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 : 10cm
Method
Info all low-res indoor outdoor lakeside sand box storage room storage room 2 tunnel
DeepPCF-MVS 93.97 1 87.63 1 82.10 1 91.31 1 91.80 1 91.34 1 77.64 1 86.56 1 90.79 1 DeepC-MVS_fast 93.61 2 86.42 2 79.88 3 90.79 2 91.32 2 90.73 2 75.68 3 84.08 4 90.32 3 DeepC-MVS 93.30 3 86.02 3 79.73 4 90.22 3 89.69 6 90.52 3 74.52 5 84.94 2 90.45 2 TAPA-MVS 90.10 7 84.80 4 80.13 2 87.91 6 88.22 8 88.54 9 75.35 4 84.91 3 86.96 14 Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo . ICCV 2019 PCF-MVS 89.48 10 84.51 5 76.96 10 89.54 4 91.30 3 88.76 7 73.35 8 80.56 12 88.57 4 Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction . GCPR 2019 TAPA-MVS(SR) 83.88 6 78.08 7 87.74 8 88.99 7 86.49 16 73.17 10 82.99 6 87.76 10 COLMAPsoft 83.59 7 77.50 9 87.65 9 86.74 12 87.68 11 73.30 9 81.70 9 88.52 5 PLC 91.00 6 83.54 8 78.38 5 86.98 13 85.38 16 87.33 13 73.45 6 83.31 5 88.23 9 Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency . Pacific Graphics 2019 GSE 83.39 9 77.82 8 87.10 12 85.70 15 87.27 14 73.41 7 82.23 7 88.34 8 COLMAP(SR) 83.28 10 76.24 12 87.97 5 87.61 9 87.90 10 72.76 11 79.71 14 88.39 7 ACMH+ 87.92 13 82.46 11 76.63 11 86.35 16 86.95 11 84.52 18 71.68 12 81.58 10 87.56 12 ACMP 89.59 9 82.31 12 75.00 14 87.18 11 84.49 18 88.57 8 68.43 18 81.58 10 88.49 6 Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo . AAAI 2020 LTVRE_ROB 88.41 12 82.23 13 78.11 6 84.97 19 86.24 13 82.83 23 76.93 2 79.30 15 85.84 16 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 89.79 8 82.20 14 75.34 13 86.76 14 84.94 17 87.68 11 68.85 17 81.84 8 87.67 11 Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo . CVPR 2019 3Dnovator 91.36 5 82.19 15 74.09 16 87.59 10 89.83 5 90.10 5 71.54 13 76.63 18 82.84 19 3Dnovator+ 91.43 4 81.91 16 73.11 19 87.78 7 90.06 4 90.17 4 70.40 15 75.82 20 83.12 17 COLMAP_ROB 87.81 14 81.57 17 74.74 15 86.13 17 83.81 19 87.08 15 69.46 16 80.01 13 87.49 13 Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo . ECCV 2016 ACMH 87.59 15 80.55 18 73.69 17 85.11 18 86.17 14 82.81 24 68.19 19 79.20 16 86.36 15 Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo . CVPR 2019 OpenMVS 89.19 11 80.38 19 71.40 20 86.37 15 87.42 10 89.29 6 67.70 20 75.10 21 82.40 21 LPCS 78.54 20 73.62 18 81.82 21 79.59 24 83.19 21 71.25 14 75.99 19 82.69 20 A-TVSNet + Gipuma 77.14 21 67.96 22 83.26 20 80.53 22 86.26 17 65.25 21 70.67 22 82.99 18 PVSNet_0 82.17 16 74.27 22 64.50 23 80.78 22 81.77 21 80.64 27 61.11 23 67.89 24 79.92 22 OpenMVS_ROB 81.14 17 73.28 23 63.11 25 80.06 23 82.70 20 81.62 26 61.11 23 65.12 28 75.87 23 CIDER 72.92 24 62.80 26 79.66 24 80.16 23 83.22 20 60.69 25 64.90 29 75.61 25 Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume . AAAI 2020 R-MVSNet 70.54 25 63.98 24 74.91 28 73.85 30 77.87 31 59.85 26 68.11 23 73.01 26 P-MVSNet 70.42 26 71.20 21 69.90 32 66.73 35 72.02 33 64.84 22 77.55 17 70.95 28 ANet-0.75 67.51 27 58.66 29 