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

DeepPCF-MVS	96.37 2	92.61 1	89.11 1	94.94 1	95.58 1	95.07 1	85.62 1	92.60 1	94.17 2

tm-dncc		91.49 3	89.01 2	93.15 5	92.89 8	91.90 21	85.54 2	92.48 2	94.65 1

DeepC-MVS_fast	96.70 1	92.27 2	88.75 3	94.63 2	95.15 3	94.80 2	85.14 4	92.35 3	93.92 3
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
TAPA-MVS	93.98 7	90.68 5	88.05 4	92.44 8	92.42 12	93.52 8	85.24 3	90.85 7	91.36 15
Andrea Romanoni, Matteo Matteucci: TAPA-MVS: Textureless-Aware PAtchMatch Multi-View Stereo. ICCV 2019
TAPA-MVS(SR)		90.57 6	87.48 5	92.63 7	92.64 10	93.17 11	83.91 6	91.04 5	92.10 13

DeepC-MVS	95.98 3	91.43 4	87.48 5	94.06 4	94.52 5	94.15 5	83.69 7	91.26 4	93.50 4
Andreas Kuhn, Christian Sormann, Mattia Rossi, Oliver Erdler, Friedrich Fraundorfer: DeepC-MVS: Deep Confidence Prediction for Multi-View Stereo Reconstruction. 3DV 2020
PLC	95.07 4	89.96 9	87.00 7	91.94 13	90.65 21	92.68 17	83.01 9	90.98 6	92.48 8
Jie Liao, Yanping Fu, Qingan Yan, Chunxia xiao: Pyramid Multi-View Stereo with Local Consistency. Pacific Graphics 2019
GSE		89.80 12	86.71 8	91.87 14	90.87 20	92.62 18	83.49 8	89.92 11	92.11 12

IB-MVS	91.98 17	88.88 17	86.61 9	90.39 21	92.23 14	92.94 15	82.93 10	90.30 8	86.00 25
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
COLMAP(base)		89.91 10	86.37 10	92.28 10	91.56 17	92.95 14	82.53 11	90.21 9	92.33 9

LTVRE_ROB	92.95 15	89.00 15	86.32 11	90.78 20	91.67 16	89.54 28	84.87 5	87.77 18	91.13 17
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
ACMH+	92.99 14	89.84 11	85.96 12	92.43 9	92.28 13	92.84 16	81.89 14	90.03 10	92.17 11

COLMAP(SR)		90.17 8	85.79 13	93.08 6	92.63 11	93.87 6	82.51 12	89.07 15	92.74 7

ACMM	93.85 9	89.42 13	85.42 14	92.10 12	91.09 19	92.98 13	81.16 16	89.68 12	92.23 10
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
PCF-MVS	93.45 11	90.50 7	85.08 15	94.11 3	95.54 2	93.44 10	82.27 13	87.88 17	93.34 5
Andreas Kuhn, Shan Lin, Oliver Erdler: Plane Completion and Filtering for Multi-View Stereo Reconstruction. GCPR 2019
ACMH	92.88 16	88.89 16	84.55 16	91.78 15	91.37 18	92.41 19	79.88 18	89.21 13	91.56 14
Qingshan Xu and Wenbing Tao: Multi-Scale Geometric Consistency Guided Multi-View Stereo. CVPR 2019
COLMAP_ROB	93.27 12	88.32 18	84.41 17	90.93 19	89.38 24	92.11 20	79.72 19	89.10 14	91.28 16
Johannes L. SchÃ¶nberger, Enliang Zheng, Marc Pollefeys, Jan-Michael Frahm: Pixelwise View Selection for Unstructured Multi-View Stereo. ECCV 2016
ACMP	93.49 10	89.04 14	84.30 18	92.21 11	90.34 22	93.47 9	79.54 20	89.07 15	92.81 6
Qingshan Xu and Wenbing Tao: Planar Prior Assisted PatchMatch Multi-View Stereo. AAAI 2020
LPCS		85.52 25	83.61 19	86.79 27	84.99 33	88.49 30	81.34 15	85.87 22	86.89 21

A-TVSNet + Gipuma		86.96 22	82.91 20	89.65 24	88.10 27	91.71 23	80.23 17	85.60 23	89.15 19

HY-MVS	93.96 8	87.12 20	82.85 21	89.96 23	89.33 25	93.07 12	78.79 22	86.90 20	87.48 20

BP-MVSNet		86.71 23	81.68 22	90.06 22	91.85 15	87.55 34	75.75 26	87.61 19	90.77 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
P-MVSNet		77.69 36	81.61 23	75.07 46	71.91 51	77.33 49	76.51 25	86.71 21	75.98 40

