Skip to main content

Table 3 Quantitative evaluations of model alignment. Euclidean distance in meters are reported for positional errors

From: Efficient video collection association using geometry-aware Bag-of-Iconics representations

Evaluated model

London eye

Westminster Abbey

Tower of London

Brandenburg gate

Average

Reference model

Big Ben

Big Ben

Tower Bridge

Reichastag

 

Orientation error (°)

6.94

5.46

4.38

8.34

6.28

Position error (m)

1.71

0.96

3.15

2.76

2.15

Scaling error (%)

3.42

4.67

9.19

2.47

4.94

  1. Rotations are converted to axis-angle representation, and errors are reported as average angle differences in degrees. Relative errors in percentage are reported for scaling
Back to article page