Low-res two-view results

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):

bad 0.5, 1.0, 2.0, 4.0 [%]: Fraction of pixels with errors larger than the given number of disparities.
Average error [px]: The per-pixel average disparity error.
Root mean square error [px]: The per-pixel root mean square disparity error.
50%, 90%, 95%, 99% error quantile [px]: The highest disparity error within the given percentage of best pixels (for 50%, this is the median error).
Time [s]: The runtime of the method.

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	all	lakes. 1l	lakes. 1s	sand box 1l	sand box 1s	stora. room 1l	stora. room 1s	stora. room 2l	stora. room 2s	stora. room 2 1l	stora. room 2 1s	stora. room 2 2l	stora. room 2 2s	stora. room 3l	stora. room 3s	tunnel 1l	tunnel 1s	tunnel 2l	tunnel 2s	tunnel 3l	tunnel 3s

ccs	3.04 7	0.39 7	3.08 10	1.78 27	0.52 16	2.04 1	0.50 12	13.09 52	13.71 58	3.54 4	5.36 6	5.50 12	2.45 11	4.81 6	2.88 12	0.09 47	0.08 60	0.12 75	0.10 69	0.20 42	0.50 71

DMCA	3.29 10	1.05 33	4.18 22	1.60 21	2.87 69	2.11 2	0.60 15	7.95 7	4.65 6	8.62 32	8.59 19	9.24 31	4.68 33	6.25 8	3.03 14	0.04 37	0.09 64	0.06 69	0.03 50	0.18 39	0.07 18

AANet_RVC	5.01 38	1.74 69	6.38 49	1.96 38	1.29 41	2.26 3	1.69 36	10.07 17	18.53 75	7.88 26	18.15 66	8.49 29	2.70 14	10.59 51	7.04 43	0.96 97	0.15 75	0.02 54	0.00 1	0.13 31	0.12 33

CFNet_RVC	3.31 12	0.94 28	2.69 5	1.50 16	2.38 59	2.81 4	0.68 20	8.35 8	7.43 18	4.45 6	9.94 24	10.20 37	4.60 31	6.49 9	3.41 22	0.00 1	0.00 1	0.03 60	0.00 1	0.22 48	0.03 9

iResNet_ROB	4.23 29	1.02 31	4.90 28	2.18 43	0.93 32	2.92 5	0.37 8	15.10 70	16.91 72	7.89 27	10.51 27	7.03 19	3.07 17	8.16 26	3.46 24	0.01 18	0.00 1	0.00 1	0.00 1	0.10 23	0.02 3

DN-CSS_ROB	2.69 5	1.40 54	5.34 33	2.31 50	0.75 22	3.14 6	0.06 1	6.11 2	3.87 3	5.34 11	12.18 39	2.34 5	1.22 5	7.84 17	1.48 4	0.03 32	0.00 1	0.00 1	0.00 1	0.35 61	0.03 9

DeepPruner_ROB	3.52 14	1.14 42	4.06 20	1.12 7	1.65 47	3.65 7	0.83 24	13.96 60	4.47 5	7.80 25	10.84 29	7.05 21	2.16 10	8.14 25	3.08 17	0.07 45	0.03 40	0.00 1	0.01 38	0.32 57	0.06 17

R-Stereo Train	2.44 3	0.32 1	1.93 1	0.94 4	0.16 4	3.67 8	0.61 16	6.37 3	3.08 1	9.14 37	17.44 64	1.80 3	0.77 3	1.76 1	0.70 1	0.00 1	0.01 25	0.00 1	0.00 1	0.01 1	0.03 9

R-Stereo	2.44 3	0.32 1	1.93 1	0.94 4	0.16 4	3.67 8	0.61 16	6.37 3	3.08 1	9.14 37	17.44 64	1.80 3	0.77 3	1.76 1	0.70 1	0.00 1	0.01 25	0.00 1	0.00 1	0.01 1	0.03 9

STTStereo	3.60 15	0.93 27	6.34 48	2.71 61	2.23 58	3.68 10	0.63 19	9.42 12	6.73 11	9.87 45	6.97 9	8.84 30	3.65 19	6.85 10	3.04 15	0.00 1	0.02 35	0.01 43	0.00 1	0.02 4	0.02 3

DeepPrunerF	6.75 54	2.69 84	23.31 99	3.68 80	7.16 102	3.78 11	4.29 57	13.42 55	20.13 82	8.13 28	10.46 25	7.18 23	8.06 56	11.10 56	9.44 64	0.24 65	0.15 75	0.29 89	0.42 90	0.66 78	0.45 66

TDLM	4.11 25	1.11 40	3.54 14	1.62 22	1.04 35	3.91 12	7.41 81	10.60 29	10.67 42	6.38 15	12.59 43	5.95 15	4.77 35	8.79 35	3.04 15	0.58 87	0.00 1	0.01 43	0.00 1	0.19 41	0.12 33

CVANet_RVC	4.16 27	1.16 43	3.60 15	1.94 37	1.46 42	3.92 13	4.68 63	10.89 33	8.34 28	7.58 23	10.84 29	10.27 38	6.62 49	8.56 31	2.69 10	0.39 77	0.00 1	0.00 1	0.01 38	0.21 47	0.09 23

