arXiv subjects

Wangmeng Zuo

Publications and source records attributed to Wangmeng Zuo.

At least 91 records · Page 5Linked to original sources

arXiv2025

Enhanced Generative Structure Prior for Chinese Text Image Super-resolution

Faithful text image super-resolution (SR) is challenging because each character has a unique structure and usually exhibits diverse font styles and layouts. While existing methods primarily focus on English text, less attention has been paid to more complex scripts like Chinese. In this paper, we introduce a high-quality text image SR framework designed to restore the precise strokes of low-resolution (LR) Chinese characters. Unlike methods that rely on character recognition priors to regularize the SR task, we propose a novel structure prior that offers structure-level guidance to enhance visual quality. Our framework incorporates this structure prior within a StyleGAN model, leveraging its generative capabilities for restoration. To maintain the integrity of character structures while accommodating various font styles and layouts, we implement a codebook-based mechanism that restricts the generative space of StyleGAN. Each code in the codebook represents the structure of a specific character, while the vector $w$ in StyleGAN controls the character's style, including typeface, orientation, and location. Through the collaborative interaction between the codebook and style, we generate a high-resolution structure prior that aligns with LR characters both spatially and structurally. Experiments demonstrate that this structure prior provides robust, character-specific guidance, enabling the accurate restoration of clear strokes in degraded characters, even for real-world LR Chinese text with irregular layouts. Our code and pre-trained models will be available at https://github.com/csxmli2016/MARCONetPlusPlus

cs.CV↗

arXiv2025

AR-GRPO: Training Autoregressive Image Generation Models via Reinforcement Learning

Inspired by the success of reinforcement learning (RL) in refining large language models (LLMs), we propose AR-GRPO, an approach to integrate online RL training into autoregressive (AR) image generation models. We adapt the Group Relative Policy Optimization (GRPO) algorithm to refine the vanilla autoregressive models' outputs by carefully designed reward functions that evaluate generated images across multiple quality dimensions, including perceptual quality, realism, and semantic fidelity. We conduct comprehensive experiments on both class-conditional (i.e., class-to-image) and text-conditional (i.e., text-to-image) image generation tasks, demonstrating that our RL-enhanced framework significantly improves both the image quality and human preference of generated images compared to the standard AR baselines. Our results show consistent improvements across various evaluation metrics, establishing the viability of RL-based optimization for AR image generation and opening new avenues for controllable and high-quality image synthesis. The source codes and models are available at: https://github.com/Kwai-Klear/AR-GRPO.

cs.CV↗

arXiv2025

AI paradigm for solving differential equations: first-principles data generation and scale-dilation operator AI solver

Many problems are governed by differential equations (DEs). Artificial intelligence (AI) is a new path for solving DEs. However, data is very scarce and existing AI solvers struggle with approximation of high frequency components (AHFC). We propose an AI paradigm for solving diverse DEs, including DE-ruled first-principles data generation methodology and scale-dilation operator (SDO) AI solver. Using either prior knowledge or random fields, we generate solutions and then substitute them into the DEs to derive the sources and initial/boundary conditions through balancing DEs, thus producing arbitrarily vast amount of, first-principles-consistent training datasets at extremely low computational cost. We introduce a reversible SDO that leverages the Fourier transform of the multiscale solutions to fix AHFC, and design a spatiotemporally coupled, attention-based Transformer AI solver of DEs with SDO. An upper bound on the Hessian condition number of the loss function is proven to be proportional to the squared 2-norm of the solution gradient, revealing that SDO yields a smoother loss landscape, consequently fixing AHFC with efficient training. Extensive tests on diverse DEs demonstrate that our AI paradigm achieves consistently superior accuracy over state-of-the-art methods. This work makes AI solver of DEs to be truly usable in broad nature and engineering fields.

cs.LG↗

arXiv2025

QR-LoRA: Efficient and Disentangled Fine-tuning via QR Decomposition for Customized Generation

