arXiv subjects

Xiaoyu Zheng

Publications and source records attributed to Xiaoyu Zheng.

At least 19 recordsLinked to original sources

arXiv2026

Logographic Character Visual Pretraining via Semantic-based Contrastive Learning

Current deep learning-based character vision studies, e.g., text recognition, character image denoising, and historical text completion, are offering new solutions for learning, managing, and utilizing character resources. However, the performance of these studies peaks only with large and balanced datasets, which is a rarity with real-world character datasets, especially for logographic character languages, e.g., Chinese. The imbalance in data distribution of logographic characters is a common issue due to differences in character usage frequency and new characters being continuously created. In this paper, we propose a novel method for logographic character recognition, which introduces a multi-modal learning approach using visual semantics and contextual semantics of characters. A novel pre-training strategy is designed to enhance deep visual representations, especially for datasets suffering from issues of imbalanced and rare instances, by extracting the contextual semantics of each character from the corresponding language models. We conduct experiments across various datasets to evaluate our character recognition method and further validate the contrastive pre-training strategy by several downstream tasks. Experimental results demonstrate the superiority of our method compared to state-of-the-art methods.

cs.CV↗

arXiv2026

Crayotter: Traceable Multi-Agent Workflows for Long-Form Video Editing

Long-form video editing over heterogeneous footage requires agents to coordinate source selection, multimodal analysis, timeline construction, narration and subtitle alignment, rendering, and revision while exposing intermediate state for inspection and repair. We present Crayotter, an open-source multimodal multi-agent demo system for prompt-driven long-form video editing. Crayotter organizes production around coverage-aware material preparation, artifact-grounded editing research, and tool-grounded timeline execution. Across these stages, retrieval reports, video analyses, editing blueprints, scheduler events, tool calls, intermediate renders, and final exports are treated as first-class artifacts rather than hidden transient state. The workbench supports local assets, agent-assisted retrieval, progress monitoring, artifact preview, failure diagnosis, interrupted-job resumption, and resource-aware asynchronous execution for long-running workflows. In a 23-theme evaluation, Crayotter achieves the highest human overall score (3.40/5) among the compared systems, with its largest margins in theme alignment, narrative coherence, and editing smoothness. These results show that long-horizon video editing agents can be made traceable, inspectable, and practically controllable through observable production artifacts. Code, traces, and examples are publicly available at https://github.com/idwts/Crayotter.

cs.CV↗

arXiv2026

Towards a Frugal Photosynthesis Sensing Toolkit for Data-Driven Plant Science Education and Exploration

Rapid environmental change and advances in data-driven analysis highlight the need not only to use computational tools, but also to foster understanding of the natural world and inspire creativity. Photosynthesis, the process that fuels nearly all life on Earth, provides a compelling context for such learning, particularly in understanding how plants alter their photosynthetic strategies in response to environmental changes. However, existing tools for studying photosynthesis are often inaccessible or limited to demonstrating its presence, rather than capturing its temporal dynamics. We present PhytoBits, a frugal in situ gas-exchange sensing toolkit for distinguishing and teaching photosynthetic strategies. PhytoBits combines leaf enclosure with accessible materials, an off-the-shelf CO2 sensor, and a low-cost microcontroller, to support multi-day monitoring of plant gas-exchange in educational and research contexts. We validated PhytoBits against research-grade gas-exchange systems, confirming that it identifies C3 and CAM (Crassulacean Acid Metabolism) photosynthetic pathways. In addition to obligate CAM, PhytoBits also resolves facultative CAM and developmental CAM dynamics in plants. This work presents an early-stage hardware validation; user deployment studies, open-source code dissemination, and automated pathway classification are planned as future work.

cs.HC↗

arXiv2026

Task-guided Spatiotemporal Network with Diffusion Augmentation for EEG-based Dementia Diagnosis and MMSE Prediction