AdaStereo	3.09 8	0.58 12	3.04 9	2.84 63	0.48 15	4.08 14	1.29 31	12.16 44	7.77 22	6.03 12	9.62 23	5.79 14	1.53 7	4.56 5	1.93 7	0.00 1	0.00 1	0.00 1	0.00 1	0.02 4	0.02 3
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.
CC-Net-ROB	3.84 20	1.07 35	5.23 31	2.65 58	2.96 70	4.22 15	0.69 21	10.43 26	7.72 20	8.78 34	8.29 17	9.61 34	4.02 25	7.16 13	3.65 26	0.13 53	0.03 40	0.02 54	0.03 50	0.05 15	0.03 9

FADNet-RVC	3.98 23	1.84 74	12.48 80	1.69 25	0.44 13	4.33 16	1.31 32	11.84 38	7.15 13	3.53 3	3.50 2	10.63 40	4.43 30	9.12 38	6.25 37	0.03 32	0.10 65	0.00 1	0.03 50	0.60 75	0.25 53

iResNetv2_ROB	4.28 31	1.43 55	7.17 59	2.91 64	1.26 39	4.36 17	1.62 34	13.64 58	10.25 41	9.83 44	11.41 34	7.68 24	4.00 24	7.75 16	1.85 6	0.00 1	0.00 1	0.00 1	0.00 1	0.37 63	0.09 23

NLCA_NET_v2_RVC	3.84 20	1.06 34	5.23 31	2.72 62	3.27 72	4.36 17	0.61 16	10.71 30	7.56 19	8.75 33	7.89 14	9.86 36	3.90 23	7.15 12	3.44 23	0.14 54	0.02 35	0.02 54	0.03 50	0.04 13	0.03 9
Zhibo Rao, Mingyi He, Yuchao Dai, Zhidong Zhu, Bo Li, and Renjie He.: NLCA-Net: A non-local context attention network for stereo matching.
MLCV	3.44 13	0.88 21	5.60 36	1.39 12	0.25 7	4.36 17	0.33 6	7.25 5	7.28 15	9.17 39	12.24 41	5.09 10	2.47 12	9.15 39	3.23 19	0.00 1	0.00 1	0.00 1	0.00 1	0.10 23	0.02 3

Anonymous Stereo	6.16 48	3.15 91	23.75 100	2.97 69	2.48 63	4.39 20	13.30 97	9.21 10	9.86 37	9.56 42	8.76 20	6.79 18	1.99 9	13.50 70	13.04 81	0.01 18	0.05 48	0.00 1	0.06 63	0.22 48	0.19 44

iResNet	3.68 17	0.91 24	7.94 64	2.97 69	0.34 9	4.44 21	0.48 10	7.70 6	9.74 36	7.72 24	12.74 44	4.03 7	2.87 15	8.05 21	3.37 21	0.02 23	0.01 25	0.00 1	0.00 1	0.10 23	0.09 23

RASNet	4.52 35	0.61 14	4.42 24	3.42 77	4.68 94	4.58 22	0.99 25	9.54 14	8.01 23	5.28 10	11.42 35	10.34 39	8.88 61	9.28 40	8.68 59	0.15 56	0.00 1	0.00 1	0.00 1	0.03 10	0.04 15

FADNet	4.23 29	1.65 67	11.75 76	1.64 24	0.80 27	4.80 23	0.77 23	13.76 59	11.65 50	3.97 5	5.24 5	9.62 35	5.14 39	8.40 29	3.78 27	0.21 63	0.04 44	0.07 70	0.05 61	1.14 90	0.10 30

DPM-Stereo	1.97 1	0.64 16	2.95 8	0.17 1	0.10 1	4.83 24	0.13 2	8.60 9	4.06 4	6.42 16	4.92 4	0.44 1	0.72 1	3.57 3	1.80 5	0.00 1	0.01 25	0.00 1	0.00 1	0.05 15	0.04 15

CFNet	3.72 18	1.10 39	5.03 29	2.49 53	1.59 44	4.90 25	0.22 5	11.38 34	9.88 38	4.80 7	11.25 32	6.44 17	3.68 20	8.33 28	3.00 13	0.00 1	0.00 1	0.00 1	0.00 1	0.22 48	0.07 18

PMTNet	1.99 2	0.32 1	2.21 3	0.39 2	0.23 6	5.08 26	0.49 11	5.84 1	8.22 26	3.07 1	3.29 1	0.73 2	0.75 2	8.18 27	0.94 3	0.00 1	0.00 1	0.00 1	0.00 1	0.03 10	0.00 1

BEATNet_4x	3.24 9	1.27 48	5.89 41	1.56 18	0.10 1	5.26 27	1.07 26	10.08 18	5.50 8	6.89 21	7.73 13	4.53 9	4.13 26	5.05 7	5.27 34	0.04 37	0.05 48	0.00 1	0.00 1	0.23 52	0.23 47

PASM	7.90 66	4.22 95	21.97 97	3.25 73	3.29 73	5.39 28	6.57 75	10.57 28	19.09 77	12.77 64	13.92 50	18.11 74	9.51 65	13.79 73	10.77 74	0.19 61	0.45 93	0.29 89	1.08 101	1.49 96	1.19 93

ccs_rob	3.63 16	1.12 41	4.42 24	2.52 54	0.91 30	5.50 29	0.21 4	10.11 21	9.11 32	6.55 19	11.28 33	8.32 28	2.55 13	7.66 14	2.01 8	0.00 1	0.00 1	0.00 1	0.00 1	0.20 42	0.08 21

