Table 1 The state transition table of occlusion determination mechanism.

From: Robust online learning based on siamese network for ship tracking

Condition

\(Z_{k+1}\)

The kth frame condition

1

0

\(S_{k}>0\), \(\Delta S_{k}>0\), \(Z_{k}=0\)

2

0

\(S_{k}>t_{h}\), \(\Delta S_{k}>0\), \(Z_{k}=1\)

3

0

\(S_{k}>0\), \(\Delta S_{k}<0\), \(Z_{k}=0\), \(\left| \Delta S_{k}\right| <tol\)

4

1

\(S_{k}<t_{h}\), \(\Delta S_{k}>0\), \(Z_{k}=1\)

5

1

\(\Delta S_{k}<0\), \(Z_{k}=1\)

6

1

\(\Delta S_{k}<0\), \(\left| \Delta S_{k}\right| >tol\)

7

1

\(S_{k}<t_{l}\), \(\Delta S_{k}>0\), \(Z_{k}=1\)

  1. According to the classification score change degree and tracking state of the current frame, the tracking state of the next frame is predicted.