Patients with dementia typically exhibit cognitive impairment, which is routinely assessed using the Mini-Mental State Examination (MMSE). Concurrently, their underlying neurophysiological abnormalities are reflected in Electroencephalography (EEG), providing a basis for joint modeling. However, traditional multi-task approaches suffer from feature entanglement, which leads to inter-task interference when handling heterogeneous objectives.To address this challenge, we propose a task-guided spatiotemporal network (TGSN) with diffusion augmentation for EEG-based dementia diagnosis and MMSE prediction. Specifically, TGSN integrates a multi-band feature fusion module to capture complementary spectral information from EEG. Meanwhile, a pre-trained data augmentation module utilizing a diffusion process is introduced toincrease sample diversity. To model the complex spatiotemporal patterns of EEG, we propose a gated spatiotemporal attention module that captures long-range spatial dependencies and temporal dynamics. Moreover, we design a task-guided query module to achieve task-specific feature extraction, thereby mitigating task interference. The effectiveness of TGSN is evaluated on the XY02 dataset. Experimental results demonstrate that the proposed network outperforms several state-of-the-art methods, achieving classification accuracies of 97.78\% for Alzheimer's Disease (AD)/Frontotemporal Dementia (FTD) and 83.93\% for AD/FTD/Vascular Cognitive Impairment (VCI), which exceed the best baselines by 16.39\% and 8.28\%, respectively. In parallel, it reduces the RMSE for MMSE prediction to 1.93 and 2.38, achieving significant error reductions of 1.44 and 1.43 compared to the best baselines. Additionally, validation on the DS004504 dataset demonstrates strong cross-dataset generalization...

cs.LG↗

arXiv2026

DiEC: Diffusion Embedded Clustering

Deep clustering methods typically rely on a single, well-defined representation for clustering. In contrast, pretrained diffusion models provide abundant and diverse multi-scale representations across network layers and noise timesteps. However, a key challenge is how to efficiently identify the most clustering-friendly representation in the layer*timestep space. To address this issue, we propose Diffusion Embedded Clustering (DiEC), an unsupervised framework that performs clustering by leveraging optimal intermediate representations from pretrained diffusion models. DiEC systematically evaluates the clusterability of representations along the trajectory of network depth and noise timesteps. Meanwhile, an unsupervised search strategy is designed for recognizing the Clustering-optimal Layer (COL) and Clustering-optimal Timestep (COT) in the layer*timestep space of pretrained diffusion models, aiming to promote clustering performance and reduce computational overhead. DiEC is fine-tuned primarily with a structure-preserving DEC-style KL-divergence objective at the fixed COL + COT, together with a random-timestep diffusion denoising objective to maintain the generative capability of the pretrained model. Without relying on augmentation-based consistency constraints or contrastive learning, DiEC achieves excellent clustering performance across multiple benchmark datasets.

cs.LG↗

arXiv2026

On extracting momentum from supports: the bullet in the beam problem

To gain insights into momentum transfer from the supporting environment, we consider the simple problem of a bullet, fired from below, into a wooden beam resting on two supports. The resulting upward velocity of the beam strongly depends on where the bullet enters the beam; this dependence is due to upward momentum extracted by the beam from its supports. Our simple example illustrates the momentum transfer mechanism exploited in the remarkable dynamics of the chain fountain.

physics.class-ph↗

arXiv2025

Beyond H&E: Unlocking Pathological Insights with Polarization Imaging

Histopathology image analysis is fundamental to digital pathology, with hematoxylin and eosin (H&E) staining as the gold standard for diagnostic and prognostic assessments. While H&E imaging effectively highlights cellular and tissue structures, it lacks sensitivity to birefringence and tissue anisotropy, which are crucial for assessing collagen organization, fiber alignment, and microstructural alterations--key indicators of tumor progression, fibrosis, and other pathological conditions. To bridge this gap, we construct a polarization imaging system and curate a new dataset of over 13,000 paired Polar-H&E images. Visualizations of polarization properties reveal distinctive optical signatures in pathological tissues, underscoring its diagnostic value. Building on this dataset, we propose PolarHE, a dual-modality fusion framework that integrates H&E with polarization imaging, leveraging the latter ability to enhance tissue characterization. Our approach employs a feature decomposition strategy to disentangle common and modality specific features, ensuring effective multimodal representation learning. Through comprehensive validation, our approach significantly outperforms previous methods, achieving an accuracy of 86.70% on the Chaoyang dataset and 89.06% on the MHIST dataset. These results demonstrate that polarization imaging is a powerful and underutilized modality in computational pathology, enriching feature representation and improving diagnostic accuracy. PolarHE establishes a promising direction for multimodal learning, paving the way for more interpretable and generalizable pathology models.

