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).
Methods with suffix _ROB may participate in the Robust Vision Challenge .
training datasets . The visualizations may not work with mobile browsers.
Coverage : dense
Set : Test
Metric : bad 1.0 [%]
Mask : non-occluded
Method
Info all
iResNetv2_ROB 4.28 8 1.43 23 7.17 24 2.91 27 1.26 20 4.36 3 1.62 11 13.64 18 10.25 14 9.83 11 11.41 9 7.68 8 4.00 8 7.75 1 1.85 2 0.00 1 0.00 1 0.00 1 0.00 1 0.37 28 0.09 5 MLCV 3.44 3 0.88 10 5.60 14 1.39 5 0.25 1 4.36 3 0.33 3 7.25 2 7.28 4 9.17 9 12.24 12 5.09 3 2.47 3 9.15 16 3.23 5 0.00 1 0.00 1 0.00 1 0.00 1 0.10 9 0.02 1 DLCB_ROB 4.51 10 0.91 11 3.78 7 2.19 20 1.07 18 6.28 7 3.09 13 9.78 4 7.72 5 10.65 13 12.97 15 13.91 18 3.71 6 8.72 12 5.30 11 0.00 1 0.00 1 0.00 1 0.00 1 0.03 3 0.10 9 DISCO 6.28 20 0.57 5 5.78 16 3.43 31 1.17 19 11.22 23 3.39 15 12.14 11 16.16 27 6.52 4 11.22 8 16.96 21 6.32 20 19.51 35 10.74 31 0.00 1 0.00 1 0.00 1 0.00 1 0.35 26 0.11 10 StereoDRNet-Refined 4.46 9 0.62 7 3.80 8 1.92 16 0.40 5 9.35 13 0.15 2 10.02 5 8.83 8 12.69 21 11.62 10 9.34 9 3.87 7 8.06 7 8.02 17 0.00 1 0.00 1 0.01 21 0.05 28 0.20 19 0.26 24 HSM 4.00 5 0.79 8 3.16 4 1.59 8 2.17 28 6.77 10 1.11 8 12.28 12 6.35 2 6.75 5 8.11 2 13.90 17 5.37 16 8.85 14 2.71 4 0.00 1 0.00 1 0.00 1 0.00 1 0.02 1 0.02 1 NCCL2 5.88 17 1.59 25 5.44 13 1.87 13 0.92 14 9.55 15 11.55 38 12.11 10 9.94 12 9.67 10 8.85 4 22.28 29 7.41 23 8.78 13 7.17 16 0.01 7 0.00 1 0.03 25 0.00 1 0.13 15 0.23 19 CBMV_ROB 4.14 6 0.52 4 3.14 3 1.30 3 0.77 11 6.92 11 1.97 12 10.11 6 9.58 9 8.92 8 14.20 20 7.12 7 5.90 19 8.65 11 3.50 8 0.01 7 0.05 19 0.00 1 0.00 1 0.04 4 0.09 5 iResNet_ROB 4.23 7 1.02 15 4.90 11 2.18 19 0.93 15 2.92 1 0.37 5 15.10 23 16.91 29 7.89 7 10.51 7 7.03 5 3.07 5 8.16 9 3.46 7 0.01 7 0.00 1 0.00 1 0.00 1 0.10 9 0.02 1 PSMNet_ROB 5.02 12 1.63 27 6.03 18 1.90 15 1.83 26 9.57 16 6.35 25 15.58 24 7.23 3 6.15 3 10.48 6 12.22 16 4.16 9 8.02 5 8.71 23 0.02 10 0.01 11 0.01 21 0.10 30 0.20 19 0.12 11 StereoDRNet 5.59 14 1.75 28 6.80 23 3.12 30 4.45 34 10.61 22 4.35 17 18.80 28 9.73 10 12.22 17 6.87 1 11.44 13 4.65 11 8.09 8 8.26 19 0.02 10 0.11 29 0.00 1 0.03 24 0.20 19 0.28 26 iResNet 3.68 4 0.91 11 7.94 27 2.97 29 0.34 3 4.44 5 0.48 6 7.70 3 9.74 11 7.72 6 12.74 14 4.03 2 2.87 4 8.05 6 3.37 6 0.02 10 0.01 11 0.00 1 0.00 1 0.10 9 0.09 5 CBMV 5.35 13 0.91 11 3.67 6 1.62 9 0.44 6 10.09 19 7.19 30 12.49 13 12.33 18 12.22 17 14.69 21 10.93 11 6.48 21 8.51 10 4.96 10 0.02 10 0.15 31 0.00 1 0.00 1 0.17 17 0.17 16 Konstantinos Batsos, Changjiang Cai, Philippos Mordohai: CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation . Computer Vision and Pattern Recognition (CVPR) 2018 XPNet_ROB 6.03 19 1.22 18 5.61 15 2.56 25 0.90 13 6.32 8 7.07 28 12.92 16 8.30 7 14.76 26 15.13 24 19.84 25 6.66 22 10.36 19 8.58 22 0.02 10 0.04 17 0.00 1 0.03 24 0.11 13 0.24 21 MDST_ROB 8.37 26 0.32 1 9.03 29 4.18 33 2.42 29 26.86 40 6.14 21 19.36 32 13.52 19 27.09 35 22.75 27 9.47 10 4.74 12 15.06 31 6.34 12 0.02 10 0.02 14 0.00 1 0.00 1 0.02 1 0.13 12 NOSS_ROB 3.30 2 0.46 3 2.62 1 2.08 18 1.01 17 5.60 6 0.74 7 10.37 7 11.48 16 5.15 1 8.43 3 5.67 4 1.73 2 7.97 4 2.34 3 0.02 10 0.06 22 0.00 1 0.00 1 0.07 7 0.14 13 DN-CSS_ROB 2.69 1 1.40 22 5.34 12 2.31 22 0.75 10 3.14 2 0.06 1 6.11 1 3.87 1 5.34 2 12.18 11 2.34 1 1.22 1 7.84 2 1.48 1 0.03 17 0.00 1 0.00 1 0.00 1 0.35 26 0.03 4 ETE_ROB 5.80 15 1.77 29 6.33 21 1.44 6 0.78 12 6.43 9 6.90 27 12.53 14 8.08 6 12.93 23 14.89 22 21.13 28 5.87 18 9.83 18 6.57 14 0.04 18 0.01 11 0.00 1 0.02 21 0.08 8 0.33 27 LALA_ROB 6.58 21 1.80 30 6.25 19 1.26 2 0.94 16 10.08 18 9.02 31 16.00 25 11.51 17 12.74 22 13.02 16 24.77 30 5.25 14 10.56 20 8.02 17 0.04 18 0.05 19 0.00 1 0.02 21 0.10 9 0.25 22 DRN-Test 5.87 16 0.98 14 5.89 17 2.69 26 3.65 31 12.37 24 3.35 14 20.07 35 10.20 13 11.93 16 12.31 13 11.06 12 5.31 15 7.89 3 9.05 25 0.04 18 0.05 19 0.04 28 0.04 26 0.18 18 0.25 22 LE_ROB 16.73 41 1.28 21 11.61 35 3.72 32 1.65 23 16.67 32 9.17 32 14.39 19 55.91 43 63.81 43 40.86 41 35.94 39 37.73 43 14.24 29 26.87 42 0.05 21 0.10 28 0.13 32 0.22 32 0.12 14 0.15 14 NaN_ROB 6.00 18 1.24 19 6.29 20 1.34 4 1.68 24 9.60 17 10.31 35 15.09 22 15.79 25 12.62 20 8.95 5 11.67 15 5.83 17 11.78 21 6.41 13 0.05 21 0.13 30 0.08 30 0.20 31 0.22 22 0.79 32 pmcnn 7.72 23 1.27 20 9.42 31 2.91 27 3.14 30 9.44 14 6.23 22 12.56 15 16.51 28 14.53 25 24.08 28 27.44 34 8.49 24 9.32 17 8.44 21 0.06 23 0.08 27 0.00 1 0.00 1 0.30 23 0.15 14 SANet 10.64 33 1.86 31 10.91 33 1.76 10 0.71 9 14.62 30 9.23 33 19.18 30 37.14 41 19.22 29 27.96 30 25.86 32 19.11 33 13.02 25 10.63 30 0.08 24 0.06 22 0.03 25 0.02 21 0.62 31 0.81 33 CSAN 7.62 22 1.60 26 6.56 22 1.83 11 0.66 8 12.40 25 10.52 37 14.45 20 21.32 32 14.19 24 15.98 25 17.84 22 13.02 28 12.32 22 8.38 20 0.09 25 0.07 25 0.03 25 0.04 26 0.33 24 0.67 31 MeshStereo 11.52 34 1.52 24 4.55 9 1.89 14 1.46 22 19.87 35 5.11 19 20.66 36 15.91 26 32.67 39 34.51 37 39.34 42 21.15 35 18.74 34 12.10 33 0.11 26 0.06 22 0.01 21 0.00 1 0.45 29 0.22 18 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 SGM-Forest 4.96 11 0.32 1 2.84 2 1.21 1 0.64 7 10.23 20 6.64 26 11.55 8 10.98 15 10.94 14 13.59 17 11.65 14 4.30 10 8.94 15 4.63 9 0.11 26 0.04 17 0.00 