Existing text-to-image models often rely on parameter fine-tuning techniques such as Low-Rank Adaptation (LoRA) to customize visual attributes. However, when combining multiple LoRA models for content-style fusion tasks, unstructured modifications of weight matrices often lead to undesired feature entanglement between content and style attributes. We propose QR-LoRA, a novel fine-tuning framework leveraging QR decomposition for structured parameter updates that effectively separate visual attributes. Our key insight is that the orthogonal Q matrix naturally minimizes interference between different visual features, while the upper triangular R matrix efficiently encodes attribute-specific transformations. Our approach fixes both Q and R matrices while only training an additional task-specific $ΔR$ matrix. This structured design reduces trainable parameters to half of conventional LoRA methods and supports effective merging of multiple adaptations without cross-contamination due to the strong disentanglement properties between $ΔR$ matrices. Experiments demonstrate that QR-LoRA achieves superior disentanglement in content-style fusion tasks, establishing a new paradigm for parameter-efficient, disentangled fine-tuning in generative models. The project page is available at: https://luna-ai-lab.github.io/QR-LoRA/.

cs.CV↗

arXiv2025

Rethinking Cross-Modal Interaction in Multimodal Diffusion Transformers

Multimodal Diffusion Transformers (MM-DiTs) have achieved remarkable progress in text-driven visual generation. However, even state-of-the-art MM-DiT models like FLUX struggle with achieving precise alignment between text prompts and generated content. We identify two key issues in the attention mechanism of MM-DiT, namely 1) the suppression of cross-modal attention due to token imbalance between visual and textual modalities and 2) the lack of timestep-aware attention weighting, which hinder the alignment. To address these issues, we propose \textbf{Temperature-Adjusted Cross-modal Attention (TACA)}, a parameter-efficient method that dynamically rebalances multimodal interactions through temperature scaling and timestep-dependent adjustment. When combined with LoRA fine-tuning, TACA significantly enhances text-image alignment on the T2I-CompBench benchmark with minimal computational overhead. We tested TACA on state-of-the-art models like FLUX and SD3.5, demonstrating its ability to improve image-text alignment in terms of object appearance, attribute binding, and spatial relationships. Our findings highlight the importance of balancing cross-modal attention in improving semantic fidelity in text-to-image diffusion models. Our codes are publicly available at \href{https://github.com/Vchitect/TACA}

cs.CV↗

arXiv2025

VitaGlyph: Vitalizing Artistic Typography with Flexible Dual-branch Diffusion Models

Artistic typography is a technique to visualize the meaning of input character in an imaginable and readable manner. With powerful text-to-image diffusion models, existing methods directly design the overall geometry and texture of input character, making it challenging to ensure both creativity and legibility. In this paper, we introduce a dual-branch, training-free method called VitaGlyph, enabling flexible artistic typography with controllable geometry changes while maintaining the readability. The key insight of VitaGlyph is to treat input character as a scene composed of a Subject and its Surrounding, which are rendered with varying degrees of geometric transformation. To enhance the visual appeal and creativity of the generated artistic typography, the subject flexibly expresses the essential concept of the input character, while the surrounding enriches relevant background without altering the shape, thus maintaining overall readability. Specifically, we implement VitaGlyph through a three-phase framework: (i) Knowledge Acquisition leverages large language models to design text descriptions for the subject and surrounding. (ii) Regional Interpretation detects the part that most closely matches the subject description and refines the structure via Semantic Typography. (iii) Attentional Compositional Generation separately renders the textures of the Subject and Surrounding regions and blends them in an attention-based manner. Experimental results demonstrate that VitaGlyph not only achieves better artistry and readability but also manages to depict multiple customized concepts, facilitating more creative and pleasing artistic typography generation. Our code will be made publicly available.

cs.CV↗

arXiv2025

LoViC: Efficient Long Video Generation with Context Compression

Despite recent advances in diffusion transformers (DiTs) for text-to-video generation, scaling to long-duration content remains challenging due to the quadratic complexity of self-attention. While prior efforts -- such as sparse attention and temporally autoregressive models -- offer partial relief, they often compromise temporal coherence or scalability. We introduce LoViC, a DiT-based framework trained on million-scale open-domain videos, designed to produce long, coherent videos through a segment-wise generation process. At the core of our approach is FlexFormer, an expressive autoencoder that jointly compresses video and text into unified latent representations. It supports variable-length inputs with linearly adjustable compression rates, enabled by a single query token design based on the Q-Former architecture. Additionally, by encoding temporal context through position-aware mechanisms, our model seamlessly supports prediction, retradiction, interpolation, and multi-shot generation within a unified paradigm. Extensive experiments across diverse tasks validate the effectiveness and versatility of our approach.