eess.IV↗

arXiv2025

OpenUS: A Fully Open-Source Foundation Model for Ultrasound Image Analysis via Self-Adaptive Masked Contrastive Learning

Ultrasound (US) is one of the most widely used medical imaging modalities, thanks to its low cost, portability, real-time feedback, and absence of ionizing radiation. However, US image interpretation remains highly operator-dependent and varies significantly across anatomical regions, acquisition protocols, and device types. These variations, along with unique challenges such as speckle, low contrast, and limited standardized annotations, hinder the development of generalizable, label-efficient ultrasound AI models. In this paper, we propose OpenUS, the first reproducible, open-source ultrasound foundation model built on a large collection of public data. OpenUS employs a vision Mamba backbone, capturing both local and global long-range dependencies across the image. To extract rich features during pre-training, we introduce a novel self-adaptive masking framework that combines contrastive learning with masked image modeling. This strategy integrates the teacher's attention map with student reconstruction loss, adaptively refining clinically-relevant masking to enhance pre-training effectiveness. OpenUS also applies a dynamic learning schedule to progressively adjust the difficulty of the pre-training process. To develop the foundation model, we compile the largest to-date public ultrasound dataset comprising over 308K images from 42 publicly available datasets, covering diverse anatomical regions, institutions, imaging devices, and disease types. Our pre-trained OpenUS model can be easily adapted to specific downstream tasks by serving as a backbone for label-efficient fine-tuning. Code is available at https://github.com/XZheng0427/OpenUS.

cs.CV↗

arXiv2025

Joint Attention Mechanism Learning to Facilitate Opto-physiological Monitoring during Physical Activity

Opto-physiological monitoring including photoplethysmography (PPG) provides non-invasive cardiac and respiratory measurements, yet motion artefacts (MAs) during physical activity degrade its signal quality and downstream estimation concurrently. An attention-mechanism-based generative adversarial network (AM-GAN) was proposed to model motion artefacts and mitigate their impact on raw PPG signals. The AM-GAN learns how to transform motion-affected PPG into artefact-reduced waveforms to align with triaxial acceleration signals corresponding to artefact components gained from a triaxial accelerometer. The AM-GAN has been validated across four experimental protocols with 43 participants performing activities from low to high intensity (6--12km/h). With the public datasets, the AM-GAN achieves mean absolute error (MAE) for heart rate (HR) of 1.81 beats/min on IEEE-SPC and 3.86 beats/min on PPGDalia. On the in-house LU dataset, it shows the MAEs < 1.37 beats/min for HR and 2.49 breaths/min for respiratory rate (RR). A further in-house C2 dataset with three oxygen levels (16%, 18%, and 21%) was applied in the AM-GAN to attain a MAE of 1.65% for SpO2. The outcome demonstrates that the AM-GAN offers a robust and reliable physiological estimation under various intensities of physical activity.

eess.SP↗

arXiv2025

On quantifying the spin angular momentum density of light

In addition to energy, light carries linear and angular momentum. These are key quantities in rapidly developing optics research and in technologies focusing on light induced forces and torques on materials. Spin angular momentum (SAM) density is of particular interest, since unlike orbital angular momentum, it is uncoupled from linear momentum. The SAM density of light was first estimated in 1909 by Poynting, using a mechanical analogy. Exact expressions, based on results from quantum mechanics and field theory were subsequently developed, and are in common use today. In this paper, we show that the SAM density of light can be obtained directly from the Coulomb force and Maxwell's equations, without reliance on quantum mechanics or field theories; it could have been calculated by Maxwell and his contemporaries. Besides its historical significance, the simple derivation of our result makes it readily accessible to non-experts in the field.

physics.optics↗

arXiv2025

Intelligent Task Management via Dynamic Multi-region Division in LEO Satellite Networks

