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 : F1 score [%]
Tolerance : 5cm
Method
Info all low-res indoor outdoor lakeside sand box storage room storage room 2 tunnel
DeepPCF-MVS 89.96 1 79.79 1 72.17 1 84.87 1 84.94 1 85.37 1 66.97 1 77.38 1 84.30 1 DeepC-MVS_fast 89.26 2 78.48 2 69.73 3 84.30 2 84.41 2 84.58 2 64.93 3 74.53 4 83.93 3 DeepC-MVS 88.90 3 78.03 3 69.73 3 83.56 3 82.29 7 84.42 3 63.56 5 75.91 3 83.96 2 TAPA-MVS 84.62 6 76.12 4 70.08 2 80.15 7 80.96 8 80.06 9 63.78 4 76.37 2 79.45 14 Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo . ICCV 2019 PCF-MVS 84.11 10 75.39 5 66.57 6 81.27 5 83.29 4 80.62 7 62.19 6 70.95 6 79.89 11 Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction . GCPR 2019 3Dnovator+ 87.14 4 74.49 6 63.75 14 81.65 4 83.81 3 82.92 4 60.57 10 66.93 17 78.22 15 3Dnovator 86.66 5 74.48 7 64.71 9 81.00 6 82.81 5 82.90 5 61.47 8 67.94 13 77.28 16 COLMAPsoft 74.23 8 65.63 8 79.96 9 78.64 11 79.30 11 61.52 7 69.74 9 81.95 4 PLC 84.53 7 73.65 9 66.64 5 78.33 15 75.81 18 78.67 13 61.39 9 71.89 5 80.49 8 Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency . Pacific Graphics 2019 COLMAP(SR) 73.53 10 63.77 13 80.04 8 79.79 9 78.44 14 60.49 11 67.05 16 81.87 5 TAPA-MVS(SR) 73.30 11 64.27 10 79.32 11 82.74 6 75.33 19 59.44 12 69.10 10 79.88 12 GSE 73.02 12 64.14 11 78.95 12 77.59 13 78.43 15 59.39 13 68.88 11 80.83 7 ACMP 84.23 8 72.76 13 62.76 15 79.42 10 76.30 16 80.59 8 54.92 18 70.60 7 81.37 6 Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo . AAAI 2020 LTVRE_ROB 82.13 12 72.22 14 66.24 7 76.21 18 77.23 14 74.60 21 65.16 2 67.32 15 76.79 18 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 ACMM 84.12 9 72.07 15 62.37 17 78.54 14 75.71 19 79.93 10 54.53 19 70.21 8 79.97 10 Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo . CVPR 2019 OpenMVS 83.78 11 71.88 16 61.70 18 78.66 13 78.88 10 81.24 6 56.67 17 66.73 18 75.86 19 COLMAP_ROB 80.39 14 71.80 17 62.64 16 77.90 16 74.64 20 78.97 12 57.39 16 67.89 14 80.10 9 Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo . ECCV 2016 ACMH+ 81.04 13 71.42 18 63.93 12 76.41 17 78.38 12 71.06 24 58.99 14 68.87 12 79.80 13 LPCS 68.73 19 60.59 19 74.16 20 72.17 21 74.57 22 57.78 15 63.39 21 75.74 20 ACMH 80.38 15 68.42 20 59.50 20 74.37 19 77.00 15 69.10 29 54.34 20 64.66 19 77.02 17 Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo . CVPR 2019 OpenMVS_ROB 74.94 16 63.99 21 53.23 22 71.15 23 75.82 17 68.57 30 51.19 22 55.27 24 69.07 24 A-TVSNet + Gipuma 63.67 22 50.37 26 72.54 21 69.76 25 76.21 18 48.72 25 52.03 28 71.64 22 PVSNet 73.20 17 63.55 23 51.85 24 71.35 22 72.10 22 69.61 28 47.57 26 56.13 23 72.33 21 CIDER 63.41 24 52.95 23 70.39 25 71.10 24 72.79 23 49.19 23 56.71 22 67.28 25 Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume . AAAI 2020 P-MVSNet 61.04 25 58.45 21 62.77 30 60.78 35 63.39 34 52.67 21 64.23 20 64.13 26 AttMVS 58.26 26 51.80 25 62.56 32 58.33 36 65.96 32 49.04 24 54.55 25 63.38 27 R-MVSNet 56.72 27 48.74 27 62.04 33 