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	lakeside	sand box	storage room	storage room 2	tunnel

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

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

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

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

BP-MVSNet		1.30 6	1.30 6	1.30 6	1.30 6	1.30 6	1.30 6	1.30 6	1.30 6
Christian Sormann, Patrick Knöbelreiter, Andreas Kuhn, Mattia Rossi, Thomas Pock, Friedrich Fraundorfer: BP-MVSNet: Belief-Propagation-Layers for Multi-View-Stereo. 3DV 2020
Pnet-blend++		9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7

Pnet_fast		9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7

Pnet-eth		9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7

Pnet-blend		9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7

Pnet-new-		9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7	9.00 7

MVSNet_plusplus		10.85 12	9.79 12	11.56 12	10.23 12	14.23 12	9.34 12	10.23 12	10.23 12

MVSNet_++		10.85 12	9.79 12	11.56 12	10.23 12	14.23 12	9.34 12	10.23 12	10.23 12

confMetMVS		86.40 14	95.00 14	80.67 14	67.00 14	112.00 14	93.00 14	97.00 14	63.00 14

metmvs_fine		86.40 14	95.00 14	80.67 14	67.00 14	112.00 14	93.00 14	97.00 14	63.00 14

unMVSmet		86.40 14	95.00 14	80.67 14	67.00 14	112.00 14	93.00 14	97.00 14	63.00 14

Cas-MVS_preliminary		283.71 17	277.25 17	288.02 17	287.28 17	287.14 17	275.09 17	279.41 17	289.65 17

FADENet		300.30 18	300.30 18	300.30 18	300.30 18	300.30 18	300.30 18	300.30 18	300.30 18

CPR_FA		300.30 18	300.30 18	300.30 18	300.30 18	300.30 18	300.30 18	300.30 18	300.30 18

unMVSv1		438.86 20	369.40 20	485.16 20	432.90 20	452.10 20	398.70 20	340.10 20	570.50 21

MVSNet + Gipuma		622.80 21	505.00 25	701.33 21	642.00 22	671.00 21	502.00 25	508.00 23	791.00 22

F/T MVSNet+Gipuma		622.80 21	505.00 25	701.33 21	642.00 22	671.00 21	502.00 25	508.00 23	791.00 22

P-MVSNet		626.74 23	503.44 24	708.94 23	634.45 21	681.77 23	490.58 24	516.29 28	810.60 24

PMVS	863.65 2	639.40 24	530.00 28	712.33 24	798.00 29	888.00 29	701.00 30	359.00 21	451.00 20
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
DPSNet		660.88 25	500.81 21	767.59 25	683.99 24	726.46 24	489.58 21	512.04 25	892.32 26

example		660.88 25	500.81 21	767.59 25	683.99 24	726.46 24	489.58 21	512.04 25	892.32 26

hgnet		660.88 25	500.81 21	767.59 25	683.99 24	726.46 24	489.58 21	512.04 25	892.32 26

CIDER		669.75 28	505.36 27	779.35 28	703.58 27	808.57 27	513.92 27	496.79 22	825.90 25
Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume. AAAI 2020
MVSNet		769.20 29	609.10 29	875.93 29	786.50 28	836.20 28	593.70 28	624.50 29	1005.10 29

ACMH+	729.97 1	894.58 30	641.24 30	1063.48 30	901.73 30	959.70 30	639.95 29	642.52 30	1329.02 30

ACMH	993.80 3	1055.91 31	719.19 31	1280.39 31	1101.67 31	1267.33 31	717.78 31	720.60 31	1472.17 31
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
TAPA-MVS(SR)		1338.82 32	969.96 32	1584.73 32	1327.62 32	1473.80 32	922.00 32	1017.91 32	1952.77 32

ACMP	1246.57 5	1631.36 33	1293.07 34	1856.88 34	1568.97 34	1868.85 34	1273.91 34	1312.23 34	2132.82 34
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
ACMM	1311.08 6	1661.53 34	1487.23 35	1777.73 33	1530.09 33	1690.76 33	1475.05 35	1499.41 35	2112.35 33
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
OpenMVS_ROB	1936.51 8	1881.32 35	1172.27 33	2354.03 35	2118.19 35	2241.51 36	1205.90 33	1138.64 33	2702.38 35

PVSNet	1207.94 4	2116.05 36	1643.70 37	2430.96 36	2140.83 36	2235.92 35	1648.06 37	1639.33 37	2916.13 36

