arXiv subjects

Lei Sun

Publications and source records attributed to Lei Sun.

At least 73 records · Page 4Linked to original sources

arXiv2026

Video-CoE: Reinforcing Video Event Prediction via Chain of Events

Despite advances in the application of MLLMs for various video tasks, video event prediction (VEP) remains relatively underexplored. VEP requires the model to perform fine-grained temporal modeling of videos and establish logical relationships between videos and future events, which current MLLMs still struggle with. In this work, we first present a comprehensive evaluation of current leading MLLMs on the VEP task, revealing the reasons behind their inaccurate predictions, including lack of logical reasoning ability for future events prediction and insufficient utilization of visual information. To address these challenges, we propose \textbf{C}hain \textbf{o}f \textbf{E}vents (\textbf{CoE}) paradigm, which constructs temporal event chains to implicitly enforce MLLM focusing on the visual content and the logical connections between videos and future events, incentivizing model's reasoning capability with multiple training protocols. Experimental results on public benchmarks demonstrate that our method outperforms both leading open-source and commercial MLLMs, establishing a new state-of-the-art on the VEP task. Codes and models will be released soon.

cs.CV↗

arXiv2026

Towards Universal Computational Aberration Correction in Photographic Cameras: A Comprehensive Benchmark Analysis

Prevalent Computational Aberration Correction (CAC) methods are typically tailored to specific optical systems, leading to poor generalization and labor-intensive re-training for new lenses. Developing CAC paradigms capable of generalizing across diverse photographic lenses offers a promising solution to these challenges. However, efforts to achieve such cross-lens universality within consumer photography are still in their early stages due to the lack of a comprehensive benchmark that encompasses a sufficiently wide range of optical aberrations. Furthermore, it remains unclear which specific factors influence existing CAC methods and how these factors affect their performance. In this paper, we present comprehensive experiments and evaluations involving 24 image restoration and CAC algorithms, utilizing our newly proposed UniCAC, a large-scale benchmark for photographic cameras constructed via automatic optical design. The Optical Degradation Evaluator (ODE) is introduced as a novel framework to objectively assess the difficulty of CAC tasks, offering credible quantification of optical aberrations and enabling reliable evaluation. Drawing on our comparative analysis, we identify three key factors -- prior utilization, network architecture, and training strategy -- that most significantly influence CAC performance, and further investigate their respective effects. We believe that our benchmark, dataset, and observations contribute foundational insights to related areas and lay the groundwork for future investigations. Benchmarks, codes, and Zemax files will be available at https://github.com/XiaolongQian/UniCAC.

cs.CV↗

arXiv2026

Learning Latent Transmission and Glare Maps for Lens Veiling Glare Removal

Beyond the commonly recognized optical aberrations, the imaging performance of simplified optical systems--including single-lens and metalens designs--is often further degraded by veiling glare caused by stray-light scattering from non-ideal optical surfaces and coatings, particularly in complex real-world environments. This compound degradation undermines traditional lens aberration correction yet remains underexplored. A major challenge is that conventional scattering models (e.g., for dehazing) fail to fit veiling glare due to its spatial-varying and depth-independent nature. Consequently, paired high-quality data are difficult to prepare via simulation, hindering application of data-driven veiling glare removal models. To this end, we propose VeilGen, a generative model that learns to simulate veiling glare by estimating its underlying optical transmission and glare maps in an unsupervised manner from target images, regularized by Stable Diffusion (SD)-based priors. VeilGen enables paired dataset generation with realistic compound degradation of optical aberrations and veiling glare, while also providing the estimated latent optical transmission and glare maps to guide the veiling glare removal process. We further introduce DeVeiler, a restoration network trained with a reversibility constraint, which utilizes the predicted latent maps to guide an inverse process of the learned scattering model. Extensive experiments on challenging simplified optical systems demonstrate that our approach delivers superior restoration quality and physical fidelity compared with existing methods. These suggest that VeilGen reliably synthesizes realistic veiling glare, and its learned latent maps effectively guide the restoration process in DeVeiler. All code and datasets will be publicly released at https://github.com/XiaolongQian/DeVeiler.