cs.CV↗

arXiv2025

RefSTAR: Blind Facial Image Restoration with Reference Selection, Transfer, and Reconstruction

Blind facial image restoration is highly challenging due to unknown complex degradations and the sensitivity of humans to faces. Although existing methods introduce auxiliary information from generative priors or high-quality reference images, they still struggle with identity preservation problems, mainly due to improper feature introduction on detailed textures. In this paper, we focus on effectively incorporating appropriate features from high-quality reference images, presenting a novel blind facial image restoration method that considers reference selection, transfer, and reconstruction (RefSTAR). In terms of selection, we construct a reference selection (RefSel) module. For training the RefSel module, we construct a RefSel-HQ dataset through a mask generation pipeline, which contains annotating masks for 10,000 ground truth-reference pairs. As for the transfer, due to the trivial solution in vanilla cross-attention operations, a feature fusion paradigm is designed to force the features from the reference to be integrated. Finally, we propose a reference image reconstruction mechanism that further ensures the presence of reference image features in the output image. The cycle consistency loss is also redesigned in conjunction with the mask. Extensive experiments on various backbone models demonstrate superior performance, showing better identity preservation ability and reference feature transfer quality. Source code, dataset, and pre-trained models are available at https://github.com/yinzhicun/RefSTAR.

cs.CV↗

arXiv2025

Motion-Aware Adaptive Pixel Pruning for Efficient Local Motion Deblurring

Local motion blur in digital images originates from the relative motion between dynamic objects and static imaging systems during exposure. Existing deblurring methods face significant challenges in addressing this problem due to their inefficient allocation of computational resources and inadequate handling of spatially varying blur patterns. To overcome these limitations, we first propose a trainable mask predictor that identifies blurred regions in the image. During training, we employ blur masks to exclude sharp regions. For inference optimization, we implement structural reparameterization by converting $3\times 3$ convolutions to computationally efficient $1\times 1$ convolutions, enabling pixel-level pruning of sharp areas to reduce computation. Second, we develop an intra-frame motion analyzer that translates relative pixel displacements into motion trajectories, establishing adaptive guidance for region-specific blur restoration. Our method is trained end-to-end using a combination of reconstruction loss, reblur loss, and mask loss guided by annotated blur masks. Extensive experiments demonstrate superior performance over state-of-the-art methods on both local and global blur datasets while reducing FLOPs by 49\% compared to SOTA models (e.g., LMD-ViT). The source code is available at https://github.com/shangwei5/M2AENet.

cs.CV↗

arXiv2025

BMMR: A Large-Scale Bilingual Multimodal Multi-Discipline Reasoning Dataset

In this paper, we introduce BMMR, a large-scale bilingual, multimodal, multi-disciplinary reasoning dataset for the community to develop and evaluate large multimodal models (LMMs). BMMR comprises 110k college-level questions spanning 300 UNESCO-defined subjects, spanning diverse formats-multiple-choice, fill-in-the-blank, and open-ended QA-and sourced from both print and digital media such as books, exams, and quizzes. All data are curated and filtered via a human-in-the-loop and scalable framework, and each instance is paired with a high-quality reasoning path. The dataset is organized into two parts: BMMR-Eval that comprises 20,458 high-quality instances to comprehensively assess LMMs' knowledge and reasoning across multiple disciplines in both Chinese and English; and BMMR-Train that contains 88,991 instances to support further research and development, extending the current focus on mathematical reasoning to diverse disciplines and domains. In addition, we propose the process-based multi-discipline verifier (i.e., BMMR-Verifier) for accurate and fine-grained evaluation of reasoning paths. Extensive experiments on 24 models reveal that (i) even SOTA models (e.g., o3 and Gemini-2.5-Pro) leave substantial headroom on BMMR-Eval; (ii) reasoning models exhibit discipline bias and outperform LMMs only on specific subjects; (iii) open-source models still trail their proprietary counterparts; and (iv) fine-tuning on BMMR-Train narrows this gap. Additionally, we conduct reasoning-chain analyses using BMMR-Verifier and other in-depth studies, uncovering the challenges LMMs currently face in multidisciplinary reasoning. We will release the data, and we hope our work can offer insights and contributions to the community.

cs.CL↗

arXiv2025

Reblurring-Guided Single Image Defocus Deblurring: A Learning Framework with Misaligned Training Pairs