HITNet	2.79 6	0.77 17	4.02 19	2.03 39	0.11 3	5.58 30	0.59 14	9.24 11	5.15 7	6.42 16	7.26 11	3.66 6	2.92 16	4.07 4	3.87 28	0.00 1	0.00 1	0.00 1	0.00 1	0.06 19	0.02 3
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
NOSS_ROB	3.30 11	0.46 8	2.62 4	2.08 40	1.01 34	5.60 31	0.74 22	10.37 25	11.48 48	5.15 9	8.43 18	5.67 13	1.73 8	7.97 19	2.34 9	0.02 23	0.06 54	0.00 1	0.00 1	0.07 20	0.14 38

FADNet_RVC	3.91 22	1.67 68	12.95 83	0.96 6	0.75 22	5.71 32	0.54 13	10.83 32	6.60 10	3.46 2	8.09 15	4.10 8	3.40 18	9.43 42	6.33 38	0.36 75	0.44 92	0.17 81	0.46 94	0.91 85	0.95 88

FADNet-RVC-Resample	3.79 19	1.62 65	12.06 77	1.43 14	0.66 18	5.94 33	2.41 39	10.18 23	8.58 30	6.28 14	4.22 3	5.33 11	4.80 37	7.71 15	3.19 18	0.17 59	0.21 80	0.17 81	0.12 71	0.41 69	0.29 59

DLCB_ROB	4.51 33	0.91 24	3.78 17	2.19 44	1.07 36	6.28 34	3.09 43	9.78 15	7.72 20	10.65 49	12.97 45	13.91 55	3.71 21	8.72 33	5.30 35	0.00 1	0.00 1	0.00 1	0.00 1	0.03 10	0.10 30

XPNet_ROB	6.03 47	1.22 44	5.61 37	2.56 57	0.90 29	6.32 35	7.07 78	12.92 51	8.30 27	14.76 73	15.13 56	19.84 81	6.66 51	10.36 48	8.58 58	0.02 23	0.04 44	0.00 1	0.03 50	0.11 27	0.24 50

RP	6.84 58	1.29 52	5.53 35	3.92 86	5.18 96	6.32 35	3.53 49	11.73 37	15.31 64	9.54 41	22.38 74	18.25 75	14.47 82	10.11 46	7.49 45	0.91 96	0.01 25	0.12 75	0.15 73	0.33 58	0.19 44

edge stereo	6.76 55	1.01 30	6.76 55	2.20 45	2.45 62	6.41 37	2.45 40	14.84 67	11.98 52	15.29 74	18.31 67	22.02 87	12.56 75	10.82 52	7.49 45	0.03 32	0.06 54	0.11 74	0.03 50	0.30 54	0.14 38

ETE_ROB	5.80 42	1.77 71	6.33 47	1.44 15	0.78 26	6.43 38	6.90 77	12.53 48	8.08 24	12.93 66	14.89 54	21.13 86	5.87 44	9.83 44	6.57 41	0.04 37	0.01 25	0.00 1	0.02 42	0.08 22	0.33 60

HSM	4.00 24	0.79 18	3.16 12	1.59 20	2.17 56	6.77 39	1.11 27	12.28 45	6.35 9	6.75 20	8.11 16	13.90 54	5.37 42	8.85 36	2.71 11	0.00 1	0.00 1	0.00 1	0.00 1	0.02 4	0.02 3

CBMV_ROB	4.14 26	0.52 9	3.14 11	1.30 10	0.77 25	6.92 40	1.97 38	10.11 21	9.58 34	8.92 36	14.20 52	7.12 22	5.90 45	8.65 32	3.50 25	0.01 18	0.05 48	0.00 1	0.00 1	0.04 13	0.09 23

HSM-Net_RVC	4.20 28	0.32 1	2.76 6	0.63 3	0.69 20	6.95 41	1.69 36	11.96 39	8.36 29	8.83 35	12.17 38	15.18 62	4.21 28	6.91 11	3.30 20	0.02 23	0.02 35	0.00 1	0.00 1	0.01 1	0.01 2
Gengshan Yang, Joshua Manela, Michael Happold, and Deva Ramanan: Hierarchical Deep Stereo Matching on High-resolution Images. CVPR 2019
Abc-Net	6.77 56	1.49 58	6.48 50	2.92 66	4.40 85	7.43 42	3.61 51	19.52 92	13.29 55	8.39 30	16.91 60	15.96 63	12.13 73	12.85 65	7.70 47	1.47 100	0.11 67	0.01 43	0.42 90	0.14 35	0.24 50

NCC-stereo	6.77 56	1.49 58	6.48 50	2.92 66	4.40 85	7.43 42	3.61 51	19.52 92	13.29 55	8.39 30	16.91 60	15.96 63	12.13 73	12.85 65	7.70 47	1.47 100	0.11 67	0.01 43	0.42 90	0.14 35	0.24 50

PWC_ROB	8.24 70	3.13 89	12.74 81	2.43 52	4.43 88	7.51 44	1.22 28	16.63 79	19.24 78	16.08 76	28.29 87	13.99 56	10.16 67	13.63 72	14.06 83	0.42 82	0.00 1	0.05 67	0.00 1	0.59 74	0.27 57