1 0.00 1 0.05 5 0.46 29 Johannes L. Schönberger, Sudipta Sinha, Marc Pollefeys: Learning to Fuse Proposals from Multiple Scanline Optimizations in Semi-Global Matching . ECCV 2018 SGM_ROB 10.08 31 0.60 6 3.42 5 2.30 21 0.32 2 19.41 33 6.33 24 18.95 29 14.64 21 25.14 34 24.32 29 33.34 37 18.79 32 19.86 36 12.55 35 0.25 28 0.26 33 0.22 34 0.24 34 0.34 25 0.40 28 Heiko Hirschmueller: Stereo processing by semiglobal matching and mutual information . TPAMI 2008, Volume 30(2), pp. 328-341 MSMD_ROB 9.28 29 1.09 17 4.65 10 1.58 7 0.39 4 16.52 31 4.41 18 13.60 17 14.87 23 22.34 31 39.89 39 25.67 31 20.71 34 12.42 23 6.98 15 0.34 29 0.03 16 0.00 1 0.00 1 0.05 5 0.09 5 PWC_ROB 8.24 25 3.13 36 12.74 36 2.43 23 4.43 33 7.51 12 1.22 9 16.63 27 19.24 31 16.08 27 28.29 31 13.99 19 10.16 27 13.63 28 14.06 36 0.42 30 0.00 1 0.05 29 0.00 1 0.59 30 0.27 25 FBW_ROB 8.50 27 1.03 16 7.98 28 1.93 17 1.28 21 13.10 27 6.23 22 22.50 38 18.98 30 18.82 28 14.91 23 19.06 24 10.04 26 18.41 33 9.83 28 0.62 31 0.22 32 1.82 38 0.82 37 0.99 32 1.36 34 DPSNet 10.14 32 1.88 32 16.82 39 1.85 12 1.73 25 24.84 38 17.20 42 19.92 34 27.41 35 12.23 19 13.62 18 16.52 20 18.35 30 14.42 30 12.50 34 0.78 32 0.54 35 0.08 30 0.25 35 1.18 33 0.59 30 WCMA_ROB 9.21 28 0.87 9 7.37 25 2.54 24 2.13 27 13.59 28 5.80 20 11.64 9 14.01 20 24.43 33 32.99 35 27.09 33 18.02 29 12.51 24 9.85 29 0.81 33 0.07 25 0.01 21 0.01 20 0.16 16 0.23 19 ELAS 16.72 40 2.14 34 9.23 30 4.92 34 4.53 36 32.66 42 15.11 40 27.40 42 28.68 38 40.27 41 44.90 42 38.33 41 30.50 42 26.44 41 21.94 41 0.88 34 1.23 37 0.67 36 0.89 38 1.49 36 2.18 36 A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching . ACCV 2010 ELAS_ROB 16.54 39 2.26 35 10.09 32 5.50 35 4.46 35 28.28 41 16.72 41 25.55 41 33.54 40 40.19 40 40.30 40 36.68 40 30.03 41 29.40 42 20.61 39 0.98 35 1.21 36 0.86 37 0.70 36 1.39 35 2.16 35 A. Geiger, M. Roser, R. Urtasun: Efficient large-scale stereo matching . ACCV 2010 MFMNet_re 13.29 35 8.60 41 18.29 40 9.75 39 7.25 38 19.65 34 14.84 39 20.71 37 30.72 39 23.03 32 28.77 32 18.85 23 26.09 40 13.55 27 9.82 27 2.44 36 1.35 38 0.34 35 0.23 33 4.78 39 6.69 39 PWCDC_ROB 7.92 24 3.17 37 7.48 26 5.73 36 4.40 32 10.45 21 0.35 4 14.52 21 28.19 36 10.36 12 31.27 34 7.04 6 9.14 25 13.22 26 8.78 24 2.74 37 0.02 14 0.00 1 0.00 1 1.31 34 0.17 16 DispFullNet 17.47 42 26.01 42 33.98 42 22.58 42 20.86 43 13.84 29 1.28 10 16.50 26 26.27 34 19.97 30 17.17 26 20.52 27 18.49 31 22.86 38 10.76 32 5.13 38 2.83 39 30.72 43 7.72 42 20.86 42 11.01 43 PDISCO_ROB 9.62 30 1.99 33 11.51 34 9.88 40 9.61 40 21.48 36 3.83 16 19.33 31 28.49 37 11.27 15 14.17 19 