3Dnovator	94.51 5	87.43 19	80.90 24	91.78 15	94.76 4	94.22 3	78.96 21	82.85 24	86.35 22

3Dnovator+	94.38 6	87.07 21	80.26 25	91.61 17	94.34 6	94.17 4	78.01 23	82.52 25	86.34 23

OpenMVS	93.04 13	86.34 24	79.08 26	91.18 18	93.50 7	93.77 7	76.56 24	81.60 28	86.26 24

R-MVSNet		81.26 29	77.44 27	83.81 33	82.82 35	86.25 36	74.52 28	80.36 30	82.37 29

tmmvs		84.10 26	76.16 28	89.39 25	92.89 8	91.90 21	74.49 29	77.82 33	83.39 27

A1Net		68.49 47	75.64 29	63.72 58	63.53 62	51.39 76	68.88 35	82.41 26	76.23 39

test_1205		80.31 31	75.46 30	83.55 34	86.48 31	90.90 24	70.38 33	80.54 29	73.26 45

PVSNet_0	88.72 19	82.59 27	75.14 31	87.55 26	88.79 26	88.38 31	73.35 30	76.93 34	85.48 26

PVSNet	91.96 18	81.42 28	74.82 32	85.83 30	87.90 28	86.77 35	74.82 27	74.81 39	82.83 28

ANet-0.75		77.73 35	74.78 33	79.70 37	77.01 44	84.71 39	67.34 38	82.21 27	77.39 37

PVSNet_LR		80.13 34	73.03 34	84.86 32	84.39 34	88.14 32	66.94 39	79.13 31	82.06 30

CIDER		80.48 30	72.14 35	86.05 28	87.66 29	89.77 26	71.75 31	72.53 44	80.72 32
Qingshan Xu and Wenbing Tao: Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume. AAAI 2020
CPR_FA		74.23 37	72.06 36	75.68 44	74.30 49	70.69 56	68.60 36	75.52 36	82.05 31

test_1126		80.30 32	71.93 37	85.88 29	89.75 23	90.41 25	70.18 34	73.68 41	77.50 36

OpenMVS_ROB	86.42 20	80.25 33	71.88 38	85.82 31	87.52 30	89.56 27	70.53 32	73.24 42	80.39 33

RMVSNet		62.69 55	69.01 39	58.48 66	69.15 57	58.56 65	59.72 42	78.30 32	47.75 73

ANet		74.07 38	68.26 40	77.94 41	77.01 44	84.71 39	64.75 40	71.76 45	72.09 46

AttMVS		72.35 39	68.13 41	75.16 45	70.98 55	80.63 44	68.09 37	68.16 47	73.87 43

metmvs_fine		60.95 56	66.50 42	57.25 68	64.00 60	51.92 72	59.04 43	73.97 40	55.84 67

Pnet-eth		60.43 58	65.87 43	56.80 69	63.59 61	45.39 80	55.58 45	76.16 35	61.43 61

unMVSv1		63.91 52	62.59 44	64.80 57	66.39 58	68.00 59	61.62 41	63.55 60	60.01 62

mvs_zhu_1030		72.30 40	62.31 45	78.96 39	77.79 42	83.27 41	55.37 46	69.25 46	75.83 41

MVE	62.14 22	59.41 62	60.38 46	58.77 65	50.39 78	69.00 58	56.98 44	63.78 56	56.91 66
Simon Fuhrmann, Fabian Langguth, Michael Goesele: MVE - A Multi-View Reconstruction Environment. EUROGRAPHICS Workshops on Graphics and Cultural Heritage (2014)
test3		57.95 64	60.13 47	56.50 71	54.75 75	51.79 73	53.99 47	66.26 48	62.98 55

MVS_test_1		68.30 48	59.98 48	73.84 48	80.46 38	73.07 52	44.68 64	75.28 37	67.99 52

SGNet		57.44 66	58.95 49	56.44 72	55.09 72	52.01 71	52.63 49	65.27 51	62.22 56

PSD-MVSNet		57.63 65	58.95 49	56.74 70	56.38 69	52.03 70	52.75 48	65.16 52	61.82 60

unsupervisedMVS_cas		68.86 45	58.51 51	75.77 43	74.54 48	80.76 43	51.51 53	65.51 50	72.00 47

test_mvsss		69.43 43	58.19 52	76.92 42	77.20 43	80.08 45	43.23 66	73.16 43	73.47 44

TVSNet		56.87 68	57.97 53	56.13 73	55.48 71	50.99 77	49.98 56	65.96 49	61.93 59

MVSNet		63.58 53	56.25 54	68.47 55	66.00 59	71.12 53	48.36 58	64.13 54	68.29 51

CasMVSNet(SR_B)		56.87 68	55.95 55	57.48 67	56.06 70	67.85 60	47.86 60	64.05 55	48.53 72

