This table lists the benchmark results for the high-res multi-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 Infoallhigh-res
multi-view
indooroutdoorbotani.boulde.bridgedoorexhibi.lectur.living.loungeobserv.old co.statueterrac.
sort bysort bysort bysort bysort bysort bysort bysorted bysort bysort bysort bysort bysort bysort bysort bysort by
DeepC-MVS79.81 287.08 286.88 287.69 391.16 772.32 490.31 593.94 177.12 482.82 394.23 172.13 297.09 184.83 295.37 293.65 4
3Dnovator+77.84 485.48 484.47 488.51 191.08 873.49 193.18 193.78 280.79 176.66 1193.37 360.40 996.75 377.20 593.73 495.29 1
DeepPCF-MVS80.84 188.10 188.56 186.73 492.24 469.03 789.57 993.39 377.53 289.79 194.12 278.98 196.58 485.66 195.72 194.58 2
DeepC-MVS_fast79.65 386.91 386.62 387.76 293.52 172.37 391.26 293.04 476.62 584.22 293.36 471.44 396.76 280.82 395.33 394.16 3
CLD-MVS82.31 781.65 784.29 1088.47 1767.73 1485.81 2092.35 575.78 778.33 686.58 2264.01 694.35 1276.05 687.48 1190.79 10
Zhaoxin Li, Wangmeng Zuo, Zhaoqi Wang, Lei Zhang: Confidence-based Large-scale Dense Multi-view Stereo. IEEE Transaction on Image Processing, 2020
3Dnovator76.31 583.38 682.31 686.59 587.94 1872.94 290.64 392.14 677.21 375.47 1392.83 558.56 1094.72 873.24 992.71 592.13 8
ACMM73.20 880.78 1179.84 1183.58 1389.31 1268.37 1189.99 691.60 770.28 1177.25 989.66 1453.37 1693.53 1574.24 782.85 1788.85 13
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
TAPA-MVS73.13 979.15 1577.94 1582.79 1789.59 1062.99 2288.16 1991.51 865.77 1877.14 1091.09 960.91 893.21 1750.26 2687.05 1292.17 7
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
ACMP74.13 681.51 1080.57 1084.36 989.42 1168.69 1089.97 791.50 974.46 975.04 1490.41 1153.82 1494.54 977.56 482.91 1689.86 11
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
PCF-MVS73.52 780.38 1378.84 1385.01 787.71 1968.99 883.65 2291.46 1063.00 2377.77 890.28 1266.10 595.09 761.40 1588.22 1090.94 9
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
OpenMVScopyleft72.83 1079.77 1478.33 1484.09 1185.17 2369.91 690.57 490.97 1166.70 1772.17 1591.91 754.70 1393.96 1361.81 1490.95 888.41 15
MAR-MVS81.84 980.70 985.27 691.32 671.53 589.82 890.92 1269.77 1478.50 586.21 2362.36 794.52 1165.36 1192.05 689.77 12
Zhenyu Xu, Yiguang Liu, Xuelei Shi, Ying Wang, Yunan Zheng: MARMVS: Matching Ambiguity Reduced Multiple View Stereo for Efficient Large Scale Scene Reconstruction. CVPR 2020
AP-MVS82.00 881.11 884.69 887.65 2067.81 1288.76 1390.84 1376.33 678.25 791.60 857.36 1193.38 1673.93 885.26 1492.88 5
MG-MVS83.41 583.45 583.28 1592.74 362.28 2388.17 1889.50 1475.22 881.49 492.74 666.75 495.11 672.85 1091.58 792.45 6
PLCcopyleft70.83 1178.05 1676.37 1683.08 1691.88 567.80 1388.19 1789.46 1564.33 2069.87 1888.38 1653.66 1593.58 1458.86 1682.73 1887.86 17
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
LTVRE_ROB69.57 1276.25 1974.54 1881.41 2188.60 1564.38 2179.24 2689.12 1670.76 1069.79 1987.86 1749.09 1993.20 1856.21 2080.16 2086.65 24
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
F-COLMAP76.38 1874.33 1982.50 1889.28 1366.95 1688.41 1589.03 1764.05 2266.83 2388.61 1546.78 2092.89 2057.48 1878.55 2187.67 18
ACMH67.68 1475.89 2173.93 2181.77 2088.71 1466.61 1788.62 1489.01 1869.81 1366.78 2486.70 2141.95 2591.51 2255.64 2178.14 2287.17 22
