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:

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.

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 bysorted bysort bysort bysort bysort bysort bysort bysort bysort bysort by
DN-CSS_ROBtwo views2.69
1		1.40
7		5.34
6		2.31
8		0.75
5		3.14
1		0.06
1		6.11
1		3.87
1		5.34
1		12.18
2		2.34
1		1.22
1		7.84
2		1.48
1		0.03
3		0.00
1		0.00
1		0.00
1		0.35
9		0.03
1
PWCDC_ROBbinarytwo views5.58
5		3.02
13		8.92
10		5.39
13		3.68
11		3.52
2		0.57
2		11.74
4		17.57
13		6.51
3		20.00
6		4.75
2		5.50
5		6.82
1		10.21
9		0.74
11		0.00
1		1.54
14		0.00
1		0.77
12		0.39
10
CBMVpermissivetwo views5.35
4		0.91
4		3.67
3		1.62
4		0.44
3		10.09
7		7.19
12		12.49
5		12.33
5		12.22
5		14.69
5		10.93
3		6.48
6		8.51
4		4.96
3		0.02
1		0.15
11		0.00
1		0.00
1		0.17
5		0.17
5
Konstantinos Batsos, Changjiang Cai, Philippos Mordohai: CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation. Computer Vision and Pattern Recognition (CVPR) 2018
SGM-Foresttwo views4.96
2		0.32
1		2.84
1		1.21
1		0.64
4		10.23
8		6.64
10		11.55
2		10.98
3		10.94
4		13.59
4		11.65
4		4.30
3		8.94
5		4.63
2		0.11
6		0.04
6		0.00
1		0.00
1		0.05
1		0.46
12
PSMNet_ROBtwo views5.02
3		1.63
9		6.03
7		1.90
6		1.83
8		9.57
5		6.35
9		15.58
8		7.23
2		6.15
2		10.48
1		12.22
5		4.16
2		8.02
3		8.71
7		0.02
1		0.01
4		0.01
7		0.10
11		0.20
6		0.12
3
PWC_ROBbinarytwo views8.24
8		3.13
14		12.74
14		2.43
9		4.43
12		7.51
3		1.22
3		16.63
10		19.24
14		16.08
8		28.29
9		13.99
6		10.16
8		13.63
10		14.06
12		0.42
10		0.00
1		0.05
10		0.00
1		0.59
11		0.27
9
LaLa_ROBtwo views6.58
6		1.80
10		6.25
8		1.26
2		0.94
6		10.08
6		9.02
13		16.00
9		11.51
4		12.74
6		13.02
3		24.77
7		5.25
4		10.56
7		8.02
5		0.04
4		0.05
7		0.00
1		0.02
10		0.10
3		0.25
8
MSMD_ROBtwo views9.28
10		1.09
5		4.65
5		1.58
3		0.39
2		16.52
10		4.41
4		13.60
7		14.87
9		22.34
9		39.89
15		25.67
8		20.71
11		12.42
8		6.98
4		0.34
9		0.03
5		0.00
1		0.00
1		0.05
1		0.09
2
WCMA_ROBtwo views9.21
9		0.87
3		7.37
9		2.54
10		2.13
9		13.59
9		5.80
6		11.64
3		14.01
6		24.43
10		32.99
11		27.09
9		18.02
9		12.51
9		9.85
8		0.81
12		0.07
9		0.01
7		0.01
9		0.16
4		0.23
7
pmcnntwo views7.72
7		1.27
6		9.42
12		2.91
11		3.14
10		9.44
4		6.23
7		12.56
6		16.51
12		14.53
7		24.08
7		27.44
10		8.49
7		9.32
6		8.44
6		0.06
5		0.08
10		0.00
1		0.00
1		0.30
7		0.15
4
SPS-STEREOcopylefttwo views15.04
13		6.23
15		13.21
15		11.34
16		11.65
17		23.30
13		7.15
11		24.16
14		15.65
10		31.78
13		29.19
10		31.62
11		21.32
13		24.62
13		19.50
14		7.59
15		4.19
16		3.22
15		1.48
15		6.99
16		6.54
15
K. Yamaguchi, D. McAllester, R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation. ECCV 2014
SGM+DAISYtwo views15.62
14		7.26
16		19.28
16		8.94
15		10.11
16		26.25
14		10.49
14		19.36
12		14.65
8		30.64
12		33.59
12		33.00
12		22.32
14		24.96
14		16.42
13		7.90
16		6.25
17		4.51
16		3.37
16		5.86
15		7.20
16
SGM_ROBbinarytwo views10.08
11		0.60
2		3.42
2		2.30
7		0.32
1		19.41
11		6.33
8		18.95
11		14.64
7		25.14
11		24.32
8		33.34
13		18.79
10		19.86
12		12.55
11		0.25
8		0.26
12		0.22
11		0.24
12		0.34
8		0.40
11
Heiko Hirschmueller: Stereo processing by semiglobal matching and mutual information. TPAMI 2008, Volume 30(2), pp. 328-341
ELAS_ROBcopylefttwo views16.54
15		2.26
12		10.09
13		5.50
14		4.46
13		28.28
15		16.72
16		25.55
15		33.54
16		40.19
15		40.30
16		36.68
14		30.03
16		29.40
16		20.61
15		0.98
14		1.21
13		0.86
13		0.70
13		1.39
13		2.16
13
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
ELAScopylefttwo views16.72
16		2.14
11		9.23
11		4.92
12		4.53
14		32.66
16		15.11
15		27.40
16		28.68
15		40.27
16		44.90
17		38.33
15		30.50
17		26.44
15		21.94
16		0.88
13		1.23
14		0.67
12		0.89
14		1.49
14		2.18
14
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
MeshStereopermissivetwo views11.52
12		1.52
8		4.55
4		1.89
5		1.46
7		19.87
12		5.11
5		20.66
13		15.91
11		32.67
14		34.51
13		39.34
16		21.15
12		18.74
11		12.10
10		0.11
6		0.06
8		0.01
7		0.00
1		0.45
10		0.22
6
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
PWCKtwo views30.53
17		44.32
17		47.25
17		29.76
17		7.23
15		40.78
17		27.10
17		44.73
17		44.32
17		47.31
17		36.37
14		47.16
17		26.05
15		41.26
17		31.87
17		21.83
17		4.03
15		29.50
17		4.67
17		27.17
17		7.80
17
MEDIAN_ROBtwo views98.41
18		99.70
18		99.30
19		97.09
18		97.02
18		96.89
18		95.77
19		97.66
18		97.28
18		98.79
18		98.94
18		99.18
18		98.14
18		96.89
18		96.88
18		99.96
18		99.16
18		100.00
18		99.99
18		99.69
18		99.88
18
AVERAGE_ROBtwo views99.62
19		99.95
19		98.81
18		100.00
19		100.00
19		98.08
19		95.47
18		100.00
19		100.00
19		100.00
19		100.00
19		100.00
19		100.00
19		100.00
19		99.99
19		100.00
19		100.00
19		100.00
18		100.00
19		100.00
19		100.00
19