DANet	6.02 46	1.23 45	8.45 66	3.86 85	3.94 83	7.64 45	1.34 33	9.51 13	7.00 12	13.39 68	15.53 57	15.99 65	7.02 52	12.14 60	12.37 78	0.19 61	0.12 71	0.02 54	0.03 50	0.13 31	0.56 75

NVstereo2D	4.51 33	0.82 19	6.86 57	3.28 74	3.38 76	8.16 46	3.13 44	10.51 27	15.15 63	4.90 8	6.89 8	7.87 25	4.78 36	9.88 45	3.91 29	0.01 18	0.00 1	0.00 1	0.06 63	0.02 4	0.58 76

AF-Net	7.78 65	1.44 56	6.68 54	3.37 75	4.50 92	8.61 47	2.69 41	17.07 81	20.17 83	9.52 40	24.02 77	20.31 83	14.59 83	11.58 58	9.84 69	0.61 88	0.00 1	0.12 75	0.00 1	0.38 64	0.12 33

PA-Net	4.98 37	1.47 57	7.42 61	2.40 51	2.14 55	8.73 48	3.64 53	12.42 46	13.11 54	7.03 22	7.57 12	7.88 26	6.52 48	10.16 47	7.82 49	0.02 23	0.03 40	0.00 1	0.00 1	0.11 27	1.07 91
Zhibo Rao, Mingyi He, Yuchao Dai, Zhelun Shen: Patch Attention Network with Generative Adversarial Model for Semi-Supervised Binocular Disparity Prediction.
RGC	6.88 59	2.23 80	6.13 44	4.05 87	4.73 95	8.94 49	2.78 42	15.19 72	11.74 51	11.13 53	19.34 69	17.86 72	10.42 68	13.02 67	8.03 53	0.73 90	0.01 25	0.24 88	0.41 89	0.31 56	0.38 62

GANet	6.22 49	1.07 35	4.07 21	2.27 47	0.89 28	9.19 50	9.52 86	12.02 40	8.13 25	10.72 50	29.09 89	13.86 53	7.52 55	11.00 55	4.39 30	0.36 75	0.00 1	0.02 54	0.02 42	0.12 29	0.08 21

StereoDRNet-Refined	4.46 32	0.62 15	3.80 18	1.92 34	0.40 11	9.35 51	0.15 3	10.02 16	8.83 31	12.69 62	11.62 36	9.34 32	3.87 22	8.06 22	8.02 51	0.00 1	0.00 1	0.01 43	0.05 61	0.20 42	0.26 56
Rohan Chabra, Julian Straub, Chris Sweeney, Richard Newcombe, Henry Fuchs: StereoDRNet. CVPR
pmcnn	7.72 64	1.27 48	9.42 69	2.91 64	3.14 71	9.44 52	6.23 70	12.56 49	16.51 70	14.53 71	24.08 78	27.44 95	8.49 58	9.32 41	8.44 57	0.06 44	0.08 60	0.00 1	0.00 1	0.30 54	0.15 40

SHD	9.61 80	2.60 83	12.46 79	3.69 81	3.54 78	9.47 53	1.25 29	20.16 97	37.84 105	18.19 80	21.24 71	16.96 68	12.83 78	14.47 78	16.05 86	0.32 73	0.13 72	0.01 43	0.08 66	0.38 64	0.48 69

NCCL2	5.88 44	1.59 63	5.44 34	1.87 30	0.92 31	9.55 54	11.55 95	12.11 41	9.94 39	9.67 43	8.85 21	22.28 88	7.41 53	8.78 34	7.17 44	0.01 18	0.00 1	0.03 60	0.00 1	0.13 31	0.23 47

PSMNet_ROB	5.02 39	1.63 66	6.03 43	1.90 33	1.83 52	9.57 55	6.35 73	15.58 75	7.23 14	6.15 13	10.48 26	12.22 47	4.16 27	8.02 20	8.71 60	0.02 23	0.01 25	0.01 43	0.10 69	0.20 42	0.12 33

NaN_ROB	6.00 45	1.24 46	6.29 46	1.34 11	1.68 49	9.60 56	10.31 91	15.09 68	15.79 66	12.62 61	8.95 22	11.67 46	5.83 43	11.78 59	6.41 40	0.05 42	0.13 72	0.08 71	0.20 76	0.22 48	0.79 84

SuperB	8.10 69	3.15 91	24.67 101	2.65 58	1.23 38	9.88 57	4.29 57	10.18 23	30.07 95	11.53 55	12.18 39	6.12 16	6.65 50	10.50 49	14.47 84	0.14 54	0.11 67	0.35 93	0.25 80	13.06 109	0.48 69

LALA_ROB	6.58 53	1.80 73	6.25 45	1.26 9	0.94 33	10.08 58	9.02 83	16.00 76	11.51 49	12.74 63	13.02 46	24.77 90	5.25 40	10.56 50	8.02 51	0.04 37	0.05 48	0.00 1	0.02 42	0.10 23	0.25 53

