95
0.79
86
1.03
85
1.37
96
1.13
82
1.30
87
1.10
83
0.85
71
2.65
92
2.08
89
0.21
55
0.24
79
0.18
42
0.18
48
0.43
83
0.32
71
CSANtwo views1.25
96
1.12
82
3.04
92
0.51
79
0.37
59
1.06
79
1.15
99
1.22
95
1.08
85
1.17
89
1.91
94
1.04
77
1.31
93
5.96
112
2.43
96
0.26
81
0.22
73
0.24
72
0.24
70
0.31
59
0.30
64
pmcnntwo views1.32
97
1.05
77
2.77
81
0.74
101
1.43
108
0.72
49
0.56
66
0.96
65
0.91
69
0.99
74
2.92
107
6.01
112
0.70
60
2.38
82
3.32
104
0.13
18
0.10
20
0.08
1
0.06
1
0.34
66
0.22
39
MSC_U_SF_DS_RVCtwo views1.34
98
1.70
107
2.50
67
0.56
88
0.40
67
2.15
106
0.99
94
1.24
97
1.26
93
1.02
75
1.10
81
1.29
91
0.88
73
3.10
102
1.08
69
1.66
111
0.83
110
0.68
108
0.88
108
1.57
110
1.80
111
MSMD_ROBtwo views1.34
98
0.96
64
2.38
56
0.39
48
0.32
45
1.96
102
0.62
72
0.91
43
1.43
98
2.94
109
2.60
103
4.77
109
1.95
103
2.43
85
1.33
77
0.29
89
0.26
86
0.32
90
0.32
93
0.30
54
0.30
64
MADNet+two views1.36
100
1.93
110
5.95
110
0.81
103
0.59
96
1.57
94
1.09
96
1.39
103
1.22
90
0.84
67
1.01
74
1.25
90
1.12
87
3.19
103
2.36
95
0.45
99
0.47
99
0.38
98
0.34
95
0.59
97
0.57
95
ELAS_RVCcopylefttwo views1.37
101
0.78
39
3.21
96
0.55
87
0.50
84
1.49
91
2.48
110
1.48
105
1.71
104
2.13
104
2.09
96
2.07
102
2.56
106
1.90
69
2.02
88
0.35
94
0.36
95
0.37
97
0.34
95
0.49
90
0.47
91
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
ELAScopylefttwo views1.39
102
0.79
40
2.86
82
0.53
82
0.50
84
2.81
109
2.20
107
1.61
108
1.50
99
2.28
105
2.35
99
1.82
100
2.13
105
1.73
58
2.22
92
0.35
94
0.37
97
0.35
92
0.35
97
0.49
90
0.47
91
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
FC-DCNNcopylefttwo views1.49
103
0.59
18
2.88
83
0.44
66
0.39
61
1.97
103
0.66
75
1.67
109
1.70
103
2.72
107
2.89
106
4.22
108
2.97
109
2.55
89
2.19
91
0.29
89
0.26
86
0.31
89
0.30
90
0.32
62
0.39
85
PWCKtwo views1.49
103
1.90
109
3.22
97
1.09
108
0.54
91
1.71
98
1.73
105
1.49
106
1.53
100
1.44
99
1.80
92
1.82
100
1.30
92
3.34
104
1.76
84
1.07
110
0.56
102
0.97
109
0.58
102
1.38
109
0.65
100
MFN_U_SF_RVCtwo views1.55
105
2.39
113
3.16
94
0.65
98
0.41
71
1.12
83
0.46
50
1.23
96
1.33
94
1.61
100
1.20
83
1.43
96
0.98
80
1.76
62
1.97
87
1.90
113
1.80
112
1.98
112
2.15
112
1.63
111
1.77
110
DispFullNettwo views1.67
106
2.10
111
5.22
107
1.53
110
8.90
113
0.86
68
0.31
29
0.96
65
0.96
76
1.37
97
1.01
74
1.13
86
1.98
104
1.51
44
0.91
59
0.54
103
0.31
91
1.18
110
0.63
106
1.13
108
0.80
105
SGM+DAISYtwo views1.83
107
1.43
103
3.81
103
0.87
105
1.01
105
2.38
107
2.53
111
1.36
102
1.75
105
2.92
108
3.22
109
2.65
104
2.93
108
2.22
78
3.10
101
0.72
107
0.69
106
0.63
105
0.62
104
0.77
105
0.98
106
SGM_RVCbinarytwo views1.89
108
0.92
52
2.50
67
0.96
106
0.93
104
1.63
97
1.37
103
1.52
107
12.43
113
1.85
102
1.69
91
2.40
103
1.74
102
2.81
99
1.71
80
0.53
102
0.47
99
0.42
99
0.47
99
0.74
102
0.74
101
Heiko Hirschmueller: Stereo processing by semiglobal matching and mutual information. TPAMI 2008, Volume 30(2), pp. 328-341
MADNet++two views2.79
109
2.11
112
2.97
89
2.49
111
2.28
110
2.95
