Abstract:Recent years have witnessed a remarkable breakthrough in synthetic aperture radar automatic target recognition (SAR ATR) with the development of deep learning (DL). Nonetheless, once deployed, the DL-based methods' ability to incrementally learn new knowledge from few-shot samples without forgetting the old is fragile, hindering them from discriminating unseen targets in real-world situations. In this article, we propose a cosine prototype learning (CPL) framework to first unlock few-shot class-incremental learning (FSCIL) in the SAR ATR field inspired by the intrinsic relationships between target azimuth-aware knowledge and semantic features under the cosine criterion. By condensing class-specific characteristics into individual prototypes, stable profiles of targets are depicted without losing generalization. For the model's plasticity, a pairwise structure separation (PSS) loss is introduced to separate old and new classes and compact intraclass features. Meanwhile, the model's transferability on new classes is guaranteed by a prototype consistency (PC) loss. For the model's stability, we propose a prototype–exemplar distillation (PED) loss and a prototype recalibration (PR) strategy to penalize semantic drifts of old-class feature spaces and alleviate the misalignment of the learned prototypes successively. At inference, a nearest-class-mean (NCM) classifier is adopted for evaluation by comparing cosine similarity scores between testing samples and class-specific prototypes. In experiments, the proposed components of our method are explored by ablation studies. Strong baselines are established, and extensive experiments conducted on the moving and stationary target acquisition and recognition (MSTAR) dataset show that our method outperforms state-of-the-art methods under various FSCIL conditions, verifying its effectiveness for the FSCIL of SAR ATR.

Improving SAR Automatic Target Recognition via Trusted Knowledge Distillation From Simulated Data

Multilevel Adaptive Knowledge Distillation Network for Incremental SAR Target Recognition

SAR Target Recognition Based on Probabilistic Meta-Learning

Simulated Data Feature Guided Evolution and Distillation for Incremental SAR ATR

Can We Trust Deep Learning Models in SAR ATR?

Dynamic Embedding Relation Distillation Network for Incremental SAR Automatic Target Recognition

Target Recognition in SAR Images by Deep Learning with Training Data Augmentation

A Bayesian Approach to Active Self-Paced Deep Learning for SAR Automatic Target Recognition

Few-Shot Class-Incremental SAR Target Recognition via Cosine Prototype Learning

SAR ATR Method with Limited Training Data via an Embedded Feature Augmenter and Dynamic Hierarchical-Feature Refiner

Dynamically Weighted Prototypical Learning Method for Few-Shot SAR ATR

DSRKD: Joint Despecking and Super-Resolution of SAR Images Via Knowledge Distillation

Capsule Broad Learning System Network for Robust Synthetic Aperture Radar Automatic Target Recognition with Small Samples

Crucial Feature Capture and Discrimination for Limited Training Data SAR ATR

SAR Target Classification Based on Knowledge Distillation

Double Reverse Regularization Network Based on Self-Knowledge Distillation for SAR Object Classification

Semi-Supervised SAR ATR Framework with Transductive Auxiliary Segmentation

A Comprehensive Survey on SAR ATR in Deep-Learning Era

High-resolution and reliable automatic target recognition based on photonic ISAR imaging system with explainable deep learning

Transductive Prototypical Attention Reasoning Network for Few-Shot SAR Target Recognition

LDCL: Low-Confidence Discriminant Contrastive Learning for Small-Sample SAR ATR