CBMV	5.35 40	0.91 24	3.67 16	1.62 22	0.44 13	10.09 59	7.19 80	12.49 47	12.33 53	12.22 58	14.69 53	10.93 41	6.48 47	8.51 30	4.96 33	0.02 23	0.15 75	0.00 1	0.00 1	0.17 38	0.17 42
Konstantinos Batsos, Changjiang Cai, Philippos Mordohai: CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation. Computer Vision and Pattern Recognition (CVPR) 2018
LSM	14.01 93	5.95 99	33.49 104	6.78 103	43.61 111	10.22 60	9.98 90	15.16 71	22.93 85	23.07 93	32.34 95	18.52 76	12.67 76	15.45 81	11.10 75	0.16 58	0.51 97	0.09 73	0.32 84	1.08 89	16.85 111

SGM-Forest	4.96 36	0.32 1	2.84 7	1.21 8	0.64 17	10.23 61	6.64 76	11.55 35	10.98 43	10.94 52	13.59 48	11.65 45	4.30 29	8.94 37	4.63 32	0.11 50	0.04 44	0.00 1	0.00 1	0.05 15	0.46 68
Johannes L. Schönberger, Sudipta Sinha, Marc Pollefeys: Learning to Fuse Proposals from Multiple Scanline Optimizations in Semi-Global Matching. ECCV 2018
PWCDC_ROB	7.92 67	3.17 93	7.48 62	5.73 98	4.40 85	10.45 62	0.35 7	14.52 65	28.19 92	10.36 48	31.27 92	7.04 20	9.14 64	13.22 69	8.78 61	2.74 105	0.02 35	0.00 1	0.00 1	1.31 94	0.17 42

StereoDRNet	5.59 41	1.75 70	6.80 56	3.12 71	4.45 89	10.61 63	4.35 59	18.80 85	9.73 35	12.22 58	6.87 7	11.44 44	4.65 32	8.09 24	8.26 55	0.02 23	0.11 67	0.00 1	0.03 50	0.20 42	0.28 58

Nwc_Net	6.97 60	1.25 47	6.63 53	3.82 84	3.37 75	10.83 64	1.67 35	19.56 94	11.35 46	8.36 29	23.62 76	17.19 70	11.44 72	11.21 57	8.08 54	0.80 92	0.00 1	0.00 1	0.02 42	0.13 31	0.09 23

DISCO	6.28 50	0.57 11	5.78 39	3.43 78	1.17 37	11.22 65	3.39 48	12.14 43	16.16 68	6.52 18	11.22 31	16.96 68	6.32 46	19.51 91	10.74 72	0.00 1	0.00 1	0.00 1	0.00 1	0.35 61	0.11 32

ADCMid	10.24 86	3.13 89	20.70 95	2.21 46	2.39 60	11.23 66	6.19 69	14.17 62	11.19 44	23.20 94	22.25 73	17.89 73	19.54 93	18.51 89	26.21 102	0.45 83	0.42 91	1.10 102	1.29 102	1.56 98	1.18 92

ADCRef	7.27 61	1.38 53	16.37 89	2.52 54	3.30 74	11.63 67	3.16 45	10.80 31	9.35 33	13.03 67	25.27 84	8.17 27	8.92 62	8.06 22	21.81 96	0.15 56	0.08 60	0.16 80	0.34 86	0.38 64	0.58 76

RTSC	9.15 76	3.00 88	13.57 85	3.72 82	1.76 51	11.82 68	0.46 9	16.95 80	36.83 102	15.80 75	15.53 57	12.91 50	7.46 54	20.01 93	21.76 95	0.31 72	0.13 72	0.01 43	0.08 66	0.57 72	0.41 65

DRN-Test	5.87 43	0.98 29	5.89 41	2.69 60	3.65 81	12.37 69	3.35 47	20.07 96	10.20 40	11.93 57	12.31 42	11.06 43	5.31 41	7.89 18	9.05 62	0.04 37	0.05 48	0.04 65	0.04 59	0.18 39	0.25 53

CSAN	7.62 62	1.60 64	6.56 52	1.83 28	0.66 18	12.40 70	10.52 93	14.45 64	21.32 84	14.19 70	15.98 59	17.84 71	13.02 79	12.32 62	8.38 56	0.09 47	0.07 58	0.03 60	0.04 59	0.33 58	0.67 82

NVStereoNet_ROB	16.04 98	6.75 101	12.90 82	6.37 102	7.42 105	12.89 71	9.74 88	22.78 101	25.12 88	30.32 99	46.19 109	34.37 103	25.38 102	21.48 94	21.38 92	5.94 108	3.10 108	6.07 109	10.09 111	4.01 103	8.54 109
Nikolai Smolyanskiy, Alexey Kamenev, Stan Birchfield: On the Importance of Stereo for Accurate Depth Estimation: An Efficient Semi-Supervised Deep Neural Network Approach. Arxiv
GANetREF_RVC	6.56 52	2.89 86	7.58 63	3.41 76	0.40 11	12.96 72	9.58 87	15.09 68	17.25 74	10.33 47	10.62 28	12.27 48	8.16 57	12.21 61	4.53 31	0.41 79	0.00 1	0.00 1	0.02 42	3.12 102	0.39 63
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_ROB	8.50 75	1.03 32	7.98 65	1.93 36	1.28 40	13.10 73	6.23 70	22.50 100	18.98 76	18.82 81	14.91 55	19.06 78	10.04 66	18.41 88	9.83 68	0.62 89	0.22 81	1.82 106	0.82 99	0.99 88	1.36 95

