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 bysorted bysort bysort bysort by
DeepPCF-MVS93.97 187.63 182.10 191.31 191.80 191.34 177.64 186.56 190.79 2
DeepC-MVS_fast93.89 286.59 280.39 290.73 291.00 390.77 275.90 384.89 490.41 3
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
3Dnovator+91.43 481.91 1773.11 2187.78 890.06 490.17 370.40 1775.82 2483.12 19
3Dnovator91.36 582.19 1674.09 1887.59 1189.83 590.10 471.54 1476.63 2182.84 22
DeepC-MVS93.07 385.88 379.73 589.98 389.82 690.09 574.93 584.53 590.03 4
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
OpenMVScopyleft89.19 1280.38 2171.40 2386.37 1687.42 1289.29 667.70 2275.10 2582.40 24
PCF-MVS89.48 1184.51 676.96 1189.54 491.30 288.76 773.35 980.56 1488.57 5
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
ACMP89.59 1082.31 1375.00 1687.18 1284.49 2488.57 868.43 2081.58 1288.49 7
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
TAPA-MVS90.10 784.80 580.13 487.91 788.22 888.54 975.35 484.91 386.96 15
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
COLMAP(SR)83.28 1176.24 1487.97 687.61 1187.90 1072.76 1279.71 1688.39 8
COLMAP(base)83.59 877.50 1087.65 1086.74 1587.68 1173.30 1081.70 1188.52 6
ACMM89.79 882.20 1575.34 1586.76 1584.94 2187.68 1168.85 1981.84 1087.67 12
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
PLCcopyleft91.00 683.54 978.38 686.98 1485.38 1987.33 1373.45 783.31 688.23 10
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
GSE83.39 1077.82 987.10 1385.70 1887.27 1473.41 882.23 988.34 9
tm-dncc85.31 480.20 388.72 587.91 987.16 1574.75 685.64 291.10 1
tmmvs77.36 2466.00 2784.93 2187.91 987.16 1564.41 2667.60 3179.73 27
COLMAP_ROBcopyleft87.81 1581.57 1874.74 1786.13 1883.81 2587.08 1769.46 1880.01 1587.49 14
Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
TAPA-MVS(SR)83.88 778.08 887.74 988.99 786.49 1873.17 1182.99 787.76 11
HY-MVS89.66 979.55 2271.86 2284.69 2284.86 2286.32 1966.45 2377.26 2082.88 21
A-TVSNet + Gipumacopyleft77.14 2567.96 2683.26 2480.53 2886.26 2065.25 2470.67 2682.99 20
test_120570.75 3262.35 3376.35 3578.70 3285.03 2155.63 3469.08 2765.34 45
IB-MVS87.33 1781.13 1976.89 1283.96 2386.75 1484.70 2271.43 1582.36 880.43 25
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
ACMH+87.92 1482.46 1276.63 1386.35 1786.95 1384.52 2371.68 1381.58 1287.56 13
test_112671.43 3157.57 3880.67 2885.16 2084.45 2455.16 3559.98 4272.40 34
CasMVSNet(SR_A)63.08 4145.33 4974.92 3674.36 3683.73 2537.04 5553.62 5066.66 43
CIDER72.92 3062.80 3279.66 3080.16 3083.22 2660.69 3164.90 3775.61 32
Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume. AAAI 2020
LPCS78.54 2373.62 2081.82 2679.59 3183.19 2771.25 1675.99 2382.69 23
Pnet_fast59.94 4335.26 7376.39 3474.15 3782.99 2825.02 7445.51 6172.03 35
LTVRE_ROB88.41 1382.23 1478.11 784.97 2086.24 1682.83 2976.93 279.30 1785.84 17
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
ACMH87.59 1680.55 2073.69 1985.11 1986.17 1782.81 3068.19 2179.20 1886.36 16
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
CasMVSNet(base)61.88 4243.72 5273.99 3874.03 3881.95 3134.99 6052.45 5265.99 44
PVSNet_LR70.64 3359.04 3678.38 3277.15 3581.75 3250.11 3967.97 2976.22 29
OpenMVS_ROBcopyleft81.14 2073.28 2963.11 3180.06 2982.70 2681.62 3361.11 2865.12 3675.87 30
PVSNet_082.17 1974.27 2764.50 2980.78 2781.77 2780.64 3461.11 2867.89 3079.92 26
ANet-0.7567.51 3658.66 3773.41 3970.12 4279.37 3550.72 3766.61 3370.74 37
ANet63.85 3952.65 4071.32 4170.12 4279.37 3548.53 4056.77 4664.48 46
Pnet-new-63.42 4042.13 5777.61 3377.71 3479.30 3741.06 4443.20 6775.82 31
mvs_zhu_103064.43 3851.10 4173.32 4071.88 4178.97 3842.87 4159.32 4369.10 40
PVSNet86.66 1873.29 2865.21 2878.67 3180.40 2977.98 3964.04 2766.38 3477.62 28
R-MVSNet70.54 3463.98 3074.91 3773.85 3977.87 4059.85 3268.11 2873.01 33
