This table lists the benchmark results for the low-res many-view scenario. The following metrics are evaluated:

(*) For exact definitions, detailing how potentially incomplete ground truth is taken into account, see our paper.

The datasets are grouped into different categories, and result averages are computed for a category and method if results of the method are available for all datasets within the category. Note that the category "all" includes both the high-res multi-view and the low-res many-view scenarios.

Methods with suffix _ROB may participate in the Robust Vision Challenge.

Click a dataset result cell to show a visualization of the reconstruction. For training datasets, ground truth and accuracy / completeness visualizations are also available. The visualizations may not work with mobile browsers.




Method Infoalllow-res
many-view
indooroutdoorlakesidesand boxstorage roomstorage room 2tunnel
sort bysorted bysort bysort bysort bysort bysort bysort bysort by
unMVSv197.23 197.65 296.96 496.94 998.74 195.80 399.50 195.19 3
PCF-MVS98.23 397.14 296.43 497.61 199.41 198.23 395.62 497.24 1495.20 2
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
DeepC-MVS_fast98.92 196.93 396.34 697.33 398.83 298.21 593.89 698.79 394.96 5
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
DeepPCF-MVS98.03 496.80 495.79 897.47 298.75 398.48 293.36 798.23 595.19 3
ACMM97.17 696.69 597.35 396.24 897.15 797.93 896.54 298.16 793.65 10
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
tm-dncc96.60 698.77 195.16 1494.86 2094.57 2798.54 198.99 296.07 1
ACMP97.00 896.57 796.40 596.68 697.45 598.22 494.57 598.23 594.36 7
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
COLMAP(SR)95.99 894.88 1196.72 597.34 698.20 691.93 1197.84 994.64 6
PLCcopyleft98.56 295.81 995.80 795.82 1195.90 1697.46 1193.21 898.39 494.11 9
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
COLMAP(base)95.66 1095.29 995.90 1096.87 1097.33 1292.61 997.97 893.51 11
ACMH+96.20 1395.49 1194.61 1296.08 997.07 897.95 791.77 1297.44 1193.22 12
DeepC-MVS97.84 595.05 1293.14 2096.32 797.72 497.02 1589.53 2296.74 1794.22 8
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
ACMH96.25 1195.02 1394.18 1495.57 1296.81 1297.62 1091.51 1496.86 1692.28 15
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
A-TVSNet + Gipumacopyleft94.87 1495.11 1094.71 1595.40 1897.17 1492.41 1097.81 1091.56 19
COLMAP_ROBcopyleft97.10 794.23 1594.16 1594.27 1695.49 1795.89 2191.39 1596.93 1591.43 20
Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
A1Net94.08 1692.17 2395.34 1396.68 1397.20 1387.75 2496.60 2092.15 17
TAPA-MVS(SR)93.88 1793.60 1894.06 1793.96 2296.07 1990.58 1796.63 1992.16 16
TAPA-MVS96.40 1093.41 1893.58 1993.30 1892.79 2495.34 2390.46 1896.70 1891.77 18
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
GSE93.24 1994.32 1392.51 2192.17 2694.21 3091.39 1597.25 1391.15 21
