Few-Shot Class-Incremental Learning For Efficient SAR Automatic Target Recognition

TL;DR

This paper introduces a novel few-shot class-incremental learning framework for SAR automatic target recognition, effectively addressing data scarcity and improving recognition accuracy with a dual-branch architecture and advanced feature fusion techniques.

Contribution

The paper presents a new FSCIL framework with a dual-branch architecture, combining local feature extraction, Fourier transforms, and cross-attention for improved SAR target recognition.

Findings

Outperforms state-of-the-art FSCIL methods on MSTAR dataset

Enhances class separation with focal and center loss functions

Maintains computational efficiency with minimal scale-shift parameters

Abstract

Synthetic aperture radar automatic target recognition (SAR-ATR) systems have rapidly evolved to tackle incremental recognition challenges in operational settings. Data scarcity remains a major hurdle that conventional SAR-ATR techniques struggle to address. To cope with this challenge, we propose a few-shot class-incremental learning (FSCIL) framework based on a dual-branch architecture that focuses on local feature extraction and leverages the discrete Fourier transform and global filters to capture long-term spatial dependencies. This incorporates a lightweight cross-attention mechanism that fuses domain-specific features with global dependencies to ensure robust feature interaction, while maintaining computational efficiency by introducing minimal scale-shift parameters. The framework combines focal loss for class distinction under imbalance and center loss for compact intra-class distributions to enhance class separation boundaries. Experimental results on the MSTAR benchmark dataset demonstrate that the proposed framework consistently outperforms state-of-the-art methods in FSCIL SAR-ATR, attesting to its effectiveness in real-world scenarios.