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
DeepPCF-MVS96.37 292.61 189.11 194.94 195.58 195.07 185.62 192.60 194.17 1
DeepC-MVS_fast96.44 191.83 287.74 394.57 295.34 394.64 284.48 491.00 493.72 3
DeepC-MVS96.23 391.57 387.58 494.23 394.49 594.42 383.58 691.59 293.78 2
TAPA-MVS93.98 790.68 488.05 292.44 792.42 1093.52 885.24 290.85 691.36 14
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
TAPA-SR90.57 587.48 592.63 692.64 893.17 1183.91 591.04 392.10 12
PCF-MVS93.45 1090.50 685.08 1394.11 495.54 293.44 1082.27 1187.88 1593.34 4
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
CMAPSRf90.17 785.79 1193.08 592.63 993.87 682.51 1089.07 1392.74 6
PLCcopyleft95.07 489.96 887.00 691.94 1290.65 1792.68 1583.01 890.98 592.48 7
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
CMAPbl89.91 986.37 892.28 991.56 1392.95 1382.53 990.21 792.33 8
ACMH+92.99 1389.84 1085.96 1092.43 892.28 1192.84 1481.89 1290.03 892.17 10
GSE89.80 1186.71 791.87 1390.87 1692.62 1683.49 789.92 992.11 11
ACMM93.85 889.42 1285.42 1292.10 1191.09 1592.98 1281.16 1489.68 1092.23 9
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
ACMP93.49 989.04 1384.30 1692.21 1090.34 1893.47 979.54 1889.07 1392.81 5
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
LTVRE_ROB92.95 1489.00 1486.32 990.78 1991.67 1289.54 2284.87 387.77 1691.13 16
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
ACMH92.88 1588.89 1584.55 1491.78 1491.37 1492.41 1779.88 1689.21 1191.56 13
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
COLMAP_ROBcopyleft93.27 1188.32 1684.41 1590.93 1889.38 1992.11 1879.72 1789.10 1291.28 15
Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
3Dnovator94.51 587.43 1780.90 2091.78 1494.76 494.22 478.96 1982.85 2086.35 19
3Dnovator+94.38 687.07 1880.26 2191.61 1694.34 694.17 578.01 2082.52 2186.34 20
A-TVSNet + Gipumacopyleft86.96 1982.91 1889.65 2088.10 2091.71 1980.23 1585.60 1989.15 17
OpenMVScopyleft93.04 1286.34 2079.08 2291.18 1793.50 793.77 776.56 2181.60 2486.26 21
LPCS85.52 2183.61 1786.79 2184.99 2388.49 2481.34 1385.87 1886.89 18
R-MVSNet81.26 2277.44 2383.81 2482.82 2486.25 2674.52 2380.36 2582.37 22
CIDER80.48 2372.14 2686.05 2287.66 2189.77 2071.75 2472.53 3080.72 24
Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume. AAAI 2020
OpenMVS_ROBcopyleft86.42 1680.25 2471.88 2885.82 2387.52 2289.56 2170.53 2573.24 2980.39 25
ANet-0.7577.73 2574.78 2579.70 2777.01 3084.71 2967.34 2982.21 2377.39 28
P-MVSNet77.69 2681.61 1975.07 3371.91 3577.33 3476.51 2286.71 1775.98 30
CPR_FA74.23 2772.06 2775.68 3174.30 3470.69 4068.60 2775.52 2882.05 23
ANet74.07 2868.26 3077.94 3077.01 3084.71 2964.75 3071.76 3172.09 32
AttMVS72.35 2968.13 3175.16 3270.98 3980.63 3168.09 2868.16 3273.87 31
Pnet-new-70.55 3053.66 4181.81 2681.27 2685.23 2851.15 3856.17 4578.92 27
cscdsrtA69.66 3155.07 3779.38 2880.08 2787.62 2546.56 4363.58 3870.44 33
Pnet_fast68.89 3247.01 4983.47 2581.84 2589.30 2338.61 5155.42 4679.28 26
cscd68.63 3353.70 4078.58 2979.92 2885.96 2744.83 4562.57 4069.87 34
A1Net68.49 3475.64 2463.72 4063.53 4551.39 5068.88 2682.41 2276.23 29
Pnet-blend++64.02 3552.13 4571.95 3475.81 3278.00 3240.66 4863.61 3662.05 38
Pnet-blend64.02 3552.13 4571.95 3475.81 3278.00 3240.66 4863.61 3662.05 38
unMVSv163.91 3762.59 3364.80 3966.39 4268.00 4361.62 3163.55 3960.01 41
MVSNet63.58 3856.25 3568.47 3866.00 4371.12 3748.36 4064.13 3368.29 36
MVSCRF63.13 3952.16 4470.45 3670.18 4075.83 3651.17 3753.16 4965.33 37
RMVSNet62.69 4069.01 2958.48 4869.15 4158.56 4659.72 3278.30 2647.75 49
Snet60.77 4147.18 4869.82 3777.90 2961.73 4539.61 5054.76 4769.82 35
Pnet-eth60.43 4265.87 3256.80 5063.59 4445.39 5155.58 3476.16 2761.43 40
hgnet59.89 4354.16 3863.70 4171.39 3670.87 3852.09 3556.24 4348.85 46
DPSNet59.89 4354.16 3863.70 4171.39 3670.87 3852.09 3556.24 4348.85 46
MVEpermissive62.14 1859.41 4560.38 3458.77 4750.39 5269.00 4256.98 3363.78 3556.91 45
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
MVSNet + Gipuma57.07 4653.08 4259.73 4662.91 4857.42 4848.41 3957.74 4158.88 42
cscdsrtB56.87 4755.95 3657.48 4956.06 5067.85 4447.86 4264.05 3448.53 48
F/T MVSNet+Gipuma55.78 4849.83 4759.76 4563.10 4657.51 4746.34 4453.31 4858.66 43
example55.06 4944.75 5061.93 4371.21 3869.85 4142.82 4746.68 5044.73 50
firsttry54.62 5052.74 4355.88 5154.90 5155.69 4948.02 4157.46 4257.06 44
PMVScopyleft61.03 1949.24 5132.69 5160.27 4462.70 4976.26 3542.99 4622.38 5141.86 51
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
CMPMVSbinary66.06 179.32 520.44 5315.23 526.17 5339.51 520.89 530.00 530.00 53
M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces. CVPR 2011
FADENet4.02 534.30 523.83 536.03 543.71 535.02 523.58 521.74 52
dnet0.00 540.00 540.00 540.00 550.00 540.00 540.00 530.00 53
UnsupFinetunedMVSNet63.10 46