eess.IV↗

arXiv2026

Layer-wise Instance Binding for Regional and Occlusion Control in Text-to-Image Diffusion Transformers

Region-instructed layout control in text-to-image generation is highly practical, yet existing methods suffer from limitations: (i) training-based approaches inherit data bias and often degrade image quality, and (ii) current techniques struggle with occlusion order, limiting real-world usability. To address these issues, we propose LayerBind. By modeling regional generation as distinct layers and binding them during the generation, our method enables precise regional and occlusion controllability. Our motivation stems from the observation that spatial layout and occlusion are established at a very early denoising stage, suggesting that rearranging the early latent structure is sufficient to modify the final output. Building on this, we structure the scheme into two phases: instance initialization and subsequent semantic nursing. (1) First, leveraging the contextual sharing mechanism in multimodal joint attention, Layer-wise Instance Initialization creates per-instance branches that attend to their own regions while anchoring to the shared background. At a designated early step, these branches are fused according to the layer order to form a unified latent with a pre-established layout. (2) Then, Layer-wise Semantic Nursing reinforces regional details and maintains the occlusion order via a layer-wise attention enhancement. Specifically, a sequential layered attention path operates alongside the standard global path, with updates composited under a layer-transparency scheduler. LayerBind is training-free and plug-and-play, serving as a regional and occlusion controller across Diffusion Transformers. Beyond generation, it natively supports editable workflows, allowing for flexible modifications like changing instances or rearranging visible orders. Both qualitative and quantitative results demonstrate LayerBind's effectiveness, highlighting its strong potential for creative applications.

cs.CV↗

arXiv2026

MultiHaystack: Benchmarking Multimodal Retrieval and Reasoning over 40K Images, Videos, and Documents

Multimodal large language models (MLLMs) achieve strong performance on benchmarks that evaluate text, image, or video understanding separately. However, these settings do not assess a critical real-world requirement, which involves retrieving relevant evidence from large, heterogeneous multimodal corpora prior to reasoning. Most existing benchmarks restrict retrieval to small, single-modality candidate sets, substantially simplifying the search space and overstating end-to-end reliability. To address this gap, we introduce MultiHaystack, the first benchmark designed to evaluate both retrieval and reasoning under large-scale, cross-modal conditions. MultiHaystack comprises over 46,000 multimodal retrieval candidates across documents, images, and videos, along with 747 open yet verifiable questions. Each question is grounded in a unique validated evidence item within the retrieval pool, requiring evidence localization across modalities and fine-grained reasoning. In our study, we find that models perform competitively when provided with the corresponding evidence, but their performance drops sharply when required to retrieve that evidence from the full corpus. Additionally, even the strongest retriever, E5-V, achieves only 40.8% Recall@1, while state-of-the-art MLLMs such as GPT-5 experience a significant drop in reasoning accuracy from 80.86% when provided with the corresponding evidence to 51.4% under top-5 retrieval. These results indicate that multimodal retrieval over heterogeneous pools remains a primary bottleneck for MLLMs, positioning MultiHaystack as a valuable testbed that highlights underlying limitations obscured by small-scale evaluations and promotes retrieval-centric advances in multimodal systems.

