Low-res two-view benchmark

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
sort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysorted bysort bysort bysort bysort bysort bysort bysort bysort by
DN-CSS_ROBtwo views2.69
1		1.40
21		5.34
12		2.31
21		0.75
9		3.14
2		0.06
1		6.11
1		3.87
1		5.34
2		12.18
10		2.34
1		1.22
1		7.84
1		1.48
1		0.03
15		0.00
1		0.00
1		0.00
1		0.35
25		0.03
3
NOSS_ROBtwo views3.30
2		0.46
3		2.62
1		2.08
17		1.01
16		5.60
4		0.74
6		10.37
6		11.48
14		5.15
1		8.43
3		5.67
3		1.73
2		7.97
3		2.34
2		0.02
8		0.06
20		0.00
1		0.00
1		0.07
7		0.14
11
iResNettwo views3.68
3		0.91
10		7.94
25		2.97
27		0.34
2		4.44
3		0.48
5		7.70
2		9.74
10		7.72
6		12.74
12		4.03
2		2.87
3		8.05
5		3.37
4		0.02
8		0.01
9		0.00
1		0.00
1		0.10
9		0.09
4
iResNet_ROBtwo views4.23
6		1.02
14		4.90
11		2.18
18		0.93
14		2.92
1		0.37
4		15.10
21		16.91
27		7.89
7		10.51
7		7.03
4		3.07
4		8.16
8		3.46
5		0.01
5		0.00
1		0.00
1		0.00
1		0.10
9		0.02
1
DLCB_ROBtwo views4.51
8		0.91
10		3.78
7		2.19
19		1.07
17		6.28
5		3.09
11		9.78
3		7.72
4		10.65
11		12.97
13		13.91
16		3.71
5		8.72
11		5.30
9		0.00
1		0.00
1		0.00
1		0.00
1		0.03
3		0.10
7
StereoDRNet-Filteredtwo views4.46
7		0.62
7		3.80
8		1.92
15		0.40
4		9.35
11		0.15
2		10.02
4		8.83
7		12.69
19		11.62
9		9.34
7		3.87
6		8.06
6		8.02
15		0.00
1		0.00
1		0.01
19		0.05
26		0.20
18		0.26
22
PSMNet_ROBtwo views5.02
10		1.63
25		6.03
17		1.90
14		1.83
24		9.57
14		6.35
23		15.58
22		7.23
3		6.15
3		10.48
6		12.22
14		4.16
7		8.02
4		8.71
21		0.02
8		0.01
9		0.01
19		0.10
28		0.20
18		0.12
9
SGM-Foresttwo views4.96
9		0.32
1		2.84
2		1.21
1		0.64
6		10.23
18		6.64
24		11.55
7		10.98
13		10.94
12		13.59
15		11.65
12		4.30
8		8.94
14		4.63
7		0.11
24		0.04
15		0.00
1		0.00
1		0.05
5		0.46
27
Johannes L. Schönberger, Sudipta Sinha, Marc Pollefeys: Learning to Fuse Proposals from Multiple Scanline Optimizations in Semi-Global Matching. ECCV 2018
StereoDRNettwo views5.59
12		1.75
26		6.80
22		3.12
28		4.45
32		10.61
20		4.35
15		18.80
26		9.73
9		12.22
15		6.87
1		11.44
11		4.65
9		8.09
7		8.26
17		0.02
8		0.11
27		0.00
1		0.03
22		0.20
18		0.28
24
MDST_ROBtwo views8.37
24		0.32
1		9.03
27		4.18
31		2.42
27		26.86
38		6.14
19		19.36
30		13.52
17		27.09
33		22.75
25		9.47
8		4.74
10		15.06
29		6.34
10		0.02
8		0.02
12		0.00
1		0.00
1		0.02
1		0.13
10
PDISCO_ROBtwo views9.62
28		1.99
31		11.51
32		9.88
38		9.61
38		21.48
34		3.83
14		19.33
29		28.49
35		11.27
13		14.17
17		19.92
24		5.02
11		16.35
30		9.18
24		5.28
37		0.41
32		0.14
31		0.09
27		2.05
35		2.36
35
LALA_ROBtwo views6.58
19		1.80
28		6.25
18		1.26
2		0.94
15		10.08
16		9.02
29		16.00
23		11.51
15		12.74
20		13.02
14		24.77
28		5.25
12		10.56
18		8.02
15		0.04
16		0.05
17		0.00
1		0.02
19		0.10
9		0.25
20
DRN-Testtwo views5.87
14		0.98
13		5.89
16		2.69
25		3.65
