This paper addresses the important and challenging task of large-scale unsupervised semantic segmentation (LUSS). We present the first attempt to unleash the power of foundation models (FMs) for the challenging, dense prediction task LUSS, and our main objective is to present simple, effective yet efficient solutions for LUSS, namely Prompting foundation models for LUSS (PLUSS). Firstly, we proposed a cascade framework PLUSS$_\alpha$α by effectively marrying CLIPS, Grounding DINO, and SAM in a zero-shot manner. This cascade architecture automatically generates semantic and spatial prompts for SAM, establishing a strong baseline that significantly outperforms previous state-of-the-art methods. Building upon this foundation, we propose PLUSS$_\beta$β, which addresses the critical bottleneck of prompt quality through two novel tuner modules: a semantic tuner that enhances fine-grained category discrimination via visual prompt tuning, and a box tuner that improves object localization through cross-modal feature fusion. Both tuners are optimized by capitalizing on the knowledge already present within the foundation models themselves, deriving self-supervised signals from internal model consistency. This approach requires no external supervision or updates to the foundation models' parameters. Extensive experiments on ImageNet-S benchmarks demonstrate that PLUSS$_\beta$β achieves remarkable performance improvements, surpassing the previous best method by 39.6%, 27.3%, and 22.6% in mIoU for 50, 300, and 919 categories respectively. Our approach exhibits robust category-shape representation across varying object sizes and dataset scales, while maintaining strong generalization capabilities for open-vocabulary tasks. The proposed framework provides a solid baseline for adapting foundation models to downstream vision tasks.

The interrupted sampling repeater jamming (ISRJ) is a widely used coherent jamming technique. A proper waveform design can effectively suppress or mitigate the ISRJ. Even if several phase-coded waveform design methods have been proposed for this purpose, frequency modulation (FM) waveforms remain the most common choice for high-power transmitters, as they do not introduce significant distortions in real radar systems. In this letter, we propose a design method for multiple-input multiple-output (MIMO) radar that simultaneously derives the optimal polyphase-coded FM (PCFM) waveforms and the receive filters to mitigate the ISRJ. Specifically, we first model the joint design problem as a nonconvex bivariate optimization problem and we minimize the matching error between the desired and practical transmit-receive correlation functions for different channels. Subsequently, we adopt an alternating strategy to update the PCFM waveforms and receive filters sequentially. More specifically, gradient-based algorithms in Euclidean space and Riemannian manifold space are adopted to derive the optimal waveforms and filters, respectively. The proposed method is characterized by a low computational cost, thanks to its FFT-based implementation. Numerical analysis shows the effectiveness of the proposed method.

With the maturity of deep learning algorithm in Synthetic Aperture Radar (SAR) target recognition filed, Convolutional Neural Network (CNN) has become the most effective model. However, the interpretability and the separability of feature maps extracted from convolution layers have not been specially analyzed neither qualitatively nor quantitatively, which makes the traditional model work like a “black box”. To alleviate the problem, a novel model based on separability measure (SM) - CNN is proposed in this letter, which introduces the principle of maximal coding rate reduction to the backbone module. SM-CNN quantitatively analyzes the separability of the feature maps and takes the value as a vital part of the loss function to guide the training process of the model. The calculation process of the separability measure values can be strictly derived mathematically, so it is more interpretable, turning the black box into a “gray box”. Additionally, the proposed model can achieve comparable recognition performance of the backbone networks with reduced computational complexity. Comparative experiments based on MSTAR and OpenSARShip data sets verify the effectiveness and practicability of the method proposed in this letter.

Waveform adaptation grants cognitive radar (CR) the ability to adapt to its environment, which requires an effective framework to synthesize waveforms sharing a desired ambiguity function (AF). In this letter, we propose a novel method for shaping the slow-time AF in order to adaptively suppress the interference power. The problem is formulated as minimizing the average value of the slow-time AF over some range Doppler bins spanned by the interference, which can be identified exploiting a plurality of knowledge sources. From a technical point of view, this is tantamount to optimizing a complex quartic-order polynomial with a constant modulus (CM) constraint on each optimization variable. To solve this problem, we proposed a quartic Riemannian trust region algorithm. This algorithm first transforms the optimization into an unconstrained one in a complex circle Riemannian manifold, then devises a new Riemannian trust region optimization algorithm that invokes Riemannian gradient and Hessian matrix to obtain an iterative solution with super-linear convergence rate and ability to escape potential saddle points. Simulation results demonstrated that our proposed algorithm outperformed state-of-the art approaches for AF shaping while being computationally less expensive.