For single image defocus deblurring, acquiring well-aligned training pairs (or training triplets), i.e., a defocus blurry image, an all-in-focus sharp image (and a defocus blur map), is a challenging task for developing effective deblurring models. Existing image defocus deblurring methods typically rely on training data collected by specialized imaging equipment, with the assumption that these pairs or triplets are perfectly aligned. However, in practical scenarios involving the collection of real-world data, direct acquisition of training triplets is infeasible, and training pairs inevitably encounter spatial misalignment issues. In this work, we introduce a reblurring-guided learning framework for single image defocus deblurring, enabling the learning of a deblurring network even with misaligned training pairs. By reconstructing spatially variant isotropic blur kernels, our reblurring module ensures spatial consistency between the deblurred image, the reblurred image and the input blurry image, thereby addressing the misalignment issue while effectively extracting sharp textures from the all-in-focus sharp image. Moreover, spatially variant blur can be derived from the reblurring module, and serve as pseudo supervision for defocus blur map during training, interestingly transforming training pairs into training triplets. To leverage this pseudo supervision, we propose a lightweight defocus blur estimator coupled with a fusion block, which enhances deblurring performance through seamless integration with state-of-the-art deblurring networks. Additionally, we have collected a new dataset for single image defocus deblurring (SDD) with typical misalignments, which not only validates our proposed method but also serves as a benchmark for future research.

cs.CV↗

arXiv2025

Position: Intelligent Science Laboratory Requires the Integration of Cognitive and Embodied AI

Scientific discovery has long been constrained by human limitations in expertise, physical capability, and sleep cycles. The recent rise of AI scientists and automated laboratories has accelerated both the cognitive and operational aspects of research. However, key limitations persist: AI systems are often confined to virtual environments, while automated laboratories lack the flexibility and autonomy to adaptively test new hypotheses in the physical world. Recent advances in embodied AI, such as generalist robot foundation models, diffusion-based action policies, fine-grained manipulation learning, and sim-to-real transfer, highlight the promise of integrating cognitive and embodied intelligence. This convergence opens the door to closed-loop systems that support iterative, autonomous experimentation and the possibility of serendipitous discovery. In this position paper, we propose the paradigm of Intelligent Science Laboratories (ISLs): a multi-layered, closed-loop framework that deeply integrates cognitive and embodied intelligence. ISLs unify foundation models for scientific reasoning, agent-based workflow orchestration, and embodied agents for robust physical experimentation. We argue that such systems are essential for overcoming the current limitations of scientific discovery and for realizing the full transformative potential of AI-driven science.

cs.AI↗

arXiv2025

Text to Image for Multi-Label Image Recognition with Joint Prompt-Adapter Learning

Benefited from image-text contrastive learning, pre-trained vision-language models, e.g., CLIP, allow to direct leverage texts as images (TaI) for parameter-efficient fine-tuning (PEFT). While CLIP is capable of making image features to be similar to the corresponding text features, the modality gap remains a nontrivial issue and limits image recognition performance of TaI. Using multi-label image recognition (MLR) as an example, we present a novel method, called T2I-PAL to tackle the modality gap issue when using only text captions for PEFT. The core design of T2I-PAL is to leverage pre-trained text-to-image generation models to generate photo-realistic and diverse images from text captions, thereby reducing the modality gap. To further enhance MLR, T2I-PAL incorporates a class-wise heatmap and learnable prototypes. This aggregates local similarities, making the representation of local visual features more robust and informative for multi-label recognition. For better PEFT, we further combine both prompt tuning and adapter learning to enhance classification performance. T2I-PAL offers significant advantages: it eliminates the need for fully semantically annotated training images, thereby reducing the manual annotation workload, and it preserves the intrinsic mode of the CLIP model, allowing for seamless integration with any existing CLIP framework. Extensive experiments on multiple benchmarks, including MS-COCO, VOC2007, and NUS-WIDE, show that our T2I-PAL can boost recognition performance by 3.47% in average above the top-ranked state-of-the-art methods.

