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

Table 1

Statistical table of model lightweight processing characteristics.

Model	Feature	Adaptive environment
Length-based	Removing redundant modules based on network structure	Redesign of network structure based on actual datasets
Depth-based	Removal of redundant feature information based on gradient	Adjust the feature extraction method of the receptive field to reduce gradient disappearance
Based on width (channel)	Removing redundant parameter based on network module	Wide range of applications, all networks available. A variety of implementations are implemented by specifying widths or using pruning techniques. Common pruning techniques include structured and unstructured pruning