arXiv subjects

Shuyu Wang

Publications and source records attributed to Shuyu Wang.

6 recordsLinked to original sources

arXiv2026

TextWand: A Unified Framework for Scene Text Editing

We propose TextWand, a general-purpose framework that unifies scene text removal, generation, and replacement into a single model. By decomposing complex editing tasks into the atomic primitives of rendering and erasure, TextWand achieves precise control over both text appearance and background integrity. Specifically, we introduce a novel design, Overlay-Reference Positional Encoding (ORPE), to enforce pixel-level layout fidelity and exemplar-driven style control, alongside a new strategy, Region-Adaptive Suppression (RAS), to ensure clean text erasure. To address the absence of a comprehensive benchmark for general-purpose scene text editing among existing single-task datasets, we construct TextWand-Bench. Extensive experiments demonstrate that TextWand outperforms existing leading open-source and closed-source models by delivering superior text content accuracy, layout and style consistency, and overall image quality across scene text removal, generation and replacement tasks.

cs.CV↗

arXiv2026

Phase spirals induced by the gas warp

The discovery of the phase space spirals in the Solar neighborhood in Gaia Data Release 2 has prompted various attempts to understand their origin. A source of bending waves, which has been neglected as a cause of the phase spiral, is irregular gas inflow along the warp. We aim to study whether perturbations by the gas warp could induce phase spirals. Accounting for this additional formation scenario for phase spirals could improve our current understanding of the perturbation history of the Milky Way disc. We use two N-body + SPH (Smooth Particle Hydrodynamics) simulations of an isolated galaxy to search for, and study, warp-induced phase spirals. We study the emergence and propagation of the detected phase spirals using Fourier decomposition. We detect strong one-armed phase spirals in the warped simulation. These phase spirals are prevalent and persist over ~10 Gyr. The morphology of these phase spirals varies with location and evolves with time. In particular, the emergence rate of the phase spiral evolves with the gas inflow at the outer disc and the bending wave amplitude, indicating that these phase spirals are a record of warp-induced bending waves. We find that these phase spirals can reach amplitudes comparable to those in the Gaia DR3. We only detect weak and stochastically distributed phase spirals in an unwarped control simulation. We conclude that phase spirals can be induced by the irregular gas accretion along the warp. These phase spirals occur globally and are long-lived.

astro-ph.GA↗

arXiv2025

MIND-Edit: MLLM Insight-Driven Editing via Language-Vision Projection

Recent advances in AI-generated content (AIGC) have significantly accelerated image editing techniques, driving increasing demand for diverse and fine-grained edits. Despite these advances, existing image editing methods still face challenges in achieving high precision and semantic accuracy in complex scenarios. Recent studies address this issue by incorporating multimodal large language models (MLLMs) into image editing pipelines. However, current MLLM-based methods mainly rely on interpreting textual instructions, leaving the intrinsic visual understanding of large models largely unexplored, thus resulting in insufficient alignment between textual semantics and visual outcomes. To overcome these limitations, we propose MIND-Edit, an end-to-end image-editing framework integrating pretrained diffusion model with MLLM. MIND-Edit introduces two complementary strategies: (1) a text instruction optimization strategy that clarifies ambiguous user instructions based on semantic reasoning from the MLLM, and (2) an MLLM insight-driven editing strategy that explicitly leverages the intrinsic visual understanding capability of the MLLM to infer editing intent and guide the diffusion process via generated visual embeddings. Furthermore, we propose a joint training approach to effectively integrate both strategies, allowing them to reinforce each other for more accurate instruction interpretation and visually coherent edits aligned with user intent. Extensive experiments demonstrate that MIND-Edit outperforms state-of-the-art image editing methods in both quantitative metrics and visual quality, particularly under complex and challenging scenarios.

cs.CV↗

arXiv2025

Study of the influence of a biased database on the prediction of standard algorithms for selecting the best candidate for an interview

Artificial intelligence is used at various stages of the recruitment process to automatically select the best candidate for a position, with companies guaranteeing unbiased recruitment. However, the algorithms used are either trained by humans or are based on learning from past experiences that were biased. In this article, we propose to generate data mimicking external (discrimination) and internal biases (self-censorship) in order to train five classic algorithms and to study the extent to which they do or do not find the best candidates according to objective criteria. In addition, we study the influence of the anonymisation of files on the quality of predictions.

cs.AI↗

arXiv2023

High-Resolution Chemical Abundances of the Nyx Stream

Nyx is a nearby, prograde, and high-eccentricity stellar stream physically contained in the thick disk but with an unknown origin. Nyx could be the remnant of a disrupted dwarf galaxy, in which case the associated dark matter substructure could affect terrestrial dark matter direct detection experiments. Alternatively, Nyx could be a signature of the Milky Way's disk formation and evolution. To determine the origin of Nyx, we obtained high-resolution spectroscopy of 34 Nyx stars using Keck/HIRES and Magellan/MIKE. A differential chemical abundance analysis shows that most Nyx stars reside in a metal-rich ($\mbox{[Fe/H]} > -1$) high-$α$ component that is chemically indistinguishable from the thick disk. This rules out an originally suggested scenario that Nyx is the remnant of a single massive dwarf galaxy merger. However, we also identify five substantially more metal-poor stars ($\mbox{[Fe/H]} \sim -2.0$) that have chemical abundances similar to the metal-weak thick disk. It remains unclear how stars chemically identical to the thick disk can be on such prograde, high-eccentricity orbits. We suggest two most likely scenarios: that Nyx is the result of an early minor dwarf galaxy merger or that it is a record of the early spin-up of the Milky Way disk -- although neither perfectly reproduces the chemodynamic observations. The most likely formation scenarios suggest that future spectroscopic surveys should find Nyx-like structures outside of the Solar Neighborhood.

astro-ph.GA↗

arXiv2023

Spatial-Temporal Data Mining for Ocean Science: Data, Methodologies, and Opportunities

With the rapid amassing of spatial-temporal (ST) ocean data, many spatial-temporal data mining (STDM) studies have been conducted to address various oceanic issues, including climate forecasting and disaster warning. Compared with typical ST data (e.g., traffic data), ST ocean data is more complicated but with unique characteristics, e.g., diverse regionality and high sparsity. These characteristics make it difficult to design and train STDM models on ST ocean data. To the best of our knowledge, a comprehensive survey of existing studies remains missing in the literature, which hinders not only computer scientists from identifying the research issues in ocean data mining but also ocean scientists to apply advanced STDM techniques. In this paper, we provide a comprehensive survey of existing STDM studies for ocean science. Concretely, we first review the widely-used ST ocean datasets and highlight their unique characteristics. Then, typical ST ocean data quality enhancement techniques are explored. Next, we classify existing STDM studies in ocean science into four types of tasks, i.e., prediction, event detection, pattern mining, and anomaly detection, and elaborate on the techniques for these tasks. Finally, promising research opportunities are discussed. This survey can help scientists from both computer science and ocean science better understand the fundamental concepts, key techniques, and open challenges of STDM for ocean science.

cs.LG↗