110
2.31
108
2.67
110
2.47
107
3.03
110
3.13
108
2.70
105
2.92
107
4.10
107
4.48
108
2.68
114
2.24
114
3.04
114
2.44
113
2.37
112
2.34
112
MANEtwo views2.90
110
1.14
85
2.96
86
1.03
107
1.01
105
5.01
112
2.45
109
5.36
113
6.88
111
3.58
111
4.76
113
5.99
111
4.99
111
4.23
108
4.61
110
0.56
104
0.67
105
0.44
101
0.78
107
0.51
94
1.03
107
SGM-ForestMtwo views2.98
111
1.70
107
3.52
101
0.67
100
0.47
82
5.18
113
3.36
112
3.30
112
2.86
110
4.44
112
3.79
110
9.80
115
5.50
112
6.29
113
6.52
112
0.27
86
0.36
95
0.24
72
0.27
86
0.38
76
0.58
96
LSMtwo views3.31
112
1.11
81
5.29
108
9.40
116
30.65
121
0.84
66
1.12
97
0.91
43
1.22
90
1.36
94
2.19
97
0.93
63
0.83
69
1.66
53
0.90
58
0.21
55
0.31
91
0.24
72
0.28
87
0.39
79
6.33
114
LE_ROBtwo views4.43
113
1.35
101
9.38
113
2.79
112
3.93
112
2.43
108
3.46
113
2.68
111
11.19
112
11.66
117
4.73
112
5.26
110
10.46
114
4.50
110
9.49
113
0.82
108
0.69
106
1.34
111
0.98
110
0.75
104
0.76
103
DGTPSM_ROBtwo views8.89
114
5.62
114
12.25
114
5.92
114
11.32
114
8.54
114
24.18
118
6.40
114
13.49
114
5.15
113
11.46
116
6.19
113
10.92
115
7.00
114
13.72
114
4.47
115
7.65
115
3.92
115
7.45
117
4.38
115
7.87
115
DPSMNet_ROBtwo views9.48
115
5.64
115
13.01
117
8.82
115
11.82
115
8.54
114
24.20
119
6.41
115
13.65
115
5.32
114
11.47
117
6.33
114
10.92
115
7.69
115
13.89
115
7.25
118
8.19
116
5.06
116
7.78
118
5.31
116
8.26
116
BEATNet-Init1two views12.10
116
11.19
116
27.29
121
2.99
113
3.51
111
21.84
121
11.10
114
14.34
117
13.83
116
20.34
118
20.33
119
26.35
122
20.03
119
19.06
119
14.80
117
1.74
112
1.81
113
2.05
113
2.49
114
3.69
114
3.12
113
DPSMtwo views12.34
117
11.21
117
12.51
115
12.91
117
12.80
116
19.84
118
28.28
120
14.38
118
14.28
117
10.58
115
11.26
114
12.03
116
11.85
117
13.15
117
15.34
118
7.11
116
8.31
117
6.50
117
7.35
115
8.84
117
8.31
117
DPSM_ROBtwo views12.34
117
11.21
117
12.51
115
12.91
117
12.80
116
19.84
118
28.28
120
14.38
118
14.28
117
10.58
115
11.26
114
12.03
116
11.85
117
13.15
117
15.34
118
7.11
116
8.31
117
6.50
117
7.35
115
8.84
117
8.31
117
DPSimNet_ROBtwo views17.97
119
20.20
119
24.98
118
17.62
119
15.92
118
21.53
120
21.00
117
13.70
116
15.61
119
23.89
120
14.69
118
22.87
118
24.58
122
12.70
116
14.56
116
18.48
120
12.91
119
20.21
120
12.05
119
17.76
119
14.17
119
MEDIAN_ROBtwo views23.96
120
27.05
121
26.39
120
22.16
120
22.13
119
16.12
116
13.97
115
20.24
120
20.05
120
25.50
121
25.31
121
22.96
119
22.63
120
24.39
120
23.24
120
28.20
121
27.23
121
28.62
121
27.75
121
27.50
121
27.76
121
LSM0two views26.58
121
22.63
120
25.88
119
30.13
122
52.21
122
39.94
122
56.74
122
28.97
122
28.72
122
21.14
119
22.26
120
24.43
120
23.88
121
26.54
121
30.87
122
14.32
119
16.55
120
13.12
119
14.71
120
17.92
120
20.60
120
AVERAGE_ROBtwo views26.93
122
30.34
122
29.27
122
25.82
121
25.34
120
19.48
117
15.17
116
23.55
121
23.00
121
28.45
122
27.87
122
26.25
121
25.86
123
26.96
122
25.55
121
31.30
122
30.07
122
31.76
122
30.81
122
30.79
122
30.93
122
MSMDNettwo views0.43
23