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 bysorted bysort bysort bysort bysort bysort bysort by
DeepC-MVS69.38 247.34 138.25 153.40 254.29 456.62 332.95 143.55 249.30 3
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
TAPA-MVS59.36 1041.97 937.54 244.93 1546.78 1646.52 1931.44 343.65 141.48 16
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
DeepPCF-MVS69.58 146.90 237.52 353.15 353.58 557.05 232.83 242.22 348.83 5
DeepC-MVS_fast68.24 346.61 334.94 454.40 155.37 357.55 131.02 438.86 450.28 1
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
3Dnovator64.47 545.24 434.37 552.48 557.34 153.53 530.63 538.11 646.56 9
PCF-MVS61.88 840.12 1333.16 644.75 1649.39 1046.16 2028.63 737.70 738.70 25
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
3Dnovator+66.72 444.84 533.14 752.63 456.23 253.78 429.20 637.07 1047.89 6
OpenMVScopyleft61.03 942.28 732.86 848.56 1051.94 750.82 1027.02 938.70 542.92 13
tm-dncc42.17 831.89 949.03 749.63 850.53 1126.59 1237.19 946.92 8
ACMP63.53 642.50 631.11 1050.08 648.62 1252.78 724.90 1837.33 848.84 4
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
COLMAP(SR)41.20 1029.58 1148.95 848.70 1148.18 1428.26 830.89 1949.98 2
TAPA-MVS(SR)40.72 1229.34 1248.31 1152.32 644.77 2426.63 1132.05 1447.84 7
ACMM61.98 740.80 1129.05 1348.64 946.87 1453.22 623.47 2234.63 1145.82 10
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
ACMH+57.40 1136.48 1728.86 1441.57 2443.89 2237.61 3826.39 1431.33 1743.20 12
PLCcopyleft56.13 1434.63 2428.70 1538.59 3336.78 3844.65 2526.03 1531.36 1634.33 35
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
LTVRE_ROB55.42 1636.88 1628.57 1642.42 2240.72 3048.22 1326.97 1030.17 2138.32 26
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
HY-MVS56.14 1335.94 1828.51 1740.89 2543.99 2138.59 3625.07 1631.96 1540.09 20
AttMVS34.35 2528.45 1838.28 3440.03 3140.06 3023.67 2033.23 1234.75 34
IB-MVS56.42 1234.86 2228.22 1939.29 3043.59 2338.19 3723.59 2132.85 1336.09 32
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
COLMAP(base)37.35 1528.09 2043.52 1842.34 2446.68 1826.55 1329.63 2241.53 15
OpenMVS_ROBcopyleft52.78 1835.39 2027.98 2140.33 2646.81 1536.10 4424.79 1931.17 1838.08 27
tmmvs38.38 1426.53 2246.29 1449.63 850.53 1125.02 1728.03 2338.71 24
CIDER32.55 3425.75 2337.08 3839.65 3240.58 2920.91 2930.59 2031.00 40
Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume. AAAI 2020
COLMAP_ROBcopyleft52.97 1733.06 3125.38 2438.18 3535.81 4045.04 2322.92 2627.84 2533.69 36
Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
LPCS33.63 2824.80 2539.52 2842.29 2539.01 3423.44 2326.15 2837.27 29
PVSNet50.76 1932.44 3524.74 2637.57 3738.43 3536.66 4323.03 2526.46 2737.62 28
PVSNet_043.31 2033.35 2923.83 2739.69 2741.22 2838.75 3521.08 2826.59 2639.10 23
ACMH55.70 1532.91 3223.69 2839.06 3141.22 2836.76 4221.84 2725.54 2939.18 22
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
GSE34.79 2323.63 2942.23 2346.31 1739.18 3323.20 2424.07 3241.20 17
PVSNet_LR34.22 2621.35 3042.80 2142.24 2645.98 2114.83 3927.87 2440.18 19
P-MVSNet29.35 3820.98 3134.93 4034.55 4436.99 4020.06 3021.91 3933.25 38
BP-MVSNet29.92 3720.53 3236.17 3937.89 3633.68 4917.55 3223.52 3636.95 30
Christian Sormann, Patrick Knöbelreiter, Andreas Kuhn, Mattia Rossi, Thomas Pock, Friedrich Fraundorfer: BP-MVSNet: Belief-Propagation-Layers for Multi-View-Stereo. 3DV 2020
CasMVSNet(SR_B)23.80 4619.79 3326.47 5024.88 5037.51 3914.94 3824.65 3017.01 52
mvs_zhu_103033.71 2719.38 3443.27 1942.20 2747.82 1615.18 3723.58 3439.77 21
CasMVSNet(SR_A)35.87 1919.38 3446.87 1246.17 1851.96 914.32 4124.45 3142.47 14
test_120528.37 4019.29 3634.43 4134.50 4541.99 2815.81 3422.78 3726.79 45