XQC	8.43 74	3.58 94	16.40 90	2.92 66	2.17 56	13.22 74	3.60 50	14.64 66	25.86 89	11.87 56	12.04 37	15.06 61	10.67 69	15.24 80	19.41 89	0.39 77	0.08 60	0.05 67	0.07 65	0.84 84	0.45 66

WCMA_ROB	9.21 77	0.87 20	7.37 60	2.54 56	2.13 54	13.59 75	5.80 67	11.64 36	14.01 59	24.43 95	32.99 96	27.09 94	18.02 87	12.51 64	9.85 70	0.81 93	0.07 58	0.01 43	0.01 38	0.16 37	0.23 47

ADCS	13.02 91	4.93 97	28.38 102	3.17 72	2.67 66	13.61 76	10.83 94	18.70 84	33.46 98	22.59 89	24.78 81	19.59 80	18.51 90	23.40 98	32.16 109	0.10 49	0.19 79	0.37 95	0.18 74	1.26 93	1.46 97

DispFullNet	17.47 105	26.01 109	33.98 105	22.58 110	20.86 110	13.84 77	1.28 30	16.50 77	26.27 90	19.97 86	17.17 63	20.52 85	18.49 89	22.86 97	10.76 73	5.13 106	2.83 107	30.72 111	7.72 109	20.86 110	11.01 110

SANet	10.64 87	1.86 75	10.91 71	1.76 26	0.71 21	14.62 78	9.23 85	19.18 87	37.14 103	19.22 82	27.96 86	25.86 93	19.11 92	13.02 67	10.63 71	0.08 46	0.06 54	0.03 60	0.02 42	0.62 76	0.81 85

AnyNet_C32	10.98 89	5.58 98	22.79 98	4.16 88	5.83 98	15.64 79	14.30 99	13.18 53	17.15 73	16.44 78	20.52 70	14.68 60	13.44 80	22.46 95	30.08 107	0.17 59	0.26 84	0.36 94	0.36 87	1.23 92	0.91 87

RYNet	6.34 51	0.89 23	5.88 40	1.41 13	4.48 91	15.97 80	4.18 56	13.41 54	16.49 69	10.81 51	7.00 10	14.33 57	8.72 60	9.43 42	13.71 82	0.00 1	0.01 25	0.00 1	0.00 1	0.02 4	0.07 18

MSMD_ROB	9.28 78	1.09 38	4.65 27	1.58 19	0.39 10	16.52 81	4.41 60	13.60 56	14.87 62	22.34 88	39.89 103	25.67 92	20.71 96	12.42 63	6.98 42	0.34 74	0.03 40	0.00 1	0.00 1	0.05 15	0.09 23

ADCP+	8.09 68	1.79 72	14.50 87	1.54 17	4.28 84	16.57 82	5.20 65	12.80 50	11.20 45	12.83 65	17.07 62	11.02 42	10.80 70	17.59 87	23.18 99	0.03 32	0.05 48	0.01 43	0.18 74	0.39 68	0.81 85

LE_ROB	16.73 103	1.28 51	11.61 75	3.72 82	1.65 47	16.67 83	9.17 84	14.39 63	55.91 111	63.81 111	40.86 106	35.94 104	37.73 110	14.24 75	26.87 103	0.05 42	0.10 65	0.13 78	0.22 77	0.12 29	0.15 40

ADCPNet	9.54 79	2.39 82	31.46 103	2.09 41	1.60 45	16.71 84	6.39 74	12.11 41	11.45 47	13.53 69	21.45 72	19.41 79	10.94 71	14.38 76	21.54 94	0.27 70	1.16 100	0.39 96	1.49 104	0.58 73	1.45 96

FC-DCNN	10.72 88	0.52 9	4.27 23	1.88 31	1.63 46	17.18 85	5.29 66	18.20 83	19.69 81	28.50 98	34.51 99	34.03 102	21.48 99	15.89 82	11.15 76	0.03 32	0.01 25	0.02 54	0.01 38	0.07 20	0.09 23

PVD	15.44 96	2.93 87	14.67 88	4.21 91	3.39 77	17.43 86	4.16 55	27.84 107	48.84 110	31.02 101	43.54 107	29.76 98	30.81 107	25.97 101	21.40 93	0.23 64	0.41 89	0.04 65	0.33 85	0.41 69	1.33 94

stereogan	7.69 63	0.88 21	7.08 58	3.49 79	3.93 82	18.98 87	3.23 46	16.52 78	19.58 80	9.93 46	18.92 68	20.50 84	9.04 63	14.07 74	6.14 36	0.26 67	0.04 44	0.21 86	0.03 50	0.63 77	0.33 60

SGM_RVC	10.08 83	0.60 13	3.42 13	2.30 49	0.32 8	19.41 88	6.33 72	18.95 86	14.64 60	25.14 96	24.32 80	33.34 101	18.79 91	19.86 92	12.55 80	0.25 66	0.26 84	0.22 87	0.24 79	0.34 60	0.40 64
Heiko Hirschmueller: Stereo processing by semiglobal matching and mutual information. TPAMI 2008, Volume 30(2), pp. 328-341
MFMNet_re	13.29 92	8.60 104	18.29 92	9.75 105	7.25 104	19.65 89	14.84 103	20.71 99	30.72 96	23.03 92	28.77 88	18.85 77	26.09 104	13.55 71	9.82 67	2.44 103	1.35 103	0.34 92	0.23 78	4.78 105	6.69 106

