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 bysort bysort bysort bysort bysort bysort bysort bysorted by
DeepPCF-MVS80.84 163.41 153.95 169.71 169.72 371.26 148.73 159.18 168.16 1
DeepC-MVS_fast79.62 262.35 251.75 469.41 269.38 470.78 246.85 256.66 468.08 2
DeepC-MVS79.50 361.99 352.12 368.57 367.79 570.58 345.96 358.27 367.34 3
COLMAP(SR)56.21 944.16 1364.25 763.87 962.07 1441.86 1046.45 1666.80 4
ACMP74.13 656.41 844.60 1264.28 661.46 1365.81 737.07 1852.13 665.56 5
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
3Dnovator+77.84 459.53 447.18 767.76 470.63 167.68 443.34 751.01 964.98 6
COLMAPsoft55.51 1144.97 1162.54 1360.85 1462.46 1242.24 947.71 1264.29 7
TAPA-MVS(SR)55.07 1243.33 1562.90 1067.41 657.81 2039.94 1346.71 1563.49 8
ACMM73.20 855.01 1343.19 1662.89 1159.60 1666.07 635.89 2050.48 1063.01 9
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
3Dnovator76.31 559.36 548.19 566.81 570.04 267.59 544.27 552.10 762.80 10
TAPA-MVS73.13 958.67 652.34 262.89 1163.68 1063.46 1145.85 458.83 261.53 11
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
ACMH+68.96 1352.02 1843.83 1457.49 2059.85 1551.27 3040.04 1247.62 1361.35 12
GSE52.52 1640.81 1860.34 1461.71 1258.23 1938.13 1743.48 1961.07 13
PCF-MVS73.52 757.06 748.10 663.03 966.85 762.36 1343.32 852.89 559.86 14
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
OpenMVScopyleft72.83 1056.18 1045.66 963.19 864.67 865.13 839.60 1451.71 859.76 15
COLMAP_ROBcopyleft66.92 1552.32 1742.45 1758.89 1556.18 2461.09 1538.61 1546.28 1759.41 16
Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
PLCcopyleft70.83 1153.50 1545.79 858.64 1756.31 2360.76 1641.70 1149.88 1158.86 17
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
LPCS50.60 1940.43 1957.38 2158.37 1856.00 2338.14 1642.71 2157.77 18
Pnet-new-42.61 2822.17 3656.24 2258.21 1953.03 2724.08 2820.25 4357.48 19
LTVRE_ROB69.57 1253.52 1445.46 1058.89 1558.76 1760.60 1843.91 647.01 1457.32 20
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
ACMH67.68 1447.97 2138.24 2354.45 2556.88 2050.08 3134.49 2241.99 2256.40 21
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
PVSNet_057.27 1746.73 2334.96 2554.57 2455.50 2551.93 2831.22 2538.71 2556.27 22
OpenMVS_ROBcopyleft64.09 1648.56 2038.68 2155.15 2361.90 1149.65 3236.02 1941.33 2353.89 23
cscdsrtA45.13 2525.99 3057.89 1856.32 2264.81 919.76 3332.23 2952.54 24
cscd44.49 2624.93 3157.53 1956.38 2164.76 1018.64 3531.23 3151.43 25
AttMVS45.85 2439.26 2050.25 3049.36 3151.75 2934.83 2143.70 1849.63 26
P-MVSNet44.46 2737.09 2449.37 3249.27 3249.30 3334.35 2339.83 2449.54 27
CIDER47.21 2238.36 2253.11 2655.37 2655.10 2633.25 2443.46 2048.87 28
Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume. AAAI 2020
A-TVSNet + Gipumacopyleft42.12 2929.09 2750.80 2848.28 3356.53 2229.14 2629.04 3247.59 29
ANet-0.7539.47 3122.61 3550.71 2949.43 2955.75 2418.21 3627.01 3646.96 30
MVSNet_plusplus28.64 3812.93 4939.11 3749.80 2721.55 466.17 5419.69 4445.98 31
ANet39.53 3024.60 3249.48 3149.42 3055.75 2422.08 2927.13 3543.25 32
Pnet_fast37.64 3317.70 3950.93 2749.72 2860.73 1710.59 4924.81 3842.34 33
MVSNet_++25.72 4017.38 4131.27 4033.47 3918.28 512.87 5731.90 3042.06 34
MVSNet38.33 3226.99 2845.89 3340.49 3757.57 2120.52 3133.47 2739.63 35
R-MVSNet36.87 3429.78 2641.61 3642.00 3646.99 3424.73 2734.83 2635.83 36
Pnet-blend++31.51 3615.02 4442.51 3447.41 3446.50 366.99 5223.05 3933.62 37
Pnet-blend31.51 3615.02 4442.51 3447.41 3446.50 366.99 5223.05 3933.62 37
A1Net23.74 4223.95 3323.61 4321.43 4719.62 4819.93 3227.96 3329.78 39
MVSCRF28.32 3918.18 3835.09 3932.16 4144.37 3914.66 3821.69 4228.73 40
Snet23.85 4117.41 4028.15 4132.72 4024.90 4412.22 4522.60 4126.83 41
CPR_FA23.28 4323.62 3423.05 4419.33 4823.00 4519.34 3427.89 3426.81 42
cscdsrtB32.16 3526.80 2935.73 3834.57 3846.99 3420.55 3033.05 2825.63 43
F/T MVSNet+Gipuma17.31 4514.39 4619.26 4616.83 5019.67 4714.03 4014.75 4821.27 44
MVSNet + Gipuma16.91 4614.12 4818.77 4916.39 5219.13 5013.98 4114.27 4920.78 45
firsttry15.88 5112.57 5018.08 5017.80 4917.14 5211.54 4713.61 5019.31 46
MVEpermissive26.22 2016.26 4916.97 4215.79 5111.75 5419.40 4914.45 3919.48 4516.21 47
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
Pnet-eth16.08 5018.30 3714.59 5221.78 447.07 569.66 5026.95 3714.91 48
PMVScopyleft37.38 1921.09 4411.49 5427.48 4224.09 4244.44 3815.98 377.01 5513.92 49
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
unMVSv113.05 5312.03 5313.73 5314.18 5316.18 5312.13 4611.93 5110.84 50
metmvs_fine10.03 5514.31 477.18 576.75 576.98 579.52 5119.10 467.81 51
unMVSmet7.42 565.82 568.48 557.60 5611.29 545.84 555.81 566.56 52
hgnet16.45 4712.31 5119.21 4721.51 4530.31 4212.80 4211.82 525.81 53
DPSNet16.45 4712.31 5119.21 4721.51 4530.31 4212.80 4211.82 525.81 53
confMetMVS4.86 583.39 575.85 585.59 586.61 582.94 563.83 575.35 55
RMVSNet11.09 5415.49 438.17 569.16 5510.05 5512.57 4418.40 475.29 56
example15.69 5210.14 5519.39 4523.73 4330.32 4111.10 489.17 544.13 57
FADENet0.17 590.15 580.18 590.31 600.16 590.19 580.10 580.08 58
dnet0.00 600.00 600.00 600.00 610.00 600.00 600.00 590.00 59
CMPMVSbinary51.72 187.38 570.03 5912.27 542.37 5934.46 400.06 590.00 590.00 59
M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces. CVPR 2011
UnsupFinetunedMVSNet16.83 50