arXiv subjects

Chaowei Wang

Publications and source records attributed to Chaowei Wang.

8 recordsLinked to original sources

arXiv2026

GrabVG: Graph-Attentive Binding for Visual Grounding in UAV Imagery

Visual grounding in Unmanned Aerial Vehicle (UAV) imagery aims to localize a target object in complex bird's-eye-view scenes according to a natural language description. However, the abundance of small, densely distributed, and visually similar objects creates high visual redundancy, while repetitive local configurations give rise to strong topological ambiguity. Existing approaches mainly focus on visual--language feature alignment or dense contextual interaction, yet they struggle to distinguish subtle inter-instance differences and effectively exploit spatial topological structures, leading to inaccurate grounding in highly crowded scenarios. To address these challenges, we propose $\textbf{GrabVG}$, a novel visual grounding framework inspired by human visual search. GrabVG explicitly decomposes grounding into two sequential stages: $\textit{preattentive hypothesis search}$ and $\textit{graph-attentive feature binding}$. Specifically, we first generate a compact set of reliable object hypotheses through distillation-guided proposal induction and text-aware hypothesis filtering, substantially reducing background distractions and semantic mismatches. These hypotheses are then organized into a sparse graph, where language-guided intra-instance visual cues and inter-instance topological relationships are jointly bound and propagated via graph attention, enabling efficient spatial reasoning and accurate target localization. Extensive experiments on AerialVG and AerialSense show that GrabVG achieves a favorable accuracy--speed trade-off, reaching 67.31$\%$ and 80.34$\%$ Acc@0.5 and outperforming the corresponding baselines by 10.55 and 8.76 percentage points, respectively.

cs.CV↗

arXiv2026

From Scene-Centric to Observer-Centric: Modeling Observer-Aware Relations for 3D Scene Graph Generation

3D Scene Graph Generation (3DSGG) represents 3D scenes as structured object--relation--object graphs for spatial understanding. In observer-centric spatial perception, the same scene may be expressed under different local observer frames while its structure remains unchanged. However, existing models typically assume a fixed scene-aligned reference frame and may produce semantically inconsistent predictions when the scene is re-expressed in another observer frame. We attribute this failure to the heterogeneous frame dependency of relational predicates. Directional predicates such as $\textit{left}$, $\textit{front}$, $\textit{right}$, and $\textit{behind}$ are $\textbf{Observer-Dependent Relations}$, whereas most contact, support, and semantic predicates, such as $\textit{standing on}$ and $\textit{attached to}$, are approximately $\textbf{Observer-Independent Relations}$. Conventional models do not distinguish these frame responses, leading to degraded relation prediction under observer-frame reorientation. We introduce $\textbf{Observer-Aware Relations (OAR)}$, which combines observer-aware geometric encoding and relation specialization, supported by frame-stable object encoding, for unified multi-label predicate prediction. Experiments on 3DSSG show that OAR consistently outperforms baselines across controlled observer-frame reorientations without training-time frame-reorientation augmentation, while remaining competitive on the standard benchmark. The project page is available at https://oar-predicate.github.io/.

cs.CV↗

arXiv2026

Realization of the SI Second Defined by Geometric Mean of Multiple Clock Transitions

The current definition of the SI second is based on the 133Cs ground-state hyperfine transition in the microwave domain, with the most accurate realizations achieving fractional frequency uncertainties of about (1-2)E16. In contrast, state-of-the-art optical clocks now demonstrate estimated uncertainties two to three orders of magnitude lower, prompting discussion on the redefinition of the SI second. Several options for the new definition have been proposed, one of which introduces a constant N defined as the weighted geometric mean of multiple clock transition frequencies. In this work, we investigate how N can be practically realized when not all defining transitions are available and when multiple optical clocks operate with different performance levels and non-overlapping uptimes. We consider two complementary realization and reconstruction routes. One route is based on geometric-mean combinations, and the other is based on arithmetic-mean combinations. We derive consistent uncertainty expressions that incorporate both measurement uncertainties and, where required, uncertainties of recommended frequencies or frequency ratios. Using analytic three-transition case studies, we identify the parameter regimes in which each route yields a lower total uncertainty and provide explicit conditions for the crossover between them. We further address the dominant role of dead time when a hydrogen maser serves as a flywheel reference by introducing a time-segmented, time-weighted combination based on coefficient and covariance matrices, which accounts for overlapping operation and correlations across measurement intervals. Our findings offer practical guidance for minimizing total uncertainty in multi-clock realizations and contribute to ongoing efforts toward redefining the SI second.

physics.atom-ph↗

arXiv2026

Towards Any-Quality Image Segmentation via Generative and Adaptive Latent Space Enhancement