29		12.37
22		3.35
12		20.07
33		10.20
12		11.93
14		12.31
11		11.06
10		5.31
13		7.89
2		9.05
23		0.04
16		0.05
17		0.04
26		0.04
24		0.18
17		0.25
20
HSMtwo views4.00
4		0.79
8		3.16
4		1.59
7		2.17
26		6.77
8		1.11
7		12.28
11		6.35
2		6.75
5		8.11
2		13.90
15		5.37
14		8.85
13		2.71
3		0.00
1		0.00
1		0.00
1		0.00
1		0.02
1		0.02
1
NaN_ROBtwo views6.00
16		1.24
18		6.29
19		1.34
4		1.68
22		9.60
15		10.31
33		15.09
20		15.79
23		12.62
18		8.95
5		11.67
13		5.83
15		11.78
19		6.41
11		0.05
19		0.13
28		0.08
28		0.20
29		0.22
21		0.79
30
ETE_ROBtwo views5.80
13		1.77
27		6.33
20		1.44
5		0.78
11		6.43
7		6.90
25		12.53
13		8.08
5		12.93
21		14.89
20		21.13
26		5.87
16		9.83
16		6.57
12		0.04
16		0.01
9		0.00
1		0.02
19		0.08
8		0.33
25
CBMV_ROBtwo views4.14
5		0.52
4		3.14
3		1.30
3		0.77
10		6.92
9		1.97
10		10.11
5		9.58
8		8.92
8		14.20
18		7.12
6		5.90
17		8.65
10		3.50
6		0.01
5		0.05
17		0.00
1		0.00
1		0.04
4		0.09
4
DISCOtwo views6.28
18		0.57
5		5.78
15		3.43
29		1.17
18		11.22
21		3.39
13		12.14
10		16.16
25		6.52
4		11.22
8		16.96
19		6.32
18		19.51
33		10.74
29		0.00
1		0.00
1		0.00
1		0.00
1		0.35
25		0.11
8
CBMVpermissivetwo views5.35
11		0.91
10		3.67
6		1.62
8		0.44
5		10.09
17		7.19
28		12.49
12		12.33
16		12.22
15		14.69
19		10.93
9		6.48
19		8.51
9		4.96
8		0.02
8		0.15
29		0.00
1		0.00
1		0.17
16		0.17
14
Konstantinos Batsos, Changjiang Cai, Philippos Mordohai: CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation. Computer Vision and Pattern Recognition (CVPR) 2018
XPNet_ROBtwo views6.03
17		1.22
17		5.61
14		2.56
24		0.90
12		6.32
6		7.07
26		12.92
15		8.30
6		14.76
24		15.13
22		19.84
23		6.66
20		10.36
17		8.58
20		0.02
8		0.04
15		0.00
1		0.03
22		0.11
12		0.24
19
NCCL2two views5.88
15		1.59
23		5.44
13		1.87
12		0.92
13		9.55
13		11.55
36		12.11
9		9.94
11		9.67
9		8.85
4		22.28
27		7.41
21		8.78
12		7.17
14		0.01
5		0.00
1		0.03
23		0.00
1		0.13
14		0.23
17
pmcnntwo views7.72
21		1.27
19		9.42
29		2.91
26		3.14
28		9.44
12		6.23
20		12.56
14		16.51
26		14.53
23		24.08
26		27.44
32		8.49
22		9.32
15		8.44
19		0.06
21		0.08
25		0.00
1		0.00
1		0.30
22		0.15
12
PWCDC_ROBbinarytwo views7.92
22		3.17
35		7.48
24		5.73
34		4.40
30		10.45
19		0.35
3		14.52
19		28.19
34		10.36
10		31.27
32		7.04
5		9.14
23		13.22
24		8.78
22		2.74
35		0.02
12		0.00
1		0.00
1		1.31
32		0.17
14
FBW_ROBtwo views8.50
25		1.03
15		7.98
26		1.93
16		1.28
19		13.10
25		6.23
20		22.50
36		18.98
28		18.82
26		14.91
21		19.06
22		10.04
24		18.41
31		9.83
26		0.62
29		0.22
30		1.82
36		0.82
35		0.99
30		1.36
32
PWC_ROBbinarytwo views8.24
23		3.13
34		12.74
34		2.43
22		4.43
31		7.51
10		1.22
8		16.63
25		19.24
29		16.08
25		28.29
29		13.99
17		10.16
25		13.63
26		14.06
34		0.42
28		0.00
1		0.05
27		0.00
1		0.59
28		0.27
23
CSANtwo views7.62
20		1.60
24		6.56
21		1.83
10		0.66