19.92 26 5.02 13 16.35 32 9.18 26 5.28 39 0.41 34 0.14 33 0.09 29 2.05 37 2.36 37 NVStereoNet_ROB 16.04 38 6.75 39 12.90 37 6.37 37 7.42 39 12.89 26 9.74 34 22.78 39 25.12 33 30.32 36 46.19 43 34.37 38 25.38 38 21.48 37 21.38 40 5.94 40 3.10 40 6.07 41 10.09 43 4.01 38 8.54 42 Nikolai Smolyanskiy, Alexey Kamenev, Stan Birchfield: On the Importance of Stereo for Accurate Depth Estimation: An Efficient Semi-Supervised Deep Neural Network Approach . Arxiv SPS-STEREO 15.04 36 6.23 38 13.21 38 11.34 41 11.65 42 23.30 37 7.15 29 24.16 40 15.65 24 31.78 38 29.19 33 31.62 35 21.32 36 24.62 39 19.50 38 7.59 41 4.19 42 3.22 39 1.48 39 6.99 41 6.54 38 K. Yamaguchi, D. McAllester, R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation . ECCV 2014 SGM+DAISY 15.62 37 7.26 40 19.28 41 8.94 38 10.11 41 26.25 39 10.49 36 19.36 32 14.65 22 30.64 37 33.59 36 33.00 36 22.32 37 24.96 40 16.42 37 7.90 42 6.25 43 4.51 40 3.37 40 5.86 40 7.20 40 PWCK 30.53 43 44.32 43 47.25 44 29.76 43 7.23 37 40.78 43 27.10 43 44.73 43 44.32 42 47.31 42 36.37 38 47.16 43 26.05 39 41.26 43 31.87 43 21.83 43 4.03 41 29.50 42 4.67 41 27.17 43 7.80 41 DPSimNet_ROB 53.45 44 64.73 44 44.39 43 53.97 44 45.39 44 53.66 44 54.83 44 55.15 44 57.87 44 64.16 44 50.83 44 63.40 44 53.34 44 46.45 44 65.81 44 63.13 44 26.54 44 57.94 44 51.11 44 45.52 44 50.69 44 DPSM 99.95 49 100.00 47 100.00 49 99.76 46 100.00 46 100.00 47 100.00 47 100.00 48 100.00 46 100.00 48 100.00 46 100.00 48 100.00 46 100.00 48 100.00 49 99.21 45 100.00 46 100.00 45 100.00 46 99.99 46 99.95 46 DPSM_ROB 99.95 49 100.00 47 100.00 49 99.76 46 100.00 46 100.00 47 100.00 47 100.00 48 100.00 46 100.00 48 100.00 46 100.00 48 100.00 46 100.00 48 100.00 49 99.21 45 100.00 46 100.00 45 100.00 46 99.99 46 99.95 46 MEDIAN_ROB 98.41 45 99.70 45 99.30 46 97.09 45 97.02 45 96.89 45 95.77 46 97.66 45 97.28 45 98.79 47 98.94 45 99.18 45 98.14 45 96.89 45 96.88 45 99.96 47 99.16 45 100.00 45 99.99 45 99.69 45 99.88 45 DGTPSM_ROB 99.90 47 100.00 47 99.99 47 99.99 48 100.00 46 100.00 47 100.00 47 99.97 46 100.00 46 98.35 45 100.00 46 99.84 46 100.00 46 99.98 46 99.99 46 99.99 48 100.00 46 100.00 45 100.00 46 100.00 48 100.00 48 AVERAGE_ROB 99.62 46 99.95 46 98.81 45 100.00 50 100.00 46 98.08 46 95.47 45 100.00 48 100.00 46 100.00 48 100.00 46 100.00 48 100.00 46 100.00 48 99.99 46 100.00 49 100.00 46 100.00 45 100.00 46 100.00 48 100.00 48 DPSMNet_ROB 99.91 48 100.00 47 99.99 47 99.99 48 100.00 46 100.00 47 100.00 47 99.98 47 100.00 46 98.35 45 100.00 46 99.84 46 100.00 46 99.98 46 99.99 46 100.00 49 100.00 46 100.00 45 100.00 46 100.00 48 100.00 48
