🤖 AI Summary
To address catastrophic forgetting in continual learning for SAR target recognition, this paper proposes a lightweight and robust method tailored for incremental scenarios. The approach features: (i) a novel ViT-CNN dual-branch late-fusion architecture; (ii) a TinyViT lightweight backbone integrated with a dynamic attention enhancement mechanism; (iii) a neural RPCA-based approximate denoising module to improve SAR image robustness; and (iv) a prototype decoupling strategy leveraging random projection and an LDA variant to disentangle class representations and preserve prior knowledge. Evaluated on three major benchmarks—MSTAR, SAR-AIRcraft-1.0, and OpenSARShip—the method achieves a mean accuracy of 99.63% and a forgetting rate of only 0.33%. Crucially, it improves retention of previously learned classes by 89% over state-of-the-art methods, significantly advancing the practical deployment of continual SAR target recognition.
📝 Abstract
Deep learning techniques have achieved significant success in Synthetic Aperture Radar (SAR) target recognition using predefined datasets in static scenarios. However, real-world applications demand that models incrementally learn new information without forgetting previously acquired knowledge. The challenge of catastrophic forgetting, where models lose past knowledge when adapting to new tasks, remains a critical issue. In this paper, we introduce IncSAR, an incremental learning framework designed to tackle catastrophic forgetting in SAR target recognition. IncSAR combines the power of a Vision Transformer (ViT) and a custom-designed Convolutional Neural Network (CNN) in a dual-branch architecture, integrated via a late-fusion strategy. Additionally, we explore the use of TinyViT to reduce computational complexity and propose an attention mechanism to dynamically enhance feature representation. To mitigate the speckle noise inherent in SAR images, we employ a denoising module based on a neural network approximation of Robust Principal Component Analysis (RPCA), leveraging a simple neural network for efficient noise reduction in SAR imagery. Moreover, a random projection layer improves the linear separability of features, and a variant of Linear Discriminant Analysis (LDA) decorrelates extracted class prototypes for better generalization. Extensive experiments on the MSTAR, SAR-AIRcraft-1.0, and OpenSARShip benchmark datasets demonstrate that IncSAR significantly outperforms state-of-the-art approaches, achieving a 99.63% average accuracy and a 0.33% performance drop, representing an 89% improvement in retention compared to existing techniques. The source code is available at https://github.com/geokarant/IncSAR.