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
DeepC-MVS_fast98.69 196.77 195.30 297.74 198.24 297.85 392.74 397.86 297.13 2
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
DeepPCF-MVS98.18 396.68 295.10 397.74 198.30 197.78 592.41 497.79 397.12 3
tm-dncc96.59 395.98 197.00 696.82 996.51 2194.07 197.88 197.68 1
DeepC-MVS98.35 296.09 494.20 1297.34 397.94 497.39 1190.85 1597.56 496.71 5
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
COLMAP(SR)96.07 594.39 997.19 596.86 897.92 291.54 1197.24 996.79 4
TAPA-MVS(SR)95.82 694.80 496.49 1096.19 1497.58 1092.04 597.56 495.71 13
PCF-MVS97.08 1495.68 793.36 1697.23 498.16 396.93 1591.43 1295.29 2296.61 6
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
ACMM97.58 595.68 794.49 796.48 1196.12 1597.20 1492.04 596.94 1296.10 10
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
ACMH+97.24 1095.61 993.77 1596.83 796.69 1297.71 790.70 1696.85 1396.09 11
TAPA-MVS97.07 1595.48 1094.55 596.10 1395.68 1897.30 1291.73 997.37 795.32 16
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
COLMAP(base)95.42 1194.03 1496.35 1295.96 1696.92 1691.10 1496.97 1096.16 9
PLCcopyleft97.94 495.35 1294.28 1196.06 1495.26 2296.59 1891.61 1096.96 1196.34 8
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
ACMH97.28 895.29 1393.25 1796.65 896.32 1397.93 189.82 1996.68 1695.69 14
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
ACMP97.20 1195.11 1493.01 2196.52 995.68 1897.26 1389.74 2096.27 1896.61 6
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
LTVRE_ROB97.16 1295.10 1594.13 1395.75 1595.69 1795.73 2492.87 295.39 2195.83 12
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
GSE95.00 1694.33 1095.45 1894.81 2396.53 2091.98 796.68 1695.02 18
A-TVSNet + Gipumacopyleft94.96 1794.54 695.24 2095.31 2196.58 1991.82 897.27 893.83 19
BP-MVSNet94.57 1893.20 1895.49 1696.77 1194.34 3389.67 2196.74 1495.35 15
Christian Sormann, Patrick Knöbelreiter, Andreas Kuhn, Mattia Rossi, Thomas Pock, Friedrich Fraundorfer: BP-MVSNet: Belief-Propagation-Layers for Multi-View-Stereo. 3DV 2020
COLMAP_ROBcopyleft97.56 694.43 1993.09 1995.33 1994.55 2496.37 2389.45 2296.72 1595.07 17
Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
IB-MVS95.67 1894.34 2094.41 894.29 2395.45 2097.77 691.37 1397.44 689.67 28
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
HY-MVS97.30 793.59 2193.02 2093.97 2492.85 2996.87 1790.03 1896.00 1992.19 21
3Dnovator97.25 992.47 2287.97 3095.47 1797.93 597.85 386.02 2689.92 3290.64 24
3Dnovator+97.12 1392.17 2387.57 3195.24 2097.49 697.68 885.24 2989.91 3390.54 25
LPCS92.07 2492.93 2291.49 3090.32 3593.29 3590.66 1795.20 2390.87 23
OpenMVScopyleft96.50 1691.77 2586.65 3495.19 2297.38 797.66 984.60 3288.69 3890.53 26
R-MVSNet90.93 2689.80 2491.68 2990.81 3494.15 3487.67 2391.93 2890.07 27
tmmvs90.82 2786.27 3593.86 2596.82 996.51 2183.91 3588.63 3988.24 31
PVSNet_094.43 1990.46 2885.77 3693.59 2693.82 2695.33 2784.62 3186.92 4191.61 22
test_112689.66 2988.06 2990.73 3393.40 2795.29 2884.99 3091.13 3083.50 39
PVSNet_LR89.56 3087.11 3291.19 3290.87 3394.45 3283.35 3790.86 3188.25 30
test_120589.53 3189.08 2789.83 3692.63 3095.71 2585.34 2892.82 2781.15 47
CPR_FA89.35 3288.37 2890.00 3589.38 3887.69 4687.29 2589.46 3692.93 20
PVSNet96.02 1789.22 3384.53 4092.35 2793.29 2894.60 3185.85 2783.20 4589.17 29
CIDER88.55 3483.13 4292.16 2893.95 2595.61 2683.79 3682.47 4886.92 33
Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume. AAAI 2020
ANet-0.7587.89 3589.58 2586.76 3884.37 5290.29 3984.51 3394.65 2485.61 37
OpenMVS_ROBcopyleft92.34 2087.46 3681.75 4391.27 3192.55 3195.21 2981.21 4182.29 5086.05 35
test_mvsss85.35 3779.90 4488.98 3790.30 3689.87 4268.09 5891.71 2986.77 34