Segment Anything Models (SAMs), known for their exceptional zero-shot segmentation performance, have garnered significant attention in the research community. Nevertheless, their performance drops significantly on severely degraded, low-quality images, limiting their effectiveness in real-world scenarios. To address this, we propose GleSAM++, which utilizes Generative Latent space Enhancement to boost robustness on low-quality images, thus enabling generalization across various image qualities. Additionally, to improve compatibility between the pre-trained diffusion model and the segmentation framework, we introduce two techniques, i.e., Feature Distribution Alignment (FDA) and Channel Replication and Expansion (CRE). However, the above components lack explicit guidance regarding the degree of degradation. The model is forced to implicitly fit a complex noise distribution that spans conditions from mild noise to severe artifacts, which substantially increases the learning burden and leads to suboptimal reconstructions. To address this issue, we further introduce a Degradation-aware Adaptive Enhancement (DAE) mechanism. The key principle of DAE is to decouple the reconstruction process for arbitrary-quality features into two stages: degradation-level prediction and degradation-aware reconstruction. Our method can be applied to pre-trained SAM and SAM2 with only minimal additional learnable parameters, allowing for efficient optimization. Extensive experiments demonstrate that GleSAM++ significantly improves segmentation robustness on complex degradations while maintaining generalization to clear images. Furthermore, GleSAM++ also performs well on unseen degradations, underscoring the versatility of our approach and dataset.

cs.CV↗

arXiv2024

Why mamba is effective? Exploit Linear Transformer-Mamba Network for Multi-Modality Image Fusion

Multi-modality image fusion aims to integrate the merits of images from different sources and render high-quality fusion images. However, existing feature extraction and fusion methods are either constrained by inherent local reduction bias and static parameters during inference (CNN) or limited by quadratic computational complexity (Transformers), and cannot effectively extract and fuse features. To solve this problem, we propose a dual-branch image fusion network called Tmamba. It consists of linear Transformer and Mamba, which has global modeling capabilities while maintaining linear complexity. Due to the difference between the Transformer and Mamba structures, the features extracted by the two branches carry channel and position information respectively. T-M interaction structure is designed between the two branches, using global learnable parameters and convolutional layers to transfer position and channel information respectively. We further propose cross-modal interaction at the attention level to obtain cross-modal attention. Experiments show that our Tmamba achieves promising results in multiple fusion tasks, including infrared-visible image fusion and medical image fusion. Code with checkpoints will be available after the peer-review process.

cs.CV↗

arXiv2024

Dynamic control of luminescence chirality through achiral metasurfaces

Circularly polarized light (CPL) sources are essential for chiroptics, spintronics, quantum optics, and asymmetric photochemistry. However, conventional approaches fail to simultaneously realize a large luminescence dissymmetry factor (glum) and wide-range tuning of glum in a compact device. Chiral luminophores usually suffer from low glum due to their small molecular sizes. Although chiral metasurfaces can enable a large glum, they lack post-fabrication tunability. Here, we demonstrate that it is possible to achieve high-purity circularly polarized luminescence using achiral metasurfaces. These metasurfaces enable optical tuning and even reversal of luminescence chirality by uncovering and utilizing giant near-field chirality. We validate our concept with upconversion nanoparticles and downshifting dye molecules, experimentally achieving a large glum of up to 1.65, which can be actively and continuously tuned between 1.65 and -1.58. Our approach promises important applications in next-generation CPL sources and detectors, and tunable quantum devices.

physics.optics↗

arXiv2023

ShapeMatcher: Self-Supervised Joint Shape Canonicalization, Segmentation, Retrieval and Deformation

In this paper, we present ShapeMatcher, a unified self-supervised learning framework for joint shape canonicalization, segmentation, retrieval and deformation. Given a partially-observed object in an arbitrary pose, we first canonicalize the object by extracting point-wise affine-invariant features, disentangling inherent structure of the object with its pose and size. These learned features are then leveraged to predict semantically consistent part segmentation and corresponding part centers. Next, our lightweight retrieval module aggregates the features within each part as its retrieval token and compare all the tokens with source shapes from a pre-established database to identify the most geometrically similar shape. Finally, we deform the retrieved shape in the deformation module to tightly fit the input object by harnessing part center guided neural cage deformation. The key insight of ShapeMaker is the simultaneous training of the four highly-associated processes: canonicalization, segmentation, retrieval, and deformation, leveraging cross-task consistency losses for mutual supervision. Extensive experiments on synthetic datasets PartNet, ComplementMe, and real-world dataset Scan2CAD demonstrate that ShapeMaker surpasses competitors by a large margin.

cs.CV↗

arXiv2016

Three-dimensional chiral microstructures fabricated by structured optical vortices in isotropic material

Optical vortices, as a kind of structured beam with helical phase wavefronts and doughnut shape intensity distribution, have been used for fabricating chiral structures in metal and spiral patterns in anisotropic polarization-dependent azobenzene polymer. However, in isotropic polymer, the fabricated microstructures are typically confined to non-chiral cylindrical geometry due to two-dimensional doughnut intensity profile of optical vortices. Here we develop a powerful strategy for realizing chiral microstructures in isotropic material by coaxial interference of a vortex beam and a plane wave, which produces three-dimensional (3D) spiral optical fields. This coaxial interference beams are creatively produced by designing the contrivable holograms consisting of azimuthal phase and equiphase loaded on liquid-crystal spatial light modulator. Then, in isotropic polymer, 3D chiral microstructures are achieved under illumination of the coaxial interference femtosecond laser beams with their chirality controlled by the topological charge. Our further investigation reveals that the spiral lobes and chirality are caused by the interfering patterns and helical phase wavefronts, respectively. This technique is simple, stable, and easy-operation, and offers broad applications in optical tweezers, optical communications and fast metamaterial fabrication.

physics.optics↗