3Dnovator	2482.06 12	2364.20 37	1597.00 36	2875.67 38	2548.00 38	2712.00 38	1635.00 36	1559.00 36	3367.00 38

R-MVSNet		2412.80 38	1890.00 39	2761.33 37	2391.00 37	2612.00 37	1879.00 39	1901.00 39	3281.00 37

OpenMVS	2352.88 11	2568.40 39	1727.00 38	3129.33 39	2833.00 39	3048.00 39	1755.00 38	1699.00 38	3507.00 39

COLMAP_ROB	2144.59 9	3311.60 40	2554.50 40	3816.33 41	3382.00 40	3555.00 41	2484.00 40	2625.00 40	4512.00 41
Johannes L. Schönberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
PVSNet_0	1661.03 7	3323.02 41	2633.37 41	3782.80 40	3505.59 41	3539.91 40	2550.07 41	2716.66 41	4302.89 40

DeepC-MVS_fast	2190.19 10	3761.23 42	3050.62 43	4234.97 42	3851.12 42	3561.12 42	3001.56 44	3099.67 43	5292.66 42
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
3Dnovator+	3414.41 15	4057.00 43	3365.00 44	4518.33 43	3909.00 43	4339.00 43	2806.00 42	3924.00 46	5307.00 43

CMPMVS	2623.62 13	4160.94 44	3037.05 42	4910.20 44	4057.20 44	5253.20 45	3109.40 45	2964.70 42	5420.20 44
M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces. CVPR 2011
PLC	2858.82 14	4457.00 45	3560.00 46	5055.00 45	4457.00 45	4671.00 44	3360.00 46	3760.00 44	6037.00 45
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
DeepPCF-MVS	3452.42 16	5328.53 46	3926.82 47	6263.00 46	5101.63 46	5379.92 47	4010.62 47	3843.02 45	8307.46 46

DeepC-MVS	3917.13 17	5746.11 47	4462.91 48	6601.57 47	5612.31 47	5379.45 46	4610.56 48	4315.25 48	8812.96 47
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
MVE	10021.00 20	8749.80 48	3439.50 45	12290.00 50	6275.00 48	16332.00 51	2923.00 43	3956.00 47	14263.00 50
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
PCF-MVS	4335.85 18	9288.52 49	5777.29 49	11629.34 49	9919.03 50	10475.60 49	5496.66 49	6057.92 49	14493.40 51
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
A-TVSNet + Gipuma		9351.57 50	7502.61 50	10584.21 48	9219.52 49	10119.20 48	7334.37 50	7670.84 50	12413.90 49

TAPA-MVS	6309.84 19	13353.70 51	10496.80 52	15258.30 51	14678.80 51	14057.00 50	10791.00 52	10202.60 52	17039.10 52
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
LTVRE_ROB	20565.47 21	13815.20 52	7669.50 51	17912.33 52	22283.00 52	21104.00 52	7445.00 51	7894.00 51	10350.00 48
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
test_1205		31740.00 53	24750.00 53	36400.00 53	32520.00 53	33720.00 53	24240.00 53	25260.00 53	42960.00 53

test_1124		31740.00 53	24750.00 53	36400.00 53	32520.00 53	33720.00 53	24240.00 53	25260.00 53	42960.00 53

test_1126		31740.00 53	24750.00 53	36400.00 53	32520.00 53	33720.00 53	24240.00 53	25260.00 53	42960.00 53

test_1120		31740.00 53	24750.00 53	36400.00 53	32520.00 53	33720.00 53	24240.00 53	25260.00 53	42960.00 53

GSE		84345.00 57	83012.00 57	85233.67 57	88513.00 57	78675.00 57	82336.00 57	83688.00 57	88513.00 57

LPCS		84345.00 57	83012.00 57	85233.67 57	88513.00 57	78675.00 57	82336.00 57	83688.00 57	88513.00 57

MVSCRF		279999.00 59	99999.00 59	399999.00 59	999999.00 59	99999.00 59	99999.00 59	99999.00 59	99999.00 59

CasMVSNet(base)		1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60

CasMVSNet(SR_B)		1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60

CasMVSNet(SR_A)		1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60

COLMAP(SR)		1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60

COLMAP(base)		1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60	1000000.00 60

Snet		10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65

dnet		10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65

AttMVS		10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65	10000000.00 65

UnsupFinetunedMVSNet					10000000.00 65

Low-resolution many-view benchmark