Deep Spiking Neural Network for Energy-Efficient SAR Ship Detection

Abstract

In this letter, we introduce the first spiking-based network optimized for synthetic aperture radar (SAR) ship detection and compare its performance with conventional neural networks (CNNs). Spiking neural networks (SNNs) offer significant advantages over traditional artificial neural networks (ANNs) by resulting in highly efficient computation. Unlike ANNs, SNNs only perform calculations when spikes occur, leading to lower power consumption and reduced computational costs, making them ideal for energy-constrained and onboard applications. Furthermore, we conduct experiments to analyze the power differences between the SNN and traditional ANN-based detection models. The results demonstrate the potential advantages of SNNs in terms of power efficiency and computational load in satellite-based target detection.