cs.CV↗

arXiv2026

OPTIAGENT: A Physics-Driven Agentic Framework for Automated Optical Design

Optical design is the process of configuring optical elements to precisely manipulate light for high-fidelity imaging. It is inherently a highly non-convex optimization problem that relies heavily on human heuristic expertise and domain-specific knowledge. While Large Language Models (LLMs) possess extensive optical knowledge, their capabilities in leveraging the knowledge in designing lens system remain significantly constrained. This work represents the first attempt to employ LLMs in the field of optical design. We bridge the expertise gap by enabling users without formal optical training to successfully develop functional lens systems. Concretely, we curate a comprehensive dataset, named OptiDesignQA, which encompasses both classical lens systems sourced from standard optical textbooks and novel configurations generated by automated design algorithms for training and evaluation. Furthermore, we inject domain-specific optical expertise into the LLM through a hybrid objective of full-system synthesis and lens completion. To align the model with optical principles, we employ Group Relative Policy Optimization Done Right (DrGRPO) guided by Optical Lexicographic Reward for physics-driven policy alignment. This reward system incorporates structural format rewards, physical feasibility rewards, light-manipulation accuracy, and LLM-based heuristics. Finally, our model integrates with specialized optical optimization routines for end-to-end fine-tuning and precision refinement. We benchmark our proposed method against both traditional optimization-based automated design algorithms and LLM counterparts, and experimental results show the superiority of our method.

cs.LG↗

arXiv2026

Gaussian Sequences with Multi-Scale Dynamics for 4D Reconstruction from Monocular Casual Videos

Understanding dynamic scenes from casual videos is critical for scalable robot learning, yet four-dimensional (4D) reconstruction under strictly monocular settings remains highly ill-posed. To address this challenge, our key insight is that real-world dynamics exhibits a multi-scale regularity from object to particle level. To this end, we design the multi-scale dynamics mechanism that factorizes complex motion fields. Within this formulation, we propose Gaussian sequences with multi-scale dynamics, a novel representation for dynamic 3D Gaussians derived through compositions of multi-level motion. This layered structure substantially alleviates ambiguity of reconstruction and promotes physically plausible dynamics. We further incorporate multi-modal priors from vision foundation models to establish complementary supervision, constraining the solution space and improving the reconstruction fidelity. Our approach enables accurate and globally consistent 4D reconstruction from monocular casual videos. Experiments of dynamic novel-view synthesis (NVS) on benchmark and real-world manipulation datasets demonstrate considerable improvements over existing methods.

cs.CV↗

arXiv2026

The faint end of the UV luminosity function at $0.4 < z < 0.7$ from the Hubble Frontier Fields

By extending the Hubble Frontier Fields (HFF) observations to the F225W band using HST WFC3/UVIS, we measure the rest-frame UV luminosity function (LF) of galaxies at $0.4 < z < 0.7$, pushing into the low-luminosity galaxy regime. In this first paper of a series, we describe the HST Cycle-27 GO-15940 F225W observations and data reduction, and present a corresponding catalog for the Abell 2744 field, which is the most data-rich HFF cluster field. Combining deep Near-UV imaging and the high magnification from strong gravitational lensing of the foreground cluster, we identify 152 faint galaxies with $-19.5 < M_{UV} < -12.1$ at $0.4 < z < 0.7$ through hybrid photometric-spectroscopic redshift selection from the Abell 2744 F225W catalog. Using a sample defined by a $50\%$ completeness cut and applying the maximum likelihood estimation, we derive the best-fit Schechter parameters for the UV LF at $z \sim 0.55$ down to $M_\text{UV} < -13.5$ mag, including a faint-end slope of $α= -1.324^{+0.072}_{-0.074}$. We incorporate a curvature parameter $δ$ in parameter estimation to account for a possible turn-over at the faint end of the UV LF, leveraging the exceedingly low luminosities probed by our sample. Our results rule out a turn-over brighter than $M_{UV} = -15.5$ at the $3σ$ confidence level.

astro-ph.GA↗

arXiv2026

OmniLens: Towards Universal Lens Aberration Correction via LensLib-to-Specific Domain Adaptation