As a key complement to terrestrial networks and a fundamental component of future 6G systems, Low Earth Orbit (LEO) satellite networks are expected to provide high-quality communication services when integrated with ground-based infrastructure, thereby attracting significant research interest. However, the limited satellite onboard resources and the uneven distribution of computational workloads often result in congestion along inter-satellite links (ISLs) that degrades task processing efficiency. Effectively managing the dynamic and large-scale topology of LEO networks to ensure balanced task distribution remains a critical challenge. To this end, we propose a dynamic multi-region division framework for intelligent task management in LEO satellite networks. This framework optimizes both intra- and inter-region routing to minimize task delay while balancing the utilization of computational and communication resources. Based on this framework, we propose a dynamic multi-region division algorithm based on the Genetic Algorithm (GA), which adaptively adjusts the size of each region based on the workload status of individual satellites. Additionally, we incorporate an adaptive routing algorithm and a task splitting and offloading scheme based on Multi-Agent Deep Deterministic Policy Gradient (MA-DDPG) to effectively accommodate the arriving tasks. Simulation results demonstrate that our proposed framework outperforms comparative methods in terms of the task delay, energy consumption per task, and task completion rate.

cs.DC↗

arXiv2025

GTransPDM: A Graph-embedded Transformer with Positional Decoupling for Pedestrian Crossing Intention Prediction

Understanding and predicting pedestrian crossing behavioral intention is crucial for the driving safety of autonomous vehicles. Nonetheless, challenges emerge when using promising images or environmental context masks to extract various factors for time-series network modeling, causing pre-processing errors or a loss of efficiency. Typically, pedestrian positions captured by onboard cameras are often distorted and do not accurately reflect their actual movements. To address these issues, GTransPDM -- a Graph-embedded Transformer with a Position Decoupling Module -- was developed for pedestrian crossing intention prediction by leveraging multi-modal features. First, a positional decoupling module was proposed to decompose pedestrian lateral motion and encode depth cues in the image view. Then, a graph-embedded Transformer was designed to capture the spatio-temporal dynamics of human pose skeletons, integrating essential factors such as position, skeleton, and ego-vehicle motion. Experimental results indicate that the proposed method achieves 92% accuracy on the PIE dataset and 87% accuracy on the JAAD dataset, with a processing speed of 0.05ms. It outperforms the state-of-the-art in comparison.

cs.CV↗

arXiv2025

XFMamba: Cross-Fusion Mamba for Multi-View Medical Image Classification

Compared to single view medical image classification, using multiple views can significantly enhance predictive accuracy as it can account for the complementarity of each view while leveraging correlations between views. Existing multi-view approaches typically employ separate convolutional or transformer branches combined with simplistic feature fusion strategies. However, these approaches inadvertently disregard essential cross-view correlations, leading to suboptimal classification performance, and suffer from challenges with limited receptive field (CNNs) or quadratic computational complexity (transformers). Inspired by state space sequence models, we propose XFMamba, a pure Mamba-based cross-fusion architecture to address the challenge of multi-view medical image classification. XFMamba introduces a novel two-stage fusion strategy, facilitating the learning of single-view features and their cross-view disparity. This mechanism captures spatially long-range dependencies in each view while enhancing seamless information transfer between views. Results on three public datasets, MURA, CheXpert and DDSM, illustrate the effectiveness of our approach across diverse multi-view medical image classification tasks, showing that it outperforms existing convolution-based and transformer-based multi-view methods. Code is available at https://github.com/XZheng0427/XFMamba.

cs.CV↗

arXiv2025

Diffusion-based Virtual Staining from Polarimetric Mueller Matrix Imaging

Polarization, as a new optical imaging tool, has been explored to assist in the diagnosis of pathology. Moreover, converting the polarimetric Mueller Matrix (MM) to standardized stained images becomes a promising approach to help pathologists interpret the results. However, existing methods for polarization-based virtual staining are still in the early stage, and the diffusion-based model, which has shown great potential in enhancing the fidelity of the generated images, has not been studied yet. In this paper, a Regulated Bridge Diffusion Model (RBDM) for polarization-based virtual staining is proposed. RBDM utilizes the bidirectional bridge diffusion process to learn the mapping from polarization images to other modalities such as H\&E and fluorescence. And to demonstrate the effectiveness of our model, we conduct the experiment on our manually collected dataset, which consists of 18,000 paired polarization, fluorescence and H\&E images, due to the unavailability of the public dataset. The experiment results show that our model greatly outperforms other benchmark methods. Our dataset and code will be released upon acceptance.

