This table lists the benchmark results for the low-res many-view scenario.
The following metrics are evaluated:
Accuracy [%]: The fraction of the reconstruction which is closer to the ground truth than the evaluation threshold distance (*). Larger is better.Completeness [%]: The fraction of the ground truth which is closer to the reconstruction than the evaluation threshold distance (*). Larger is better.F1 score [%]: The harmonic mean of accuracy and completeness, used to rank methods based on both metrics. Larger is better.Time [s]: The runtime of the method. Smaller is better.
(*) 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 .
training datasets , ground truth and accuracy / completeness visualizations are also available. The visualizations may not work with mobile browsers.
Set : Test
Metric : Time [s]
Method
Info all low-res indoor outdoor lakeside sand box storage room storage room 2 tunnel
ANet 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 ANet-0.75 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 A1Net 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 0.21 1 firsttry 1.03 4 1.03 4 1.03 4 1.03 4 1.03 4 1.03 4 1.03 4 1.03 4 RMVSNet 1.03 4 1.03 4 1.03 4 1.03 4 1.03 4 1.03 4 1.03 4 1.03 4 Pnet-blend++ 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 Pnet-eth 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 Pnet-new- 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 Pnet_fast 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 Pnet-blend 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 9.00 6 CPR_FA 300.30 11 300.30 11 300.30 11 300.30 11 300.30 11 300.30 11 300.30 11 300.30 11 FADENet 300.30 11 300.30 11 300.30 11 300.30 11 300.30 11 300.30 11 300.30 11 300.30 11 unMVSv1 438.86 13 369.40 13 485.16 13 432.90 13 452.10 13 398.70 13 340.10 13 570.50 14 F/T MVSNet+Gipuma 622.80 14 505.00 18 701.33 14 642.00 15 671.00 14 502.00 18 508.00 16 791.00 15 MVSNet + Gipuma 622.80 14 505.00 18 701.33 14 642.00 15 671.00 14 502.00 18 508.00 16 791.00 15 P-MVSNet 626.74 16 503.44 17 708.94 16 634.45 14 681.77 16 490.58 17 516.29 21 810.60 17 PMVS 863.65 2 639.40 17 530.00 21 712.33 17 798.00 22 888.00 22 701.00 23 359.00 14 451.00 13 Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis . PAMI (2010) DPSNet 660.88 18 500.81 14 767.59 18 683.99 17 726.46 17 489.58 14 512.04 18 892.32 19 hgnet 660.88 18 500.81 14 767.59 18 683.99 17 726.46 17 489.58 14 512.04 18 892.32 19 example 660.88 18 500.81 14 767.59 18 683.99 17 726.46 17 489.58 14 512.04 18 892.32 19 CIDER 669.75 21 505.36 20 779.35 21 703.58 20 808.57 20 513.92 20 496.79 15 825.90 18 Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume . AAAI 2020 MVSNet 769.20 22 609.10 22 875.93 22 786.50 21 836.20 21 593.70 21 624.50 22 1005.10 22 ACMH+ 729.97 1 894.58 23 641.24 23 1063.48 23 901.73 23 959.70 23 639.95 22 642.52 23 1329.02 23 ACMH 993.80 3 1055.91 24 719.19 24 1280.39 24 1101.67 24 1267.33 24 717.78 24 720.60 24 1472.17 24 Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo . CVPR 2019 TAPA-SR 1338.82 25 969.96 25 1584.73 25 1327.62 25 1473.80 25 922.00 25 1017.91 25 1952.77 25 ACMP 1246.57 4 1631.36 26 1293.07 27 1856.88 27 1568.97 27 1868.85 27 1273.91 27 1312.23 27 2132.82 27 Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo . AAAI 2020 ACMM 1311.08 5 1661.53 27 1487.23 28 1777.73 26 1530.09 26 1690.76 26 1475.05 28 1499.41 28 2112.35 26 Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo . CVPR 2019 OpenMVS_ROB 1936.51 6 1881.32 28 1172.27 26 2354.03 28 2118.19 28 2241.51 28 1205.90 26 1138.64 26 2702.38 28 3Dnovator 2482.06 10 2364.20 29 1597.00 29 2875.67 30 2548.00 30 2712.00 30 1635.00 29 1559.00 29 3367.00 30 R-MVSNet 2412.80 30 1890.00 31 2761.33 29 2391.00 29 2612.00 29 1879.00 31 1901.00 31 3281.00 29 OpenMVS 2352.88 9 2568.40 31 1727.00 30 3129.33 31 2833.00 31 3048.00 31 1755.00 30 1699.00 30 3507.00 31 COLMAP_ROB 2144.59 8 3311.60 32 2554.50 32 3816.33 32 3382.00 32 3555.00 32 2484.00 32 2625.00 32 4512.00 32 Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo . ECCV 2016 DeepC-MVS_fast 2027.08 7 3835.62 33 3029.20 33 4373.23 33 3841.29 33 4046.30 33 2982.26 35 3076.14 34 5232.11 33 3Dnovator+ 3414.41 13 4057.00 34 3365.00 35 4518.33 34 3909.00 34 4339.00 34 2806.00 33 3924.00 37 5307.00 34 CMPMVS 2623.62 11 4160.94 35 3037.05 34 4910.20 35 4057.20 35 5253.20 36 3109.40 36 2964.70 33 5420.20 35 M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces . CVPR 2011 PLC 2858.82 12 4457.00 36 3560.00 37 5055.00 36 4457.00 36 4671.00 35 3360.00 37 3760.00 35 6037.00 36 Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency . Pacific Graphics 2019 DeepPCF-MVS 3452.42 14 5328.53 37 3926.82 38 6263.00 37 5101.63 37 5379.92 37 4010.62 38 3843.02 36 8307.46 37 MVE 10021.00 18 8749.80 38 3439.50 36 12290.00 41 6275.00 38 16332.00 42 2923.00 34 3956.00 38 14263.00 41 Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment . EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014) PCF-MVS 4335.85 15 9288.52 39 5777.29 39 11629.34 40 9919.03 40 10475.60 39 5496.66 39 6057.92 39 14493.40 42 Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction . GCPR 2019 A-TVSNet + Gipuma 9351.57 40 7502.61 41 10584.21 38 9219.52 39 10119.20 38 7334.37 41 7670.84 41 12413.90 39 DeepC-MVS 6017.64 16 9766.85 41 7401.53 40 11343.73 39 10563.30 41 10709.80 40 7263.16 40 7539.89 40 12758.10 40 TAPA-MVS 6309.84 17 13353.70 42 10496.80 43 15258.30 42 14678.80 42 14057.00 41 10791.00 43 10202.60 43 17039.10 43 Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo . ICCV 2019 LTVRE_ROB 20565.47 19 13815.20 43 7669.50 42 17912.33 43 22283.00 43 21104.00 43 7445.00 42 7894.00 42 10350.00 38 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 GSE 84345.00 44 83012.00 44 85233.67 44 88513.00 44 78675.00 44 82336.00 44 83688.00 44 88513.00 44 LPCS 84345.00 44 83012.00 44 85233.67 44 88513.00 44 78675.00 44 82336.00 44 83688.00 44 88513.00 44 MVSCRF 279999.00 46 99999.00 46 399999.00 46 999999.00 46 99999.00 46 99999.00 46 99999.00 46 99999.00 46 CMAPbl 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 cscdsrtB 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 CMAPSRf 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 cscdsrtA 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 cscd 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 1000000.00 47 dnet 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 Snet 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 AttMVS 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 10000000.00 52 UnsupFinetunedMVSNet 10000000.00 52