7		12.40
23		10.52
35		14.45
18		21.32
30		14.19
22		15.98
23		17.84
20		13.02
26		12.32
20		8.38
18		0.09
23		0.07
23		0.03
23		0.04
24		0.33
23		0.67
29
WCMA_ROBtwo views9.21
26		0.87
9		7.37
23		2.54
23		2.13
25		13.59
26		5.80
18		11.64
8		14.01
18		24.43
31		32.99
33		27.09
31		18.02
27		12.51
22		9.85
27		0.81
31		0.07
23		0.01
19		0.01
18		0.16
15		0.23
17
testNettwo views10.14
30		1.88
30		16.82
37		1.85
11		1.73
23		24.84
36		17.20
40		19.92
32		27.41
33		12.23
17		13.62
16		16.52
18		18.35
28		14.42
28		12.50
32		0.78
30		0.54
33		0.08
28		0.25
33		1.18
31		0.59
28
DispFullNettwo views17.47
40		26.01
40		33.98
40		22.58
40		20.86
41		13.84
27		1.28
9		16.50
24		26.27
32		19.97
28		17.17
24		20.52
25		18.49
29		22.86
36		10.76
30		5.13
36		2.83
37		30.72
41		7.72
40		20.86
40		11.01
41
SGM_ROBbinarytwo views10.08
29		0.60
6		3.42
5		2.30
20		0.32
1		19.41
31		6.33
22		18.95
27		14.64
19		25.14
32		24.32
27		33.34
35		18.79
30		19.86
34		12.55
33		0.25
26		0.26
31		0.22
32		0.24
32		0.34
24		0.40
26
Heiko Hirschmueller: Stereo processing by semiglobal matching and mutual information. TPAMI 2008, Volume 30(2), pp. 328-341
SANettwo views10.64
31		1.86
29		10.91
31		1.76
9		0.71
8		14.62
28		9.23
31		19.18
28		37.14
39		19.22
27		27.96
28		25.86
30		19.11
31		13.02
23		10.63
28		0.08
22		0.06
20		0.03
23		0.02
19		0.62
29		0.81
31
MSMD_ROBtwo views9.28
27		1.09
16		4.65
10		1.58
6		0.39
3		16.52
29		4.41
16		13.60
16		14.87
21		22.34
29		39.89
37		25.67
29		20.71
32		12.42
21		6.98
13		0.34
27		0.03
14		0.00
1		0.00
1		0.05
5		0.09
4
MeshStereopermissivetwo views11.52
32		1.52
22		4.55
9		1.89
13		1.46
20		19.87
33		5.11
17		20.66
34		15.91
24		32.67
37		34.51
35		39.34
40		21.15
33		18.74
32		12.10
31		0.11
24		0.06
20		0.01
19		0.00
1		0.45
27		0.22
16
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
SPS-STEREOcopylefttwo views15.04
34		6.23
36		13.21
36		11.34
39		11.65
40		23.30
35		7.15
27		24.16
38		15.65
22		31.78
36		29.19
31		31.62
33		21.32
34		24.62
37		19.50
36		7.59
39		4.19
40		3.22
37		1.48
37		6.99
39		6.54
36
K. Yamaguchi, D. McAllester, R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation. ECCV 2014
SGM+DAISYtwo views15.62
35		7.26
38		19.28
39		8.94
36		10.11
39		26.25
37		10.49
34		19.36
30		14.65
20		30.64
35		33.59
34		33.00
34		22.32
35		24.96
38		16.42
35		7.90
40		6.25
41		4.51
38		3.37
38		5.86
38		7.20
38
NVStereoNet_ROBtwo views16.04
36		6.75
37		12.90
35		6.37
35		7.42
37		12.89
24		9.74
32		22.78
37		25.12
31		30.32
34		46.19
41		34.37
36		25.38
36		21.48
35		21.38
38		5.94
38		3.10
38		6.07
39		10.09
41		4.01
36		8.54
40
Nikolai Smolyanskiy, Alexey Kamenev, Stan Birchfield: On the Importance of Stereo for Accurate Depth Estimation: An Efficient Semi-Supervised Deep Neural Network Approach. Arxiv
PWCKtwo views30.53
41		44.32
41		47.25