BP-MVSNet76.76 2668.58 2582.22 2584.51 2377.66 4160.98 3076.18 2284.48 18
Christian Sormann, Patrick Knöbelreiter, Andreas Kuhn, Mattia Rossi, Thomas Pock, Friedrich Fraundorfer: BP-MVSNet: Belief-Propagation-Layers for Multi-View-Stereo. 3DV 2020
AttMVS65.89 3760.14 3469.73 4364.68 4975.04 4258.80 3361.47 4069.47 39
vp_mvsnet45.13 6025.47 7958.23 5553.74 6073.85 4319.74 8031.20 7647.10 63
P-MVSNet70.42 3571.20 2469.90 4266.73 4672.02 4464.84 2577.55 1970.95 36
unsupervisedMVS_cas57.32 4544.73 5165.71 4563.76 5071.21 4536.03 5753.44 5162.15 49
test_112450.28 5531.72 7562.65 5158.29 5470.82 4627.23 7236.21 7558.84 50
Pnet-blend++53.68 5137.04 6664.76 4668.91 4470.02 4724.13 7549.96 5755.36 55
Pnet-blend53.68 5137.04 6664.76 4668.91 4470.02 4724.13 7549.96 5755.36 55
test_mvsss54.02 5040.84 5862.81 4962.37 5169.60 4927.56 7054.13 4756.46 53
MVSCRF51.99 5337.64 6361.55 5258.88 5268.35 5034.99 6040.28 7057.43 52
PMVScopyleft53.92 2242.74 7126.50 7753.57 5754.03 5968.05 5136.35 5616.65 8238.64 69
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
MVSNet56.22 4746.49 4562.70 5057.79 5567.16 5238.93 4854.05 4863.16 48
CasMVSNet(SR_B)48.99 5745.75 4751.15 5949.07 6662.78 5337.57 4953.93 4941.61 67
MVS_test_156.68 4646.21 4663.66 4872.26 4062.56 5431.06 6561.35 4156.15 54
hgnet44.61 6538.48 6148.69 6057.33 5660.08 5537.17 5339.80 7228.65 78
DPSNet44.61 6538.48 6148.69 6057.33 5660.08 5537.17 5339.80 7228.65 78
example40.49 7430.54 7647.11 6458.75 5358.88 5729.97 6631.12 7723.71 82
CPR_FA58.26 4456.35 3959.54 5456.43 5855.34 5850.61 3862.09 3966.83 42
MVEpermissive50.73 2344.67 6245.57 4844.07 7135.90 7952.30 5940.89 4550.26 5544.00 66
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
Snet49.32 5636.02 7258.18 5666.02 4750.24 6027.93 6944.12 6258.27 51
unMVSv145.02 6143.12 5446.29 6547.83 6750.14 6142.67 4243.57 6340.92 68
MVSNet_plusplus54.10 4936.12 7166.09 4478.36 3349.55 6220.40 7951.84 5370.36 38
CCVNet38.86 7631.75 7443.59 7244.13 6949.27 6325.48 7338.01 7437.38 72
MVSNet_++50.85 5437.40 6459.81 5365.34 4846.28 649.19 8465.60 3567.81 41
F/T MVSNet+Gipuma43.59 7037.31 6547.78 6250.12 6345.64 6534.25 6340.37 6947.60 62
MVSNet + Gipuma44.11 6938.76 6047.67 6350.05 6545.26 6634.93 6242.60 6847.71 61
test345.71 5844.95 5046.21 6642.67 7244.26 6739.04 4750.87 5451.69 57
firsttry41.11 7236.23 6844.36 7043.82 7043.77 6832.28 6440.18 7145.51 65
PSD-MVSNet44.35 6843.10 5545.19 6842.64 7343.61 6937.46 5148.74 6049.32 60
SGNet44.49 6743.48 5345.16 6942.20 7543.60 7037.53 5049.43 5949.69 59
TVSNet44.65 6442.86 5645.84 6743.41 7143.54 7135.52 5850.20 5650.56 58
QQQNet40.79 7340.37 5941.07 7442.37 7443.46 7237.29 5243.44 6437.38 72
A1Net54.85 4859.98 3551.43 5847.46 6842.97 7352.72 3667.24 3263.86 47
unMVSmet32.09 7925.89 7836.22 7836.96 7842.02 7423.79 7727.99 7829.68 77
RMVSNet44.67 6250.67 4240.67 7550.79 6241.21 7542.32 4359.02 4430.02 76
SVVNet37.41 7736.23 6838.19 7637.28 7638.77 7629.02 6743.44 6438.53 70
ternet37.41 7736.23 6838.19 7637.28 7638.77 7629.02 6743.44 6438.53 70
CMPMVSbinary62.92 218.44 830.22 8413.92 834.26 8537.50 780.45 860.00 840.00 85
M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces. CVPR 2011
test_1120copyleft27.94 8022.22 8131.76 8020.92 8437.22 7918.54 8325.90 8037.14 74
Cas-MVS_preliminary25.33 8219.98 8228.90 8124.65 8234.26 8018.62 8221.35 8127.78 80
confMetMVS26.62 8123.72 8028.55 8228.79 8133.00 8121.33 7826.10 7923.86 81
metmvs_fine39.41 7548.46 4433.38 7935.53 8031.67 8239.28 4657.64 4532.95 75
Pnet-eth45.62 5949.15 4343.27 7353.31 6129.96 8335.03 5963.26 3846.53 64
test_robustmvs13.65 8422.17 8312.51 8419.49 816.28 83
FADENet1.48 841.79 831.27 852.11 861.06 852.23 851.34 830.65 84
dnet0.00 850.00 850.00 860.00 870.00 860.00 870.00 840.00 85
test_MVS27.53 71
UnsupFinetunedMVSNet50.12 63