Emerging universal Computational Aberration Correction (CAC) paradigms provide an inspiring solution to light-weight and high-quality imaging with a universal model trained on a lens library (LensLib) to address arbitrary lens optical aberrations blindly. However, the limited coverage of existing LensLibs leads to poor generalization of the trained models to unseen lenses, whose fine-tuning pipeline is also confined to the lens-descriptions-known case. In this work, we introduce OmniLens, a flexible solution to universal CAC via (i) establishing a convincing LensLib with comprehensive coverage for pre-training a robust base model, and (ii) adapting the model to any specific lens designs with unknown lens descriptions via fast LensLib-to-specific domain adaptation. To achieve these, an Evolution-based Automatic Optical Design (EAOD) pipeline is proposed to generate a rich variety of lens samples with realistic aberration behaviors. Then, we design an unsupervised regularization term for efficient domain adaptation on a few easily accessible real-captured images based on the statistical observation of dark channel priors in degradation induced by lens aberrations. Extensive experiments demonstrate that the LensLib generated by EAOD effectively develops a universal CAC model with strong generalization capabilities, which can also improve the non-blind lens-specific methods by 0.35~1.81dB in PSNR. Additionally, the proposed domain adaptation method significantly improves the base model, especially in severe aberration cases (at most 2.59dB in PSNR). The code and data will be available at https://github.com/zju-jiangqi/OmniLens.

physics.optics↗

arXiv2026

TEXTS-Diff: TEXTS-Aware Diffusion Model for Real-World Text Image Super-Resolution

Real-world text image super-resolution aims to restore overall visual quality and text legibility in images suffering from diverse degradations and text distortions. However, the scarcity of text image data in existing datasets results in poor performance on text regions. In addition, datasets consisting of isolated text samples limit the quality of background reconstruction. To address these limitations, we construct Real-Texts, a large-scale, high-quality dataset collected from real-world images, which covers diverse scenarios and contains natural text instances in both Chinese and English. Additionally, we propose the TEXTS-Aware Diffusion Model (TEXTS-Diff) to achieve high-quality generation in both background and textual regions. This approach leverages abstract concepts to improve the understanding of textual elements within visual scenes and concrete text regions to enhance textual details. It mitigates distortions and hallucination artifacts commonly observed in text regions, while preserving high-quality visual scene fidelity. Extensive experiments demonstrate that our method achieves state-of-the-art performance across multiple evaluation metrics, exhibiting superior generalization ability and text restoration accuracy in complex scenarios. All the code, model, and dataset will be released.

cs.CV↗

arXiv2026

Tunable spin-phonon polarons in a chiral molecular qubit framework

Chiral structures that produce asymmetric spin-phonon coupling can theoretically generate spin-phonon polarons -- quasiparticles exhibiting non-degenerate spin states with phonon displacements. These quasiparticles are speculated to be the origin of chirality-induced spin selectivity and presumably can display exotic dynamic behaviors. However, direct experimental evidence of spin-phonon polarons has been lacking. Using a chiral molecular qubit framework embedding stable semiquinone-like radicals, we report spin dynamic signatures that indicate the formation of spin-phonon polarons for the first time. Our non-adiabatic model reveals that these quasiparticles introduce an active spin relaxation channel when polaron reorganization energy approaches Zeeman splitting. This new channel manifests itself as anomalous, temperature-independent spin relaxation, which can be suppressed by high magnetic fields or pore-filling solvents (e.g. CH2Cl2, CS2). Such field- and guest-tunable relaxation is unattainable in conventional spin systems. Harnessing this mechanism could boost repetition rates in spin-based quantum information technologies without compromising coherence or quantum sensing performance.

