A Robust and Lightweight Detector for Ship Target with Complex Background in SAR Image

Algorithm 2

FPGM description.

Preparations:   input training data:X

         given pruning rate R.

1: Initialize:model parameter W=1

2: for epoch =1; epoch ≤ epoch_max; epoch ++ do

3:   Update the model parameter W based on X

4:      for i =1; i ≤ L; i ++ do

5:         Calculate: the sum of N Euclidean distances

6:      end for

7:      Find the relatively small number of N R filters

8:      Zeroize the selected filter gradient

9:   end for

10: Obtain the pruning model with zeroing model W^# from W

11: Remove zero parameters from the model W

Output: The pruning model and its parameters W