IB-MVS96.24 1293.10 2094.00 1692.50 2292.76 2597.79 991.65 1396.35 2186.93 24
Christian Sormann, Mattia Rossi, Andreas Kuhn and Friedrich Fraundorfer: IB-MVS: An Iterative Algorithm for Deep Multi-View Stereo based on Binary Decisions. BMVC 2021
BP-MVSNet92.81 2192.74 2192.86 2095.28 1990.85 3989.66 2195.81 2392.46 14
Christian Sormann, Patrick Knöbelreiter, Andreas Kuhn, Mattia Rossi, Thomas Pock, Friedrich Fraundorfer: BP-MVSNet: Belief-Propagation-Layers for Multi-View-Stereo. 3DV 2020
LTVRE_ROB95.29 1692.81 2192.60 2292.96 1993.47 2392.40 3589.87 2095.32 2493.01 13
Andreas Kuhn, Heiko Hirschmüller, Daniel Scharstein, Helmut Mayer: A TV Prior for High-Quality Scalable Multi-View Stereo Reconstruction. International Journal of Computer Vision 2016
CPR_FA90.55 2388.48 2791.92 2490.55 2994.73 2687.56 2589.40 3090.48 22
HY-MVS96.53 989.40 2489.89 2589.07 3187.26 3494.23 2986.76 2693.03 2785.74 25
test_mvsss89.03 2587.86 2889.81 2889.81 3095.40 2279.81 3495.90 2284.23 26
3Dnovator95.63 1487.41 2680.54 3492.00 2396.84 1196.13 1778.30 3982.78 3783.03 29
3Dnovator+95.58 1587.31 2780.52 3591.84 2596.45 1496.11 1877.89 4083.15 3682.96 31
LPCS87.27 2890.71 2484.98 3683.44 4087.91 4387.85 2393.57 2683.58 28
PVSNet_093.57 1986.78 2979.85 3891.40 2790.80 2896.34 1681.38 3378.33 4287.07 23
OpenMVScopyleft95.20 1786.65 3079.07 3991.71 2696.24 1595.92 2077.14 4281.00 4182.97 30
R-MVSNet86.00 3185.66 3086.23 3384.90 3790.99 3885.10 2886.21 3182.78 32
test_120585.92 3286.91 2985.25 3587.23 3592.99 3382.56 3091.26 2975.54 40
tmmvs85.21 3378.69 4089.56 2994.86 2094.57 2775.94 4581.44 3979.26 35
ANet-0.7584.53 3493.78 1778.36 4576.21 5182.53 5390.17 1997.39 1276.33 38
PVSNet94.91 1884.37 3576.88 4489.36 3089.15 3195.01 2481.51 3272.25 5183.92 27
test_112682.97 3681.27 3284.10 4087.91 3391.40 3777.35 4185.20 3272.99 45
PVSNet_LR82.84 3780.11 3684.66 3883.98 3990.58 4076.03 4484.20 3379.43 34
CIDER82.80 3873.82 4688.78 3291.08 2794.80 2576.99 4370.65 5480.46 33
Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume. AAAI 2020
metmvs_fine80.20 3980.57 3379.95 4279.03 4885.63 4978.78 3682.37 3875.20 41
OpenMVS_ROBcopyleft88.34 2079.83 4071.69 4985.26 3487.13 3692.62 3472.32 5071.05 5376.04 39
unsupervisedMVS_cas79.34 4177.46 4380.59 4182.05 4486.11 4678.48 3776.43 4673.62 44
ANet77.67 4280.06 3776.08 5276.21 5182.53 5379.07 3581.06 4069.50 52
PSD-MVSNet77.01 4378.31 4176.15 5075.37 5479.17 5778.36 3878.25 4373.92 42
RMVSNet77.00 4482.85 3173.11 6077.13 5085.46 5074.25 4791.44 2856.73 70
P-MVSNet76.80 4589.15 2668.57 6665.42 6770.78 7084.62 2993.68 2569.51 51
MVS_test_175.95 4672.76 4878.07 4678.02 4990.51 4161.59 5983.93 3465.68 57
SGNet75.22 4775.91 4574.77 5573.83 5677.61 6274.98 4676.83 4572.87 46
hgnet74.50 4868.37 5378.58 4383.21 4185.96 4768.21 5368.54 5566.57 54
DPSNet74.50 4868.37 5378.58 4383.21 4185.96 4768.21 5368.54 5566.57 54
test373.46 5073.44 4773.47 5872.10 6076.90 6371.62 5175.25 4771.43 50