QQQNet		52.14 74	55.46 56	49.93 77	54.81 74	51.46 75	52.58 50	58.34 62	43.53 78

CasMVSNet(SR_A)		69.66 42	55.07 57	79.38 38	80.08 39	87.62 33	46.56 61	63.58 59	70.44 48

hgnet		59.89 60	54.16 58	63.70 59	71.39 52	70.87 54	52.09 51	56.24 67	48.85 70

DPSNet		59.89 60	54.16 58	63.70 59	71.39 52	70.87 54	52.09 51	56.24 67	48.85 70

CasMVSNet(base)		68.63 46	53.70 60	78.58 40	79.92 40	85.96 37	44.83 63	62.57 61	69.87 49

Pnet-new-		70.55 41	53.66 61	81.81 36	81.27 37	85.23 38	51.15 55	56.17 69	78.92 35

MVSNet + Gipuma		57.07 67	53.08 62	59.73 64	62.91 65	57.42 67	48.41 57	57.74 65	58.88 63

firsttry		54.62 73	52.74 63	55.88 74	54.90 73	55.69 68	48.02 59	57.46 66	57.06 65

MVSCRF		63.13 54	52.16 64	70.45 51	70.18 56	75.83 51	51.17 54	53.16 74	65.33 53

Pnet-blend++		64.02 50	52.13 65	71.95 49	75.81 46	78.00 47	40.66 71	63.61 57	62.05 57

Pnet-blend		64.02 50	52.13 65	71.95 49	75.81 46	78.00 47	40.66 71	63.61 57	62.05 57

SVVNet		47.76 78	49.94 67	46.31 79	48.51 79	45.06 81	41.53 69	58.34 62	45.36 75

ternet		47.76 78	49.94 67	46.31 79	48.51 79	45.06 81	41.53 69	58.34 62	45.36 75

F/T MVSNet+Gipuma		55.78 70	49.83 69	59.76 63	63.10 63	57.51 66	46.34 62	53.31 73	58.66 64

MVSNet_plusplus		64.09 49	48.06 70	74.77 47	85.29 32	62.00 61	31.11 80	65.01 53	77.03 38

Snet		60.77 57	47.18 71	69.82 52	77.90 41	61.73 62	39.61 74	54.76 71	69.82 50

Pnet_fast		68.89 44	47.01 72	83.47 35	81.84 36	89.30 29	38.61 75	55.42 70	79.28 34

CCVNet		49.22 76	46.91 73	50.75 76	53.16 77	55.57 69	40.21 73	53.61 72	43.53 78

MVSNet_++		59.91 59	45.04 74	69.82 52	74.28 50	60.08 63	14.97 84	75.11 38	75.10 42

example		55.06 72	44.75 75	61.93 61	71.21 54	69.85 57	42.82 68	46.68 77	44.73 77

test_1124		58.90 63	44.17 76	68.72 54	62.60 68	78.44 46	38.12 76	50.22 75	65.10 54

confMetMVS		44.56 80	41.82 77	46.39 78	46.63 81	51.78 74	37.38 77	46.25 78	40.77 81

unMVSmet		48.01 77	40.73 78	52.86 75	53.90 76	58.72 64	36.72 78	44.74 79	45.96 74

vp_mvsnet		55.09 71	38.17 79	66.36 56	62.91 65	81.28 42	29.09 83	47.26 76	54.90 68

Cas-MVS_preliminary		38.91 82	36.73 80	40.37 82	37.57 82	48.76 79	30.59 81	42.87 80	34.76 82

test_1120		40.25 81	35.81 81	43.21 81	30.29 84	50.38 78	29.89 82	41.72 81	48.95 69

PMVS	61.03 23	49.24 75	32.69 82	60.27 62	62.70 67	76.26 50	42.99 67	22.38 82	41.86 80
Y. Furukawa, J. Ponce: Accurate, dense, and robust multiview stereopsis. PAMI (2010)
FADENet		4.02 84	4.30 83	3.83 85	6.03 86	3.71 85	5.02 85	3.58 83	1.74 84

CMPMVS	66.06 21	9.32 83	0.44 84	15.23 84	6.17 85	39.51 83	0.89 86	0.00 84	0.00 85
M. Jancosek, T. Pajdla: Multi-View Reconstruction Preserving Weakly-Supported Surfaces. CVPR 2011
dnet		0.00 85	0.00 85	0.00 86	0.00 87	0.00 86	0.00 87	0.00 84	0.00 85

test_MVS							43.40 65

test_robustmvs				23.92 83	37.28 83	22.69 84	32.97 79		11.78 83

UnsupFinetunedMVSNet					63.10 63

Low-resolution many-view benchmark