eess.IV↗

arXiv2024

The effects of network architecture on the photomechanical performance of azo-acrylate liquid crystal elastomers

Azo-containing liquid crystal elatomers are photomechanical materials which can be actuated by illumination. The photomechanical response is a result of the photoisomerization of the azo moiety, which produces bulk stresses in the material. These stresses arise via two distinct and competing mechanisms: order parameter change induced stress and direct contractile stress. We describe thermomechanical and photomechanical experiments aimed at assessing the relative contributions of these. we discuss our results and summarize our findings.

cond-mat.soft↗

arXiv2024

TextHawk: Exploring Efficient Fine-Grained Perception of Multimodal Large Language Models

Multimodal Large Language Models (MLLMs) have shown impressive results on various multimodal tasks. However, most existing MLLMs are not well suited for document-oriented tasks, which require fine-grained image perception and information compression. In this paper, we present TextHawk, a MLLM that is specifically designed for document-oriented tasks, while preserving the general capabilities of MLLMs. TextHawk is aimed to explore efficient fine-grained perception by designing four dedicated components. Firstly, a ReSampling and ReArrangement (ReSA) module is proposed to reduce the redundancy in the document texts and lower the computational cost of the MLLM. We explore encoding the positions of each local feature by presenting Scalable Positional Embeddings (SPEs), which can preserve the scalability of various image sizes. A Query Proposal Network (QPN) is then adopted to initialize the queries dynamically among different sub-images. To further enhance the fine-grained visual perceptual ability of the MLLM, we design a Multi-Level Cross-Attention (MLCA) mechanism that captures the hierarchical structure and semantic relations of document images. Furthermore, we create a new instruction-tuning dataset for document-oriented tasks by enriching the multimodal document data with Gemini Pro. We conduct extensive experiments on both general and document-oriented MLLM benchmarks, and show that TextHawk outperforms the state-of-the-art methods, demonstrating its effectiveness and superiority in fine-grained document perception and general abilities.

cs.CV↗

arXiv2023

A confined rod: mean field theory for hard rod-like particles

In this paper, we model the configurations of a system of hard rods by viewing each rod in a cell formed by its neighbors. By minimizing the free energy in the model and performing molecular dynamics, where, in both cases, the shape of the cell is a free parameter, we obtain the equilibrium orientational order parameter, free energy and pressure of the system. Our model enables the calculation of anisotropic stresses exerted on the walls of the cell due to shape change of the rod in photoisomerization. These results are a key step towards understanding molecular shape change effects in photomechanical systems under illumination.

cond-mat.soft↗

arXiv2022

Optical characterization of dyed liquid crystal cells

The guest-host liquid crystal display, first proposed in 1968, relies on controlling the orientation of dichroic dyes dissolved in a nematic liquid crystal host. Controlling the orientation of the liquid crystal and of the dissolved dye with an electric field allows control of the transmittance of the cell. Knowing the dielectric properties at optical frequencies of the dye and liquid crystal mixtures is crucial for the optimal design of guest-host liquid crystal devices. In this work, the dielectric functions of various layers in liquid crystal cells are described by models obeying the Kramers-Kronig relations: the Sellmeier equation for transparent layers and causal Gaussian oscillator model for absorbing layers. We propose a systematic way to accurately model the dielectric response of each layer by minimizing the sum of squared differences between the measured transmittance spectrum of a guest-host cell in the near-UV/vis range and the prediction of the transmittance of the modeled multilayer structure. By measuring the transmittance for incident light polarized parallel and perpendicular to the nematic director allows us to separately characterize the two principal dielectric functions of the uniaxial sample. Our results show that the causal Gaussian oscillator model can accurately characterize the dielectric functions of dyes in liquid crystals.

cond-mat.soft↗