Low-res many-view results

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.

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	all	low-res many-view	indoor	outdoor	delivery area	electro	forest	playground	terrains

DeepPCF-MVS	66.49 1	46.32 3	51.86 1	42.63 4	31.46 3	45.24 4	52.14 8	30.51 3	72.26 1

DeepC-MVS	66.32 2	46.73 2	50.79 2	44.02 2	32.46 2	46.47 2	53.30 7	32.29 1	69.12 3
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
DeepC-MVS_fast	65.08 3	47.08 1	50.48 3	44.81 1	33.22 1	47.35 1	55.12 3	31.95 2	67.74 5
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
3Dnovator+	62.63 4	44.72 5	47.81 5	42.67 3	27.63 11	45.47 3	55.38 2	27.16 11	67.98 4

ACMP	61.42 5	42.42 7	46.75 7	39.54 10	30.27 4	45.02 5	45.28 17	28.32 6	63.23 11
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
3Dnovator	60.86 6	44.93 4	48.84 4	42.32 5	28.50 9	44.85 7	54.53 5	27.58 9	69.17 2

ACMM	60.30 7	41.38 10	42.72 15	40.48 8	22.52 20	45.00 6	46.51 14	29.94 4	62.93 12
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
PCF-MVS	59.98 8	40.92 11	46.33 8	37.31 14	29.32 5	37.64 15	46.82 13	27.49 10	63.34 10
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
OpenMVS	57.13 9	42.36 8	47.69 6	38.82 12	28.51 8	42.20 10	49.31 11	24.94 14	66.86 7

TAPA-MVS	54.74 10	38.87 15	45.22 9	34.64 18	22.75 18	34.37 18	45.53 16	24.02 15	67.70 6
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
IB-MVS	54.11 11	43.08 6	44.40 11	42.19 6	24.33 15	44.51 8	55.78 1	26.30 13	64.46 9
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
ACMH+	53.71 12	39.30 14	43.58 12	36.45 15	27.46 12	40.49 13	41.20 20	27.67 7	59.70 14

ACMH	52.42 13	37.28 18	40.03 20	35.45 17	22.59 19	36.89 16	41.81 19	27.65 8	57.47 20
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
PLC	52.09 14	33.56 21	40.06 19	29.22 22	21.22 23	30.18 22	35.57 25	21.92 21	58.91 16
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
COLMAP_ROB	46.52 15	32.32 24	37.61 24	28.79 24	16.26 25	28.69 23	39.99 23	17.68 27	58.96 15
Johannes L. SchÃ¶nberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
LTVRE_ROB	44.17 16	36.66 19	42.98 14	32.44 20	19.43 24	36.54 17	40.06 22	20.72 24	66.52 8
Andreas Kuhn, Heiko HirschmÃ¼ller, Daniel Scharstein, Helmut Mayer: A TV Prior for High-Quality Scalable Multi-View Stereo Reconstruction. International Journal of Computer Vision 2016
CMPMVS	37.70 17	7.71 34	19.19 30	0.05 34	0.00 34	0.00 34	0.16 34	0.00 34	38.37 29
M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces. CVPR 2011
PMVS	27.84 18	12.34 29	24.68 28	4.10 31	1.67 33	2.37 30	7.55 30	2.39 31	47.69 27
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
MVE	12.28 19	9.06 31	14.41 34	5.49 30	2.18 32	7.44 29	4.17 33	4.86 29	26.64 34
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
unsupervisedMVS_cas		21.46 28	21.49 29	21.45 28	12.65 28	17.48 27	30.63 28	16.24 28	30.33 31

BP-MVSNet		38.01 17	41.54 17	35.66 16	25.02 14	34.14 19	49.43 10	23.40 16	58.05 18
Christian Sormann, Patrick KnÃ¶belreiter, Andreas Kuhn, Mattia Rossi, Thomas Pock, Friedrich Fraundorfer: BP-MVSNet: Belief-Propagation-Layers for Multi-View-Stereo. 3DV 2020
CasMVSNet(SR_A)		31.87 25	38.46 22	27.48 27	22.44 21	16.75 28	43.19 18	22.50 18	54.48 23

CasMVSNet(SR_B)		39.59 13	38.46 22	40.35 9	22.44 21	43.48 9	55.07 4	22.50 18	54.48 23

TAPA-MVS(SR)		42.33 9	44.67 10	40.77 7	27.98 10	41.86 11	50.64 9	29.80 5	61.36 13

CasMVSNet(base)		38.28 16	37.24 25	38.97 11	23.48 16	40.53 12	54.14 6	22.23 20	51.01 26

GSE		33.30 22	41.78 16	27.64 26	29.07 6	27.59 24	35.00 26	20.32 25	54.50 22

LPCS		33.27 23	39.58 21	29.06 23	27.11 13	31.78 21	34.61 27	20.79 23	52.06 25

COLMAP(SR)		40.00 12	43.43 13	37.71 13	29.04 7	38.56 14	48.20 12	26.37 12	57.82 19

COLMAP(base)		35.81 20	40.69 18	32.55 19	22.89 17	33.51 20	40.95 21	23.20 17	58.48 17

dnet		0.00 35	0.00 35	0.00 35	0.00 34	0.00 34	0.00 35	0.00 34	0.00 35

CIDER		30.56 27	29.04 27	31.57 21	13.34 27	27.52 25	46.33 15	20.87 22	44.74 28
Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume. AAAI 2020
A-TVSNet + Gipuma		31.10 26	35.58 26	28.12 25	15.64 26	26.11 26	38.18 24	20.05 26	55.52 21

hgnet		7.97 32	14.91 32	3.35 32	2.62 29	2.34 31	6.42 31	1.29 32	27.20 32

example		10.98 30	18.98 31	5.65 29	2.55 31	2.27 33	10.87 29	3.81 30	35.41 30

DPSNet		7.97 32	14.91 32	3.35 32	2.62 29	2.34 31	6.42 31	1.29 32	27.20 32

Low-resolution many-view benchmark