P-MVSNet85.35 3791.57 2381.21 5078.76 6082.87 5687.56 2495.58 2082.00 44
ANet85.16 3985.20 3885.14 4284.37 5290.29 3982.00 3988.39 4080.78 49
unMVSv183.89 4084.37 4183.58 4584.00 5685.63 4983.17 3885.56 4281.11 48
metmvs_fine83.50 4184.90 3982.56 4884.72 4781.80 5981.09 4288.72 3781.17 46
unsupervisedMVS_cas83.12 4278.75 4586.04 4085.85 4289.90 4175.64 4681.86 5282.36 42
A1Net82.90 4389.39 2678.57 5985.19 4463.18 7484.36 3494.42 2587.34 32
mvs_zhu_103082.18 4478.09 5084.91 4382.35 5788.99 4371.71 5484.46 4483.39 40
RMVSNet80.62 4586.71 3376.57 6184.14 5478.09 6180.22 4393.20 2667.48 71
MVS_test_180.50 4677.21 5282.70 4786.63 4082.92 5564.61 6389.81 3478.53 51
Pnet_fast80.38 4765.52 7290.29 3490.03 3794.91 3060.54 7270.51 7585.93 36
AttMVS80.32 4878.27 4881.68 4978.98 5986.16 4879.76 4476.79 6579.89 50
Pnet-new-79.46 4969.54 6586.08 3984.76 4690.31 3865.13 6273.95 7083.16 41
CasMVSNet(SR_A)79.00 5069.56 6485.29 4185.95 4192.82 3662.08 6977.05 6177.11 52
hgnet78.82 5176.20 5480.57 5484.50 4882.32 5775.56 4776.84 6274.88 59
DPSNet78.82 5176.20 5480.57 5484.50 4882.32 5775.56 4776.84 6274.88 59
Pnet-blend++78.13 5373.62 5981.13 5184.46 5088.20 4466.50 5980.74 5470.74 69
Pnet-blend78.13 5373.62 5981.13 5184.46 5088.20 4466.50 5980.74 5470.74 69
CasMVSNet(base)78.10 5568.51 6884.50 4485.47 4391.56 3759.79 7577.22 6076.47 53
Pnet-eth77.75 5685.52 3772.57 6775.57 6365.89 7281.44 4089.61 3576.26 54
MVSCRF77.21 5774.57 5778.98 5879.29 5883.87 5375.01 4974.12 6973.78 64
MVEpermissive76.82 2176.63 5877.52 5176.03 6470.40 7485.49 5175.68 4579.36 5672.20 66
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
MVSNet_plusplus76.50 5966.38 7083.25 4691.25 3273.79 6653.62 8179.14 5784.70 38
example74.48 6065.59 7180.41 5684.10 5583.16 5462.82 6868.36 7773.98 63
Snet74.39 6164.28 7481.13 5188.04 3973.18 6960.15 7368.41 7682.18 43
MVSNet73.65 6270.01 6376.07 6376.39 6177.59 6261.93 7078.09 5974.23 62
PSD-MVSNet73.38 6378.51 4669.96 7072.98 6862.49 7674.33 5082.69 4774.41 61
SGNet73.32 6478.38 4769.94 7172.09 6962.62 7573.95 5182.81 4675.11 58
test373.10 6578.15 4969.73 7271.16 7162.14 7873.89 5282.42 4975.89 56
MVSNet + Gipuma72.61 6671.20 6173.54 6575.39 6473.46 6865.60 6176.81 6471.78 67
firsttry72.49 6775.56 5670.44 6968.80 7770.25 7069.31 5681.82 5372.28 65
TVSNet72.34 6876.78 5369.38 7370.64 7361.58 8071.28 5582.27 5175.92 55
MVSNet_++71.81 6959.04 8080.33 5784.86 4574.23 6533.06 8485.02 4381.88 45
test_112471.63 7063.97 7576.73 6069.39 7685.55 5056.73 7671.21 7275.26 57
CasMVSNet(SR_B)70.07 7171.18 6269.33 7469.73 7577.06 6364.06 6478.31 5861.20 75
F/T MVSNet+Gipuma69.80 7264.46 7373.36 6675.05 6573.57 6760.74 7168.18 7871.46 68
vp_mvsnet69.76 7360.05 7976.24 6276.16 6287.57 4748.96 8371.14 7364.99 73
QQQNet67.18 7474.04 5862.60 7771.02 7262.30 7773.86 5374.22 6654.48 79
unMVSmet66.94 7560.67 7871.11 6871.91 7079.70 6054.62 7866.73 7961.74 74
CCVNet64.24 7667.70 6961.94 7865.09 7866.24 7163.67 6571.73 7154.48 79
confMetMVS63.51 7760.95 7765.21 7660.95 8276.48 6455.53 7766.38 8058.21 76
SVVNet62.40 7868.72 6658.19 8063.36 7954.86 8163.22 6674.22 6656.34 77
ternet62.40 7868.72 6658.19 8063.36 7954.86 8163.22 6674.22 6656.34 77
test_1120copyleft58.74 8057.03 8159.88 7949.42 8464.79 7349.65 8264.41 8165.45 72
PMVScopyleft70.75 2258.19 8143.22 8268.18 7573.06 6784.09 5254.23 7932.21 8247.37 81
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
Cas-MVS_preliminary56.71 8262.57 7652.80 8250.07 8361.96 7954.15 8071.00 7446.38 82
FADENet18.58 8315.51 8320.63 8421.45 8532.74 8515.09 8515.93 837.68 84
CMPMVSbinary69.68 2311.01 840.97 8417.70 8510.09 8643.01 841.93 860.00 840.00 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_MVS69.14 57
test_robustmvs44.81 8363.28 8145.49 8359.95 7425.66 83
UnsupFinetunedMVSNet75.05 65