MeshStereo	11.52 90	1.52 60	4.55 26	1.89 32	1.46 42	19.87 90	5.11 64	20.66 98	15.91 67	32.67 103	34.51 99	39.34 107	21.15 97	18.74 90	12.10 77	0.11 50	0.06 54	0.01 43	0.00 1	0.45 71	0.22 46
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
PDISCO_ROB	9.62 81	1.99 77	11.51 74	9.88 106	9.61 107	21.48 91	3.83 54	19.33 88	28.49 93	11.27 54	14.17 51	19.92 82	5.02 38	16.35 84	9.18 63	5.28 107	0.41 89	0.14 79	0.09 68	2.05 99	2.36 103

ADCL	10.16 85	2.11 78	19.36 94	1.92 34	1.88 53	22.23 92	8.91 82	14.04 61	23.56 86	14.62 72	26.19 85	12.75 49	13.59 81	16.06 83	22.95 98	0.26 67	0.18 78	0.75 99	0.65 95	0.69 79	0.58 76

SPS-STEREO	15.04 95	6.23 100	13.21 84	11.34 107	11.65 109	23.30 93	7.15 79	24.16 102	15.65 65	31.78 102	29.19 90	31.62 99	21.32 98	24.62 99	19.50 90	7.59 109	4.19 110	3.22 107	1.48 103	6.99 108	6.54 105
K. Yamaguchi, D. McAllester, R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation. ECCV 2014
MFN_U_SF_DS_RVC	9.78 82	4.27 96	14.47 86	2.29 48	2.85 68	23.40 94	13.62 98	13.60 56	19.54 79	19.42 83	24.27 79	16.74 67	8.59 59	17.05 86	7.98 50	1.25 99	1.68 104	0.17 81	2.63 106	0.72 80	1.04 89

DPSNet	10.14 84	1.88 76	16.82 91	1.85 29	1.73 50	24.84 95	17.20 107	19.92 95	27.41 91	12.23 60	13.62 49	16.52 66	18.35 88	14.42 77	12.50 79	0.78 91	0.54 98	0.08 71	0.25 80	1.18 91	0.59 79

G-Net	8.41 72	1.54 61	10.97 72	5.73 98	3.60 79	26.19 96	4.41 60	10.10 19	7.42 16	19.71 84	24.99 82	14.38 58	15.83 84	10.99 53	9.53 65	0.50 85	0.46 94	0.19 84	0.25 80	0.80 82	0.66 80

STTStereo_v2	8.41 72	1.54 61	10.97 72	5.73 98	3.60 79	26.19 96	4.41 60	10.10 19	7.42 16	19.71 84	24.99 82	14.38 58	15.83 84	10.99 53	9.53 65	0.50 85	0.46 94	0.19 84	0.25 80	0.80 82	0.66 80

SGM+DAISY	15.62 97	7.26 103	19.28 93	8.94 104	10.11 108	26.25 98	10.49 92	19.36 89	14.65 61	30.64 100	33.59 97	33.00 100	22.32 100	24.96 100	16.42 87	7.90 110	6.25 111	4.51 108	3.37 107	5.86 106	7.20 107

MDST_ROB	8.37 71	0.32 1	9.03 67	4.18 90	2.42 61	26.86 99	6.14 68	19.36 89	13.52 57	27.09 97	22.75 75	9.47 33	4.74 34	15.06 79	6.34 39	0.02 23	0.02 35	0.00 1	0.00 1	0.02 4	0.13 37

ELAS_RVC	16.54 101	2.26 81	10.09 70	5.50 97	4.46 90	28.28 100	16.72 106	25.55 104	33.54 99	40.19 106	40.30 105	36.68 105	30.03 105	29.40 103	20.61 91	0.98 98	1.21 101	0.86 101	0.70 97	1.39 95	2.16 100
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
AnyNet_C01	16.12 99	10.81 108	59.36 109	4.42 92	2.49 64	30.06 101	15.15 105	17.51 82	16.51 70	17.88 79	37.69 102	24.04 89	17.54 86	29.60 104	33.29 110	0.28 71	0.38 87	0.43 97	0.42 90	2.57 101	1.98 99

MANE	19.47 108	1.27 48	5.07 30	4.69 93	5.55 97	30.49 102	9.94 89	34.01 109	37.27 104	44.13 108	51.57 112	52.51 109	40.41 111	33.58 105	24.81 101	0.89 95	0.86 99	1.11 103	9.72 110	0.38 64	1.06 90

SGM-ForestM	16.99 104	1.08 37	5.74 38	2.12 42	0.75 22	31.63 103	12.21 96	27.80 106	32.25 97	37.88 105	39.99 104	52.96 110	35.20 109	33.60 106	24.47 100	0.26 67	0.39 88	0.31 91	0.39 88	0.26 53	0.53 74

ELAS	16.72 102	2.14 79	9.23 68	4.92 94	4.53 93	32.66 104	15.11 104	27.40 105	28.68 94	40.27 107	44.90 108	38.33 106	30.50 106	26.44 102	21.94 97	0.88 94	1.23 102	0.67 98	0.89 100	1.49 96	2.18 102
A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching. ACCV 2010
SAMSARA	14.63 94	2.74 85	12.38 78	12.65 108	6.74 101	36.50 105	72.93 113	19.36 89	23.77 87	16.20 77	13.04 47	29.21 96	12.78 77	16.98 85	15.21 85	0.11 50	0.26 84	0.03 60	0.14 72	0.76 81	0.77 83