cond-mat.mes-hall↗

arXiv2025

Real-world Reinforcement Learning from Suboptimal Interventions

Real-world reinforcement learning (RL) offers a promising approach to training precise and dexterous robotic manipulation policies in an online manner, enabling robots to learn from their own experience while gradually reducing human labor. However, prior real-world RL methods often assume that human interventions are optimal across the entire state space, overlooking the fact that even expert operators cannot consistently provide optimal actions in all states or completely avoid mistakes. Indiscriminately mixing intervention data with robot-collected data inherits the sample inefficiency of RL, while purely imitating intervention data can ultimately degrade the final performance achievable by RL. The question of how to leverage potentially suboptimal and noisy human interventions to accelerate learning without being constrained by them thus remains open. To address this challenge, we propose SiLRI, a state-wise Lagrangian reinforcement learning algorithm for real-world robot manipulation tasks. Specifically, we formulate the online manipulation problem as a constrained RL optimization, where the constraint bound at each state is determined by the uncertainty of human interventions. We then introduce a state-wise Lagrange multiplier and solve the problem via a min-max optimization, jointly optimizing the policy and the Lagrange multiplier to reach a saddle point. Built upon a human-as-copilot teleoperation system, our algorithm is evaluated through real-world experiments on diverse manipulation tasks. Experimental results show that SiLRI effectively exploits human suboptimal interventions, reducing the time required to reach a 90% success rate by at least 50% compared with the state-of-the-art RL method HIL-SERL, and achieving a 100% success rate on long-horizon manipulation tasks where other RL methods struggle to succeed. Project website: https://silri-rl.github.io/.

cs.RO↗

arXiv2025

FLUX-Text: A Simple and Advanced Diffusion Transformer Baseline for Scene Text Editing

Scene text editing aims to modify or add texts on images while ensuring text fidelity and overall visual quality consistent with the background. Recent methods are primarily built on UNet-based diffusion models, which have improved scene text editing results, but still struggle with complex glyph structures, especially for non-Latin ones (\eg, Chinese, Korean, Japanese). To address these issues, we present \textbf{FLUX-Text}, a simple and advanced multilingual scene text editing DiT method. Specifically, our FLUX-Text enhances glyph understanding and generation through lightweight Visual and Text Embedding Modules, while preserving the original generative capability of FLUX. We further propose a Regional Text Perceptual Loss tailored for text regions, along with a matching two-stage training strategy to better balance text editing and overall image quality. Benefiting from the DiT-based architecture and lightweight feature injection modules, FLUX-Text can be trained with only $0.1$M training examples, a \textbf{97\%} reduction compared to $2.9$M required by popular methods. Extensive experiments on multiple public datasets, including English and Chinese benchmarks, demonstrate that our method surpasses other methods in visual quality and text fidelity. All the code is available at https://github.com/AMAP-ML/FluxText.

cs.CV↗

arXiv2025

SCALAR: Scale-wise Controllable Visual Autoregressive Learning

Controllable image synthesis, which enables fine-grained control over generated outputs, has emerged as a key focus in visual generative modeling. However, controllable generation remains challenging for Visual Autoregressive (VAR) models due to their hierarchical, next-scale prediction style. Existing VAR-based methods often suffer from inefficient control encoding and disruptive injection mechanisms that compromise both fidelity and efficiency. In this work, we present SCALAR, a controllable generation method based on VAR, incorporating a novel Scale-wise Conditional Decoding mechanism. SCALAR leverages a pretrained image encoder to extract semantic control signal encodings, which are projected into scale-specific representations and injected into the corresponding layers of the VAR backbone. This design provides persistent and structurally aligned guidance throughout the generation process. Building on SCALAR, we develop SCALAR-Uni, a unified extension that aligns multiple control modalities into a shared latent space, supporting flexible multi-conditional guidance in a single model. Extensive experiments show that SCALAR achieves superior generation quality and control precision across various tasks. The code is released at https://github.com/AMAP-ML/SCALAR.

cs.CV↗

arXiv2025

Variational Calculations of the Excited States of the Charged NV-center in Diamond Using a Hybrid Functional