MVEpermissive68.59 2171.88 5169.57 5173.41 5966.25 6688.47 4267.60 5571.54 5265.52 59
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
example71.40 5262.28 5977.48 4780.61 4786.65 4557.51 6167.05 6065.18 60
mvs_zhu_103071.20 5366.84 5674.10 5770.30 6380.28 5660.09 6073.60 4971.71 48
Snet71.04 5450.26 7584.90 3781.53 4694.02 3146.31 7454.21 7779.14 36
Pnet_fast70.97 5550.67 7484.50 3982.73 4391.73 3646.07 7755.28 7679.04 37
firsttry70.92 5669.37 5271.96 6170.42 6280.33 5564.15 5874.59 4865.15 61
TVSNet69.80 5769.63 5069.92 6469.08 6472.29 6965.78 5773.47 5068.39 53
MVSNet_plusplus69.70 5860.08 6076.11 5184.56 3870.03 7353.06 6567.09 5973.74 43
MVSNet_++68.46 5955.01 7177.42 4881.98 4578.45 6132.64 8477.38 4471.83 47
AttMVS68.12 6066.47 5769.22 6565.33 6875.76 6469.66 5263.29 6566.57 54
CasMVSNet(SR_A)68.03 6156.91 6775.44 5476.18 5386.99 4449.31 7164.52 6463.16 62
Pnet-new-67.47 6254.64 7276.02 5373.83 5682.59 5249.74 7059.53 7071.64 49
Pnet-eth67.00 6377.95 4259.70 7461.15 7252.38 8272.56 4883.33 3565.56 58
CasMVSNet(base)66.90 6455.99 6974.18 5675.37 5484.67 5146.93 7365.06 6262.49 63
Pnet-blend++66.07 6559.38 6170.53 6273.22 5879.12 5850.55 6868.21 5759.25 65
Pnet-blend66.07 6559.38 6170.53 6273.22 5879.12 5850.55 6868.21 5759.25 65
MVSCRF65.50 6762.71 5867.35 6768.68 6574.27 6766.41 5659.02 7259.11 68
QQQNet64.15 6867.54 5561.89 7271.25 6175.49 6572.33 4962.75 6738.94 80
MVSNet + Gipuma60.80 6957.55 6662.97 6861.50 7170.61 7152.16 6662.94 6656.80 69
MVSNet59.21 7055.77 7061.50 7361.91 6963.44 7746.98 7264.56 6359.16 67
F/T MVSNet+Gipuma57.20 7149.16 7662.57 7161.05 7370.37 7246.27 7552.04 7856.29 71
test_112456.89 7247.96 7762.84 6953.21 8174.83 6639.90 8056.03 7460.47 64
vp_mvsnet56.54 7347.12 7862.82 7061.74 7078.50 6038.70 8255.54 7548.23 74
CCVNet56.43 7456.46 6856.42 7660.81 7569.50 7453.44 6459.47 7138.94 80
CasMVSNet(SR_B)55.71 7558.78 6353.66 7853.67 8062.71 7851.51 6766.05 6144.59 76
SVVNet54.76 7658.44 6452.30 7960.30 7655.64 8054.12 6262.75 6740.96 78
ternet54.76 7658.44 6452.30 7960.30 7655.64 8054.12 6262.75 6740.96 78
unMVSmet52.48 7845.67 8057.03 7556.69 7969.24 7539.82 8151.52 7945.15 75
confMetMVS48.96 7945.93 7950.97 8144.11 8267.03 7641.02 7950.84 8041.78 77
PMVScopyleft60.66 2347.91 8037.76 8254.68 7758.99 7873.94 6846.15 7629.37 8331.10 82
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
Cas-MVS_preliminary43.58 8150.71 7338.83 8334.27 8351.77 8343.48 7857.95 7330.45 83
test_1120copyleft42.17 8240.52 8143.26 8232.91 8447.96 8433.26 8347.78 8148.91 73
FADENet31.23 8331.77 8330.87 8417.26 8559.54 7930.66 8532.88 8215.80 84
CMPMVSbinary66.12 227.25 840.67 8411.64 855.59 8629.32 851.34 860.00 840.00 85
M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces. CVPR 2011
dnet0.00 850.00 850.00 860.00 870.00 860.00 870.00 840.00 85
test_MVS85.16 27
test_robustmvs77.12 4988.60 3293.21 3282.52 3149.55 72
UnsupFinetunedMVSNet61.05 73