MVSNet27.45 4218.54 3733.39 4329.32 4747.13 1713.18 4323.89 3323.72 46
CasMVSNet(base)35.32 2118.44 3846.56 1346.00 2052.64 813.32 4223.57 3541.04 18
R-MVSNet23.84 4518.30 3927.54 4727.82 4833.79 4814.57 4022.03 3821.00 50
test_112633.11 3018.21 4043.04 2047.43 1345.33 2215.33 3621.10 4036.35 31
A-TVSNet + Gipumacopyleft26.68 4316.93 4133.18 4431.29 4639.82 3217.28 3316.57 4728.43 43
ANet30.02 3616.73 4238.88 3239.32 3344.06 2615.61 3517.85 4333.26 37
Pnet-new-32.65 3316.09 4343.68 1746.15 1939.98 3118.51 3113.67 5544.91 11
CPR_FA14.43 5714.90 4414.12 6511.38 6214.82 6811.91 4517.89 4216.15 58
A1Net14.51 5614.40 4514.59 5813.30 5912.51 7011.95 4416.85 4517.96 51
ANet-0.7529.06 3913.44 4639.47 2939.32 3344.06 2610.78 4716.11 4935.02 33
unsupervisedMVS_cas21.41 4913.42 4726.73 4925.53 4931.05 5210.34 4816.50 4823.60 47
Pnet_fast27.80 4112.25 4838.17 3637.73 3747.91 157.47 6217.03 4428.88 42
MVSCRF18.49 5112.01 4922.81 5220.51 5532.84 509.61 4914.41 5415.08 62
Snet16.01 5311.97 5018.70 5422.09 5317.12 598.46 5515.48 5116.90 55
test313.22 5911.25 5114.54 5911.46 6117.02 609.59 5012.91 5715.13 60
MVS_test_115.08 5510.97 5217.82 5523.84 5116.69 615.32 7016.61 4612.93 63
TVSNet13.02 6010.86 5314.46 6111.60 6016.67 628.79 5312.93 5615.10 61
Pnet-eth8.45 7510.49 547.10 789.98 722.92 835.14 7115.84 508.40 72
MVSNet_++16.58 5210.34 5520.75 5322.25 5211.60 721.64 8319.05 4128.39 44
SGNet11.73 649.85 5612.98 6610.55 6815.85 648.23 5711.47 5812.53 64
MVEpermissive17.77 239.03 749.66 578.62 756.03 7811.40 748.30 5611.02 598.42 71
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
QQQNet12.59 619.43 5814.70 5710.88 6616.24 638.64 5410.22 6116.97 53
PSD-MVSNet11.22 659.34 5912.48 6710.37 6915.20 677.81 6110.86 6011.87 65
SVVNet12.22 629.22 6014.22 6310.32 7015.62 658.22 5810.22 6116.70 56
ternet12.22 629.22 6014.22 6310.32 7015.62 658.22 5810.22 6116.70 56
Pnet-blend++21.89 479.21 6230.34 4534.58 4234.44 453.76 7614.67 5222.00 48
Pnet-blend21.89 479.21 6230.34 4534.58 4234.44 453.76 7614.67 5222.00 48
test_112423.88 448.92 6433.86 4235.80 4136.79 4110.99 466.84 7228.99 41
F/T MVSNet+Gipuma9.78 688.13 6510.88 739.23 7311.92 718.81 517.45 7011.49 66
MVSNet + Gipuma9.47 717.91 6610.52 748.91 7511.46 738.80 527.01 7111.18 69
CCVNet13.25 587.81 6716.88 5613.52 5720.14 557.96 607.67 6916.97 53
RMVSNet4.93 807.73 683.06 823.93 803.22 825.95 699.52 652.04 81
firsttry9.57 697.40 6911.02 7211.05 6310.54 756.97 657.82 6811.49 66
metmvs_fine4.94 797.02 703.56 813.18 823.86 804.62 739.43 663.63 75
MVSNet_plusplus19.10 506.64 7127.41 4836.63 3914.19 693.12 7910.15 6431.41 39
hgnet9.31 726.52 7211.17 7010.91 6420.13 567.15 635.89 732.47 79
DPSNet9.31 726.52 7211.17 7010.91 6420.13 567.15 635.89 732.47 79
unMVSv16.85 766.06 747.37 777.56 769.33 766.35 685.76 755.22 74
test_mvsss11.00 666.00 7514.32 6213.56 5619.16 584.03 757.98 6710.25 70
example9.54 705.97 7611.92 6813.40 5820.72 546.89 665.05 761.63 82
PMVScopyleft28.69 2210.60 674.70 7714.53 6010.82 6730.05 536.66 672.74 782.71 77
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
test_1120copyleft6.40 784.10 787.94 766.69 775.86 784.95 723.26 7711.26 68
vp_mvsnet15.26 542.83 7923.55 5121.00 5434.21 473.45 782.22 7915.43 59
unMVSmet3.60 822.44 804.38 803.82 816.35 772.69 812.18 802.97 76
Cas-MVS_preliminary4.12 811.79 815.67 794.49 794.19 792.85 800.74 828.32 73
confMetMVS2.34 831.33 823.02 833.03 833.47 811.14 841.52 812.56 78
FADENet0.06 840.04 830.07 850.12 860.08 850.05 850.03 830.02 84
CMPMVSbinary42.80 216.77 770.01 8411.27 691.93 8531.86 510.03 860.00 840.00 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_MVS4.13 74
test_robustmvs1.65 842.89 841.35 842.69 810.72 83
UnsupFinetunedMVSNet9.23 73