The excited electronic states involved in the optical cycle preparation of a pure spin state of the negatively charged NV-defect in diamond are calculated using the HSE06 hybrid density functional and variational optimization of the orbitals. This includes the energy of the excited triplet as well as the two lowest singlet states with respect to the ground triplet state. In addition to the vertical excitation, the effect of structural relaxation is also estimated using analytical atomic forces. The lowering of the energy in the triplet excited state and the resulting zero-phonon line triplet excitation energy are both within 0.1 eV of the experimental estimates. An analogous relaxation in the lower energy singlet state using spin purified atomic forces is estimated to be 0.06 eV. These results, obtained with a hybrid density functional, improve on previously published results using local and semi-local functionals, which are known to underestimate the band gap. The good agreement with experimental estimates demonstrates how time-independent variational calculations of excited states using density functionals can give accurate results and, thereby, provide a powerful screening tool for identifying other defect systems as candidates for quantum technologies.

cond-mat.mtrl-sci↗

arXiv2025

Neuro-inspired automated lens design

The highly non-convex optimization landscape of modern lens design necessitates extensive human expertise, resulting in inefficiency and constrained design diversity. While automated methods are desirable, existing approaches remain limited to simple tasks or produce complex lenses with suboptimal image quality. Drawing inspiration from the synaptic pruning mechanism in mammalian neural development, this study proposes OptiNeuro--a novel automated lens design framework that first generates diverse initial structures and then progressively eliminates low-performance lenses while refining remaining candidates through gradient-based optimization. By fully automating the design of complex aspheric imaging lenses, OptiNeuro demonstrates quasi-human-level performance, identifying multiple viable candidates with minimal human intervention. This advancement not only enhances the automation level and efficiency of lens design but also facilitates the exploration of previously uncharted lens architectures.

physics.optics↗

arXiv2025

Federated Split Learning for Resource-Constrained Robots in Industrial IoT: Framework Comparison, Optimization Strategies, and Future Directions

Federated split learning (FedSL) has emerged as a promising paradigm for enabling collaborative intelligence in industrial Internet of Things (IoT) systems, particularly in smart factories where data privacy, communication efficiency, and device heterogeneity are critical concerns. In this article, we present a comprehensive study of FedSL frameworks tailored for resource-constrained robots in industrial scenarios. We compare synchronous, asynchronous, hierarchical, and heterogeneous FedSL frameworks in terms of workflow, scalability, adaptability, and limitations under dynamic industrial conditions. Furthermore, we systematically categorize token fusion strategies into three paradigms: input-level (pre-fusion), intermediate-level (intra-fusion), and output-level (post-fusion), and summarize their respective strengths in industrial applications. We also provide adaptive optimization techniques to enhance the efficiency and feasibility of FedSL implementation, including model compression, split layer selection, computing frequency allocation, and wireless resource management. Simulation results validate the performance of these frameworks under industrial detection scenarios. Finally, we outline open issues and research directions of FedSL in future smart manufacturing systems.

cs.RO↗

arXiv2025

DriveAction: A Benchmark for Exploring Human-like Driving Decisions in VLA Models

Vision-Language-Action (VLA) models have advanced autonomous driving, but existing benchmarks still lack scenario diversity, reliable action-level annotation, and evaluation protocols aligned with human preferences. To address these limitations, we introduce DriveAction, the first action-driven benchmark specifically designed for VLA models, comprising 16,185 QA pairs generated from 2,610 driving scenarios. DriveAction leverages real-world driving data proactively collected by drivers of autonomous vehicles to ensure broad and representative scenario coverage, offers high-level discrete action labels collected directly from drivers' actual driving operations, and implements an action-rooted tree-structured evaluation framework that explicitly links vision, language, and action tasks, supporting both comprehensive and task-specific assessment. Our experiments demonstrate that state-of-the-art vision-language models (VLMs) require both vision and language guidance for accurate action prediction: on average, accuracy drops by 3.3% without vision input, by 4.1% without language input, and by 8.0% without either. Our evaluation supports precise identification of model bottlenecks with robust and consistent results, thus providing new insights and a rigorous foundation for advancing human-like decisions in autonomous driving.

cs.CV↗