With the rapid development of aerospace technology, especially the rise of the commercial aerospace industry, the number of space targets in orbit has increased rapidly. As aerospace activities are unprecedentedly active, developing space situation awareness technology has become an urgent need for space security. High-resolution imaging of space targets to obtain space target information is critical in space situational awareness. Compared to traditional 2D imaging, 3D imaging can provide complete information about a space target, which makes it a key technology for space situational awareness. However, due to the limitations of observation setups and imaging algorithms, all current 3D imaging methods tend to construct sparse 3D geometries and cannot infer 2D novel view images with geometry consistency. In this paper, we will focus on far-field imaging and propose ISAR-NeRF to address these problems, which encodes the 3D physical geometry of the target into an implicit neural network. According to the physical mechanisms of ISAR imaging, we can effectively determine sampling points from ISAR images and feed these sampling points into an implicit neural network to output densities and backscattering coefficients. We propose a novel differentiable radar rendering algorithm that exploits densities and backscattering coefficients to render multi-view consistent images and construct 3D geometries without any explicit association algorithms. The comparative experiments qualitatively and quantitatively show that ISAR-NeRF outperforms any other methods in both 3D reconstruction and novel view synthesis. In addition, the experiments of low SNR and noisy projection vectors demonstrate the robustness of our methods.

Aircraft detection is a valuable but challenging task in synthetic aperture radar (SAR) automatic target recognition (ATR). Because of the complicated electromagnetic imaging mechanism of SAR, the SAR image of aircraft appears as a distributed collection of discrete scattering points that varies significantly with imaging conditions, like different incident angles, bringing great challenges to existing convolution neural network (CNN)-based detection methods for accurate aircraft detection. To address these challenges, we analyze and leverage the scattering characteristics of multiscale SAR aircraft to propose a novel SAR aircraft detector named deformable scattering feature correlation network (DSFCN). First, to deal with the discreteness of SAR aircraft, we propose a new Transformer-based backbone named scalable Swin Transformer backbone (SSTB) to replace a conventional CNN-based one, to effectively extract hierarchical scattering features from multiscale aircraft. Second, to cope with the varying image appearance of SAR aircraft, we design a deformable region correlation module (DRCM) to flexibly correlate strong scattering regions (SSRs) that carry aircraft salient features. Various interpretable experiments conducted on a real-measured Gaofen-3 SAR aircraft dataset demonstrate the superiority and reliability of our DSFCN over other representative CNN-based methods.

Text generation is a key component of many natural language tasks. Motivated by the success of generative adversarial networks (GANs) for image generation, many text-specific GANs have been proposed. However, due to the discrete nature of text, these text GANs often use reinforcement learning (RL) or continuous relaxations to calculate gradients during learning, leading to high-variance or biased estimation. Furthermore, the existing text GANs often suffer from mode collapse (i.e., they have limited generative diversity). To tackle these problems, we propose a new text GAN model named text feature GAN (TFGAN), where adversarial learning is performed in a continuous text feature space. In the adversarial game, GPT2 provides the “true” features, while the generator of TFGAN learns from them. TFGAN is trained by maximum likelihood estimation on text space and adversarial learning on text feature space, effectively combining them into a single objective, while alleviating mode collapse. TFGAN achieves appealing performance in text generation tasks, and it can also be used as a flexible framework for learning text representations.

Mixture models serve as one fundamental tool with versatile applications. However, their training techniques, like the popular Expectation Maximization (EM) algorithm, are notoriously sensitive to parameter initialization and often suffer from bad local optima that could be arbitrarily worse than the optimal. To address the long-lasting bad-local-optima challenge, we draw inspiration from the recent ground-breaking foundation models and propose to leverage their underlying big learning principle to upgrade the EM. Specifically, we present the Big Learning EM (BigLearn-EM), an EM upgrade that simultaneously performs joint, marginal, and orthogonally transformed marginal matchings between data and model distributions. Through simulated experiments, we empirically show that the BigLearn-EM is capable of delivering the optimal with high probability; comparisons on benchmark clustering datasets further demonstrate its effectiveness and advantages over existing techniques. The code is available at https://github.com/YulaiCong/Big-Learning-Expectation-Maximization.

Synthetic Aperture Radar (SAR) supports all-weather, day-and-night Earth observation, making it vital for military and civilian applications. While deep learning has advanced SAR target detection, state-of-the-art (SOTA) methods often directly adopt optical data augmentations, such as rotation, scaling, and Mixup, which violate SAR imaging physics and lead to artifacts, resolution distortion, and physically implausible scattering patterns. This work reveals that such physics-violating augmentations consistently compromise the detection reliability of real-time YOLO-series detectors, with the most severe drops occurring under strict localization criteria corresponding to higher IoU thresholds. To address this issue, we propose physics-preserving augmentations (PPA) that respect SAR domain priors. Extensive experiments on three representative datasets (SAR-AIRcraft-1.0, SIVED, and SRSDD) demonstrate the consistent superiority of PPA over optical augmentations in the context of SAR target detection. Notably, PPA enables the lightweight YOLOv11n to achieve 79.2% mAP75 on SAR-AIRcraft-1.0, comparable to or superior to existing SOTA SAR detectors while retaining real-time efficiency.

Recent work has shown generative adversarial networks (GANs) can generate highly realistic images, that are often indistinguishable (by humans) from real images. Most images so generated are not contained in the training dataset, suggesting potential for augmenting training sets with GAN-generated data. While this scenario is of particular relevance when there are limited data available, there is still the issue of training the GAN itself based on that limited data. To facilitate this, we leverage existing GAN models pretrained on large-scale datasets (like ImageNet) to introduce additional knowledge (which may not exist within the limited data), following the concept of transfer learning. Demonstrated by natural-image generation, we reveal that low-level filters (those close to observations) of both the generator and discriminator of pretrained GANs can be transferred to facilitate generation in a perceptually-distinct target domain with limited training data. To further adapt the transferred filters to the target domain, we propose adaptive filter modulation (AdaFM). An extensive set of experiments is presented to demonstrate the effectiveness of the proposed techniques on generation with limited data.