61.37 32 66.93 31 43.74 27 53.73 26 57.82 34 cscdsrtA 56.09 28 36.56 34 69.11 26 67.36 26 77.93 16 28.99 35 44.12 35 62.05 29 Pnet-new- 55.36 29 32.32 38 70.71 24 71.82 23 69.62 27 32.54 32 32.10 46 70.69 23 ANet-0.75 55.32 30 41.08 29 64.81 29 61.58 30 70.76 25 34.14 31 48.01 30 62.08 28 cscd 55.02 31 35.18 35 68.25 27 67.01 28 76.54 17 27.28 37 43.09 36 61.20 31 ANet 53.16 32 38.86 31 62.70 31 61.58 30 70.76 25 34.74 29 42.99 37 55.74 36 Pnet_fast 50.82 33 26.87 44 66.78 28 64.56 29 74.79 20 17.08 51 36.66 43 61.01 32 MVSNet 49.13 34 37.77 32 56.70 34 50.43 39 63.92 33 30.66 33 44.88 33 55.77 35 Pnet-blend 44.30 35 26.07 46 56.45 35 60.87 33 61.19 35 14.59 52 37.56 40 47.29 39 Pnet-blend++ 44.30 35 26.07 46 56.45 35 60.87 33 61.19 35 14.59 52 37.56 40 47.29 39 MVSNet_plusplus 43.14 37 25.17 50 55.12 37 67.36 26 36.20 47 12.78 55 37.55 42 61.80 30 cscdsrtB 42.10 38 37.13 33 45.41 40 43.33 42 57.07 39 29.42 34 44.85 34 35.82 45 MVSCRF 41.80 39 27.23 42 51.52 38 48.62 40 58.89 37 23.51 45 30.95 47 47.03 41 CPR_FA 41.50 40 40.70 30 42.03 43 37.78 47 40.09 43 34.69 30 46.72 32 48.23 38 A1Net 40.64 41 42.66 28 39.29 44 35.99 48 32.67 50 36.23 28 49.09 29 49.23 37 MVSNet_++ 40.21 42 29.12 41 47.60 39 52.26 37 32.34 52 5.74 57 52.50 27 58.20 33 Snet 37.53 43 26.97 43 44.56 42 51.19 38 38.17 44 19.63 50 34.31 45 44.33 42 PMVS 47.18 19 35.10 44 20.52 54 44.82 41 43.36 41 58.67 38 28.88 36 12.16 56 32.42 47 Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis . PAMI (2010) Pnet-eth 31.53 45 34.06 36 29.85 52 41.28 44 17.75 56 20.71 48 47.41 31 30.53 49 F/T MVSNet+Gipuma 31.25 46 25.73 49 34.93 45 34.16 49 33.91 48 23.61 44 27.84 49 36.73 43 MVSNet + Gipuma 31.15 47 26.01 48 34.58 46 33.87 51 33.39 49 23.75 43 28.26 48 36.47 44 DPSNet 30.74 48 25.12 51 34.49 47 40.98 45 47.29 40 24.56 41 25.68 52 15.18 54 hgnet 30.74 48 25.12 51 34.49 47 40.98 45 47.29 40 24.56 41 25.68 52 15.18 54 MVE 39.65 20 30.69 50 31.26 39 30.31 51 23.68 55 36.67 45 26.92 38 35.60 44 30.60 48 Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment . EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014) firsttry 28.73 51 23.72 53 32.06 50 31.74 52 30.80 53 21.15 47 26.30 51 33.65 46 unMVSv1 28.02 52 26.35 45 29.13 53 30.17 53 32.63 51 26.14 40 26.55 50 24.58 50 example 27.92 53 19.45 55 33.56 49 42.92 43 46.29 42 19.84 49 19.07 54 11.46 57 RMVSNet 26.78 54 32.62 37 22.89 54 27.05 54 26.01 54 26.90 39 38.35 39 15.62 53 metmvs_fine 22.70 55 30.59 40 17.43 56 17.52 57 16.63 58 22.55 46 38.64 38 18.14 51 unMVSmet 17.94 56 14.89 56 19.98 55 19.45 56 24.72 55 13.90 54 15.87 55 15.75 52 confMetMVS 13.05 57 10.92 57 14.47 57 14.73 58 17.12 57 10.20 56 11.64 57 11.56 56 CMPMVS 59.16 18 7.91 58 0.09 59 13.12 58 3.14 59 36.21 46 0.19 59 0.00 59 0.00 59 M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces . CVPR 2011 FADENet 0.56 59 0.62 58 0.51 59 0.86 60 0.42 59 0.78 58 0.45 58 0.27 58 dnet 0.00 60 0.00 60 0.00 60 0.00 61 0.00 60 0.00 60 0.00 59 0.00 59 UnsupFinetunedMVSNet 34.16 49
