Skip to main content

Table 1 Statistics for the angular error between the normal maps recovered for various methods and the ground truth, for three synthesized datasets. All numbers are shown in degrees

From: Robust surface normal estimation via greedy sparse regression

 

25 % Quantile

Median

Mean

75 % Quantile

STD

Dataset Caesar

LS

4.669

8.586

9.672

14.32

5.875

LMS mode

1.313

6.769

8.148

13.46

7.620

LMS

1.335

2.692

3.700

4.305

4.242

IRL1

1.854

3.339

5.276

6.440

5.080

SBL

1.576

2.441

4.487

4.488

5.158

OMP

0.7571

2.624

2.424

2.690

3.332

Dataset Sphere

LS

7.255

12.23

13.55

20.79

7.566

LMS mode

1.658

4.939

6.211

9.907

4.901

LMS

2.588

3.601

3.261

4.030

1.103

IRL1

3.045

4.253

4.960

5.662

3.119

SBL

2.762

3.620

3.671

4.239

1.822

OMP

2.617

3.051

3.179

3.991

0.9324

Dataset Buddha

LS

9.039

14.03

15.29

20.19

8.25

LMS mode

7.304

14.70

17.42

25.03

13.79

LMS

1.985

4.736

8.454

6.391

15.35

IRL1

4.066

7.354

10.08

13.28

9.178

SBL

2.412

5.295

8.034

9.333

9.479

OMP

1.289

3.049

7.065

5.495

14.17