73.41 30 70.12 31 79.37 28 50.72 29 66.61 26 70.74 29 AttMVS 65.89 28 60.14 27 69.73 33 64.68 38 75.04 32 58.80 27 61.47 32 69.47 31 ANet 63.85 29 52.65 31 71.32 31 70.12 31 79.37 28 48.53 31 56.77 35 64.48 36 Pnet-new- 63.42 30 42.13 41 77.61 25 77.71 26 79.30 30 41.06 34 43.20 47 75.82 24 cscdsrtA 63.08 31 45.33 38 74.92 27 74.36 27 83.73 19 37.04 41 53.62 38 66.66 34 cscd 61.88 32 43.72 39 73.99 29 74.03 29 81.95 25 34.99 44 52.45 39 65.99 35 Pnet_fast 59.94 33 35.26 53 76.39 26 74.15 28 82.99 22 25.02 51 45.51 44 72.03 27 CPR_FA 58.26 34 56.35 30 59.54 40 56.43 44 55.34 43 50.61 30 62.09 31 66.83 33 MVSNet 56.22 35 46.49 35 62.70 37 57.79 41 67.16 38 38.93 37 54.05 36 63.16 38 A1Net 54.85 36 59.98 28 51.43 43 47.46 53 42.97 52 52.72 28 67.24 25 63.86 37 MVSNet_plusplus 54.10 37 36.12 51 66.09 34 78.36 25 49.55 47 20.40 56 51.84 40 70.36 30 Pnet-blend 53.68 38 37.04 48 64.76 35 68.91 33 70.02 34 24.13 52 49.96 42 55.36 41 Pnet-blend++ 53.68 38 37.04 48 64.76 35 68.91 33 70.02 34 24.13 52 49.96 42 55.36 41 MVSCRF 51.99 40 37.64 45 61.55 38 58.88 39 68.35 36 34.99 44 40.28 50 57.43 40 MVSNet_++ 50.85 41 37.40 46 59.81 39 65.34 37 46.28 48 9.19 57 65.60 27 67.81 32 Snet 49.32 42 36.02 52 58.18 41 66.02 36 50.24 45 27.93 50 44.12 45 58.27 39 cscdsrtB 48.99 43 45.75 36 51.15 44 49.07 51 62.78 39 37.57 38 53.93 37 41.61 48 Pnet-eth 45.62 44 49.15 33 43.27 53 53.31 46 29.96 58 35.03 43 63.26 30 46.53 45 unMVSv1 45.02 45 43.12 40 46.29 50 47.83 52 50.14 46 42.67 32 43.57 46 40.92 49 RMVSNet 44.67 46 50.67 32 40.67 54 50.79 47 41.21 54 42.32 33 59.02 33 30.02 52 MVE 50.73 20 44.67 46 45.57 37 44.07 52 35.90 56 52.30 44 40.89 35 50.26 41 44.00 47 Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment . EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014) hgnet 44.61 48 38.48 43 48.69 45 57.33 42 60.08 40 37.17 39 39.80 52 28.65 54 DPSNet 44.61 48 38.48 43 48.69 45 57.33 42 60.08 40 37.17 39 39.80 52 28.65 54 MVSNet + Gipuma 44.11 50 38.76 42 47.67 48 50.05 50 45.26 50 34.93 46 42.60 48 47.71 43 F/T MVSNet+Gipuma 43.59 51 37.31 47 47.78 47 50.12 48 45.64 49 34.25 47 40.37 49 47.60 44 PMVS 53.92 19 42.74 52 26.50 55 53.57 42 54.03 45 68.05 37 36.35 42 16.65 57 38.64 50 Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis . PAMI (2010) firsttry 41.11 53 36.23 50 44.36 51 43.82 54 43.77 51 32.28 48 40.18 51 45.51 46 example 40.49 54 30.54 54 47.11 49 58.75 40 58.88 42 29.97 49 31.12 54 23.71 57 metmvs_fine 39.41 55 48.46 34 33.38 56 35.53 57 31.67 57 39.28 36 57.64 34 32.95 51 unMVSmet 32.09 56 25.89 56 36.22 55 36.96 55 42.02 53 23.79 54 27.99 55 29.68 53 confMetMVS 26.62 57 23.72 57 28.55 57 28.79 58 33.00 56 21.33 55 26.10 56 23.86 56 CMPMVS 62.92 18 8.44 58 0.22 59 13.92 58 4.26 59 37.50 55 0.45 59 0.00 59 0.00 59 M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces . CVPR 2011 FADENet 1.48 59 1.79 58 1.27 59 2.11 60 1.06 59 2.23 58 1.34 58 0.65 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 50.12 48
