25 September 2023 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Integrating an automatic target recognition (ATR) system into real-world applications presents a challenge as it may frequently encounter new samples from unseen classes. To overcome this challenge, it is necessary to adopt incremental learning, which enables the continuous acquisition of new knowledge while retaining previous knowledge. This paper introduces a novel, multi-purpose interpretability metric for ATR systems that employ synthetic aperture radar (SAR) images. The metric leverages the Local Interpretable Model-Agnostic Explanation (LIME) algorithm, enhancing human decision-making by providing a secondary measure alongside the conventional classification score. Additionally, the proposed metric is employed to analyze the robustness of Convolutional Neural Networks (CNNs) by examining the impact of target features and irrelevant background correlations on recognition results. Finally, we demonstrate the effectiveness of the proposed metric in the context of incremental learning. By utilizing the proposed interpretability metric, we select exemplars in an incremental learning scenario, resulting in improved performance and showcasing the application potential of our proposed methodology. The network is fine-tuned sequentially with unknown samples recognized by the Openmax classifier and exemplars from the old known classes, which are selected based on the proposed interpretability metric. The effectiveness of this approach is demonstrated using the publicly available MSTAR dataset.
8 January 2025 in IEEE Transactions on Radar Systems
Three-Dimensional Interferometric Synthetic Aperture Radar (3D-InISAR) imaging provides a more complete and reliable representation of targets compared to traditional 2D-ISAR, overcoming limitations related to the geometry of the radar-target system and relative motion. This paper presents the application of a Point Cloud Transformer (PCT) for Automatic Target Recognition (ATR) using 3D-InISAR data. The PCT model, originally developed to classify LIDAR’s point clouds, is trained on sparse synthetic point cloud datasets representing various military vehicles, including cars, tanks, and trucks. The synthetic data is carefully generated from CAD models, incorporating techniques such as voxel downsampling and data augmentation to ensure high fidelity and diversity. Initial testing on synthetic data demonstrates the PCT’s robustness and high accuracy when used for ATR. To bridge the gap between synthetic and real data, a transfer learning approach is employed, which operates a fine-tuning on the pre-trained model by using real 3D-InISAR point clouds obtained from the publicly available SDMS-AFRL dataset. Results show significant improvements in classification accuracy post fine-tuning, validating the effectiveness of the PCT model for real-world ATR applications. The findings highlight the potential of transformer-based models in enhancing target recognition systems for future ATR systems based on 3D radar images.
14 November 2025 in TechRxiv
Deep Neural Networks (DNNs), which play a significant role in a wide range of both civilian and military applications, are often regarded as "black-boxes" due to their opaque decision-making processes, and this raises concerns about their reliability. Explainable AI (XAI) aims to address this issue by providing human-understandable explanations for their decisions. In low-risk tolerance applications like automatic target recognition (ATR), ensuring reliable predictions is imperative. While some reliability measures exist, such as scalar thresholds and model ensembles, they often fall short in practical settings. This paper addresses the challenge of enhancing reliability in the classification of Synthetic Aperture Radar (SAR) images by incorporating XAI techniques. Specifically, the SHapley Additive exPlanations (SHAP) algorithm is employed to distinguish between positive and negative contributions of individual pixels to the outcomes of Convolutional Neural Networks (CNNs). Reliability is associated with the model's ability to focus on target features rather than spurious correlations from background clutter. The paper proposes a novel numerical reliability index based on the SHAP analysis to enhance the decision-making process. This index quantifies the reliability of predictions based on pixel contributions. The theoretical findings are evaluated using the OpenSARShip dataset, which contains ship patches extracted from SAR images.
18 October 2023 in TechRxiv
Navigational systems, the heart of maritime oper- ations, face escalating cybersecurity risks due to their system of systems nature and reliance on diverse suppliers. Amid concerns about supply chain attacks and insider threats, the potential for malicious radar image injections, including Inverse Synthetic Aperture Radar (ISAR) images, is emerging. Such manipulations can critically undermine navigational integrity through the generation of decoy targets, the strategic relocation of existing ones, and the effective concealment of additional targets. We have identified where an Advanced Persistent Threat (APT) could be concealed within the processing chain of a modern radar system. This study demonstrates the potential of APTs to exploit Generative Adversarial Networks (GANs) for the creation of deceptive ISAR images, thereby spotlighting this previously unexplored threat vector. Our findings offer novel insights into bolstering maritime cybersecurity in an increasingly AI-dominated landscape.