42		29.76
41		7.23
35		40.78
41		27.10
41		44.73
41		44.32
40		47.31
40		36.37
36		47.16
41		26.05
37		41.26
41		31.87
41		21.83
41		4.03
39		29.50
40		4.67
39		27.17
41		7.80
39
MFMNet_retwo views13.29
33		8.60
39		18.29
38		9.75
37		7.25
36		19.65
32		14.84
37		20.71
35		30.72
37		23.03
30		28.77
30		18.85
21		26.09
38		13.55
25		9.82
25		2.44
34		1.35
36		0.34
33		0.23
31		4.78
37		6.69
37
ELAS_ROBcopylefttwo views16.54
37		2.26
33		10.09
30		5.50
33		4.46
33		28.28
39		16.72
39		25.55
39		33.54
38		40.19
38		40.30
38		36.68
38		30.03
39		29.40
40		20.61
37		0.98
33		1.21
34		0.86
35		0.70
34		1.39
33		2.16
33
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
ELAScopylefttwo views16.72
38		2.14
32		9.23
28		4.92
32		4.53
34		32.66
40		15.11
38		27.40
40		28.68
36		40.27
39		44.90
40		38.33
39		30.50
40		26.44
39		21.94
39		0.88
32		1.23
35		0.67
34		0.89
36		1.49
34		2.18
34
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
LE_ROBtwo views16.73
39		1.28
20		11.61
33		3.72
30		1.65
21		16.67
30		9.17
30		14.39
17		55.91
41		63.81
41		40.86
39		35.94
37		37.73
41		14.24
27		26.87
40		0.05
19		0.10
26		0.13
30		0.22
30		0.12
13		0.15
12
DPSimNet_ROBtwo views53.45
42		64.73
42		44.39
41		53.97
42		45.39
42		53.66
42		54.83
42		55.15
42		57.87
42		64.16
42		50.83
42		63.40
42		53.34
42		46.45
42		65.81
42		63.13
42		26.54
42		57.94
42		51.11
42		45.52
42		50.69
42
MEDIAN_ROBtwo views98.41
43		99.70
43		99.30
44		97.09
43		97.02
43		96.89
43		95.77
44		97.66
43		97.28
43		98.79
45		98.94
43		99.18
43		98.14
43		96.89
43		96.88
43		99.96
45		99.16
43		100.00
43		99.99
43		99.69
43		99.88
43
AVERAGE_ROBtwo views99.62
44		99.95
44		98.81
43		100.00
48		100.00
44		98.08
44		95.47
43		100.00
46		100.00
44		100.00
46		100.00
44		100.00
46		100.00
44		100.00
46		99.99
44		100.00
47		100.00
44		100.00
43		100.00
44		100.00
46		100.00
46
DPSMtwo views99.95
47		100.00
45		100.00
47		99.76
44		100.00
44		100.00
45		100.00
45		100.00
46		100.00
44		100.00
46		100.00
44		100.00
46		100.00
44		100.00
46		100.00
47		99.21
43		100.00
44		100.00
43		100.00
44		99.99
44		99.95
44
DPSM_ROBtwo views99.95
47		100.00
45		100.00
47		99.76
44		100.00
44		100.00
45		100.00
45		100.00
46		100.00
44		100.00
46		100.00
44		100.00
46		100.00
44		100.00
46		100.00
47		99.21
43		100.00
44		100.00
43		100.00
44		99.99
44		99.95
44
DPSMNet_ROBtwo views99.91
46		100.00
45		99.99
45		99.99
46		100.00
44		100.00
45		100.00
45		99.98
45		100.00
44		98.35
43		100.00
44		99.84
44		100.00
44		99.98
44		99.99
44		100.00
47		100.00
44		100.00
43		100.00
44		100.00
46		100.00
46
DGTPSM_ROBtwo views99.90
45		100.00
45		99.99
45		99.99
46		100.00
44		100.00
45		100.00
45		99.97
44		100.00
44		98.35
43		100.00
44		99.84
44		100.00
44		99.98
44		99.99
44		99.99
46		100.00
44		100.00
43		100.00
44		100.00
46		100.00
46