MSC_U_SF_DS_RVC	16.41 100	6.93 102	21.83 96	5.94 101	2.81 67	38.71 106	14.59 100	24.55 103	34.87 100	33.66 104	34.35 98	29.24 97	24.20 101	22.59 96	17.95 88	2.52 104	2.81 106	1.17 104	1.51 105	5.89 107	2.16 100

PWCK	30.53 111	44.32 111	47.25 108	29.76 111	7.23 103	40.78 107	27.10 109	44.73 111	44.32 109	47.31 109	36.37 101	47.16 108	26.05 103	41.26 110	31.87 108	21.83 111	4.03 109	29.50 110	4.67 108	27.17 111	7.80 108

RTS	18.87 106	9.32 106	86.48 111	4.95 95	6.10 99	42.08 108	14.70 101	15.49 73	41.06 107	22.65 90	32.32 93	13.77 51	19.54 93	37.98 107	28.96 104	0.41 79	0.23 82	0.00 1	0.02 42	0.91 85	0.50 71

RTSA	18.87 106	9.32 106	86.48 111	4.95 95	6.10 99	42.08 108	14.70 101	15.49 73	41.06 107	22.65 90	32.32 93	13.77 51	19.54 93	37.98 107	28.96 104	0.41 79	0.23 82	0.00 1	0.02 42	0.91 85	0.50 71

BEATNet-Init1	23.31 109	9.03 105	41.67 106	4.17 89	2.53 65	45.68 110	19.47 108	33.43 108	38.45 106	47.59 110	49.10 110	59.31 111	41.80 112	38.35 109	29.21 106	0.47 84	0.50 96	0.81 100	0.66 96	2.10 100	1.86 98

MADNet+	27.07 110	33.84 110	90.97 113	20.14 109	7.47 106	48.43 111	47.10 110	35.43 110	36.46 101	20.11 87	30.05 91	25.29 91	35.08 108	45.50 111	50.28 111	2.13 102	2.00 105	1.19 105	0.76 98	4.71 104	4.43 104

DPSimNet_ROB	53.45 112	64.73 112	44.39 107	53.97 112	45.39 112	53.66 112	54.83 111	55.15 112	57.87 112	64.16 112	50.83 111	63.40 112	53.34 113	46.45 112	65.81 112	63.13 112	26.54 112	57.94 112	51.11 112	45.52 112	50.69 112

MADNet++	82.84 113	82.38 113	73.57 110	87.72 113	82.97 113	93.14 113	69.15 112	86.42 113	82.50 113	93.46 113	86.70 113	86.28 113	80.92 114	88.34 113	88.84 113	86.83 113	84.17 113	72.64 113	68.92 113	80.47 113	81.42 113

MEDIAN_ROB	98.41 114	99.70 114	99.30 115	97.09 114	97.02 114	96.89 114	95.77 115	97.66 114	97.28 114	98.79 116	98.94 114	99.18 114	98.14 115	96.89 114	96.88 114	99.96 116	99.16 114	100.00 114	99.99 114	99.69 114	99.88 114

AVERAGE_ROB	99.62 115	99.95 115	98.81 114	100.00 119	100.00 115	98.08 115	95.47 114	100.00 117	100.00 115	100.00 117	100.00 115	100.00 117	100.00 116	100.00 117	99.99 115	100.00 118	100.00 115	100.00 114	100.00 115	100.00 117	100.00 118

DPSM	99.95 118	100.00 116	100.00 118	99.76 115	100.00 115	100.00 116	100.00 116	100.00 117	100.00 115	100.00 117	100.00 115	100.00 117	100.00 116	100.00 117	100.00 118	99.21 114	100.00 115	100.00 114	100.00 115	99.99 115	99.95 115

DPSMNet_ROB	99.91 117	100.00 116	99.99 116	99.99 117	100.00 115	100.00 116	100.00 116	99.98 116	100.00 115	98.35 114	100.00 115	99.84 115	100.00 116	99.98 115	99.99 115	100.00 118	100.00 115	100.00 114	100.00 115	100.00 117	100.00 118

DPSM_ROB	99.95 118	100.00 116	100.00 118	99.76 115	100.00 115	100.00 116	100.00 116	100.00 117	100.00 115	100.00 117	100.00 115	100.00 117	100.00 116	100.00 117	100.00 118	99.21 114	100.00 115	100.00 114	100.00 115	99.99 115	99.95 115

DGTPSM_ROB	99.90 116	100.00 116	99.99 116	99.99 117	100.00 115	100.00 116	100.00 116	99.97 115	100.00 115	98.35 114	100.00 115	99.84 115	100.00 116	99.98 115	99.99 115	99.99 117	100.00 115	100.00 114	100.00 115	100.00 117	100.00 118

LSM0	100.00 120	100.00 116	100.00 118	100.00 119	100.00 115	100.00 116	100.00 116	100.00 117	100.00 115	100.00 117	100.00 115	100.00 117	100.00 116	100.00 117	100.00 118	100.00 118	100.00 115	100.00 114	100.00 115	100.00 117	99.99 117

MSMDNet													1.26 6

Low-res two-view benchmark