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
AdaColmapbinary80.58 1279.42 1284.06 1293.09 268.91 989.36 1088.97 1969.27 1575.70 1289.69 1357.20 1295.77 563.06 1288.41 987.50 20
ACMH+68.96 1376.01 2074.01 2082.03 1988.60 1565.31 2088.86 1287.55 2070.25 1267.75 2187.47 1941.27 2693.19 1958.37 1775.94 2687.60 19
LS3D76.95 1774.82 1783.37 1490.45 967.36 1589.15 1186.94 2161.87 2669.52 2090.61 1051.71 1794.53 1046.38 2986.71 1388.21 16
MSDG73.36 2270.99 2380.49 2384.51 2465.80 1880.71 2386.13 2265.70 1965.46 2583.74 2544.60 2190.91 2451.13 2576.89 2384.74 26
COLMAP_ROBcopyleft66.92 1573.01 2470.41 2580.81 2287.13 2165.63 1988.30 1684.19 2362.96 2463.80 2787.69 1838.04 2792.56 2146.66 2874.91 2784.24 27
Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
CMPMVSbinary51.72 1970.19 2768.16 2776.28 2773.15 3057.55 2779.47 2583.92 2448.02 3156.48 3084.81 2443.13 2386.42 2662.67 1381.81 1984.89 25
M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces. CVPR 2011
ITE_SJBPF78.22 2581.77 2660.57 2583.30 2569.25 1667.54 2287.20 2036.33 2987.28 2554.34 2374.62 2886.80 23
OpenMVS_ROBcopyleft64.09 1770.56 2668.19 2677.65 2680.26 2759.41 2685.01 2182.96 2658.76 2865.43 2682.33 2637.63 2891.23 2345.34 3076.03 2582.32 28
ANet_high50.57 3146.10 3163.99 3148.67 3539.13 3570.99 2980.85 2761.39 2731.18 3257.70 3217.02 3173.65 3331.22 3215.89 3879.18 29
PVSNet64.34 1672.08 2570.87 2475.69 2886.21 2256.44 2974.37 2880.73 2862.06 2570.17 1782.23 2742.86 2483.31 2854.77 2284.45 1587.32 21
Gipumacopyleft45.18 3241.86 3255.16 3577.03 2951.52 3132.50 3680.52 2932.46 3427.12 3335.02 379.52 3475.50 3222.31 3560.21 3138.45 37
S. Galliani, K. Lasinger, K. Schindler: Massively Parallel Multiview Stereopsis by Surface Normal Diffusion. ICCV 2015
patch_net73.12 2371.33 2278.49 2483.18 2560.85 2479.63 2478.57 3064.13 2171.73 1679.81 2951.20 1885.97 2757.40 1976.36 2488.66 14
LF4IMVS64.02 2862.19 2869.50 2970.90 3153.29 3076.13 2777.18 3152.65 3058.59 2880.98 2823.55 3076.52 3053.06 2466.66 2978.68 30
PVSNet_057.27 1861.67 2959.27 2968.85 3079.61 2857.44 2868.01 3173.44 3255.93 2958.54 2970.41 3144.58 2277.55 2947.01 2735.91 3371.55 31
FPMVS53.68 3051.64 3059.81 3265.08 3251.03 3269.48 3069.58 3341.46 3240.67 3172.32 3016.46 3270.00 3524.24 3465.42 3058.40 33
wuykxyi23d39.76 3433.18 3559.51 3346.98 3644.01 3357.70 3467.74 3424.13 3613.98 3834.33 381.27 3871.33 3434.23 3118.23 3663.18 32
PMVScopyleft37.38 2044.16 3340.28 3355.82 3440.82 3842.54 3465.12 3263.99 3534.43 3324.48 3457.12 333.92 3576.17 3117.10 3755.52 3248.75 34
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
PNet_i23d38.26 3535.42 3446.79 3658.74 3335.48 3659.65 3351.25 3632.45 3523.44 3647.53 352.04 3758.96 3625.60 3318.09 3745.92 36
MVEpermissive26.22 2130.37 3625.89 3643.81 3744.55 3735.46 3728.87 3739.07 3718.20 3718.58 3740.18 362.68 3647.37 3717.07 3823.78 3548.60 35
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
DeepMVS_CXcopyleft27.40 3840.17 3926.90 3924.59 3817.44 3823.95 3548.61 349.77 3326.48 3818.06 3624.47 3428.83 38
wuyk23d16.82 3715.94 3719.46 3958.74 3331.45 3839.22 353.74 396.84 396.04 392.70 391.27 3824.29 3910.54 3914.40 392.63 39