cs.CV↗

arXiv2025

Image Demoiréing Using Dual Camera Fusion on Mobile Phones

When shooting electronic screens, moiré patterns usually appear in captured images, which seriously affects the image quality. Existing image demoiréing methods face great challenges in removing large and heavy moiré. To address the issue, we propose to utilize Dual Camera fusion for Image Demoiréing (DCID), \ie, using the ultra-wide-angle (UW) image to assist the moiré removal of wide-angle (W) image. This is inspired by two motivations: (1) the two lenses are commonly equipped with modern smartphones, (2) the UW image generally can provide normal colors and textures when moiré exists in the W image mainly due to their different focal lengths. In particular, we propose an efficient DCID method, where a lightweight UW image encoder is integrated into an existing demoiréing network and a fast two-stage image alignment manner is present. Moreover, we construct a large-scale real-world dataset with diverse mobile phones and monitors, containing about 9,000 samples. Experiments on the dataset show our method performs better than state-of-the-art methods. Code and dataset are available at https://github.com/Mrduckk/DCID.

cs.CV↗

arXiv2025

Application of convolutional neural networks in image super-resolution

Due to strong learning abilities of convolutional neural networks (CNNs), they have become mainstream methods for image super-resolution. However, there are big differences of different deep learning methods with different types. There is little literature to summarize relations and differences of different methods in image super-resolution. Thus, summarizing these literatures are important, according to loading capacity and execution speed of devices. This paper first introduces principles of CNNs in image super-resolution, then introduces CNNs based bicubic interpolation, nearest neighbor interpolation, bilinear interpolation, transposed convolution, sub-pixel layer, meta up-sampling for image super-resolution to analyze differences and relations of different CNNs based interpolations and modules, and compare performance of these methods by experiments. Finally, this paper gives potential research points and drawbacks and summarizes the whole paper, which can facilitate developments of CNNs in image super-resolution.

cs.CV↗

arXiv2025

Voyager: Long-Range and World-Consistent Video Diffusion for Explorable 3D Scene Generation

Real-world applications like video gaming and virtual reality often demand the ability to model 3D scenes that users can explore along custom camera trajectories. While significant progress has been made in generating 3D objects from text or images, creating long-range, 3D-consistent, explorable 3D scenes remains a complex and challenging problem. In this work, we present Voyager, a novel video diffusion framework that generates world-consistent 3D point-cloud sequences from a single image with user-defined camera path. Unlike existing approaches, Voyager achieves end-to-end scene generation and reconstruction with inherent consistency across frames, eliminating the need for 3D reconstruction pipelines (e.g., structure-from-motion or multi-view stereo). Our method integrates three key components: 1) World-Consistent Video Diffusion: A unified architecture that jointly generates aligned RGB and depth video sequences, conditioned on existing world observation to ensure global coherence 2) Long-Range World Exploration: An efficient world cache with point culling and an auto-regressive inference with smooth video sampling for iterative scene extension with context-aware consistency, and 3) Scalable Data Engine: A video reconstruction pipeline that automates camera pose estimation and metric depth prediction for arbitrary videos, enabling large-scale, diverse training data curation without manual 3D annotations. Collectively, these designs result in a clear improvement over existing methods in visual quality and geometric accuracy, with versatile applications.

cs.CV↗

arXiv2025

MIRAGE: Assessing Hallucination in Multimodal Reasoning Chains of MLLM

Multimodal hallucination in multimodal large language models (MLLMs) restricts the correctness of MLLMs. However, multimodal hallucinations are multi-sourced and arise from diverse causes. Existing benchmarks fail to adequately distinguish between perception-induced hallucinations and reasoning-induced hallucinations. This failure constitutes a significant issue and hinders the diagnosis of multimodal reasoning failures within MLLMs. To address this, we propose the {\dataset} benchmark, which isolates reasoning hallucinations by constructing questions where input images are correctly perceived by MLLMs yet reasoning errors persist. {\dataset} introduces multi-granular evaluation metrics: accuracy, factuality, and LLMs hallucination score for hallucination quantification. Our analysis reveals that (1) the model scale, data scale, and training stages significantly affect the degree of logical, fabrication, and factual hallucinations; (2) current MLLMs show no effective improvement on spatial hallucinations caused by misinterpreted spatial relationships, indicating their limited visual reasoning capabilities; and (3) question types correlate with distinct hallucination patterns, highlighting targeted challenges and potential mitigation strategies. To address these challenges, we propose {\method}, a method that combines curriculum reinforcement fine-tuning to encourage models to generate logic-consistent reasoning chains by stepwise reducing learning difficulty, and collaborative hint inference to reduce reasoning complexity. {\method} establishes a baseline on {\dataset}, and reduces the logical hallucinations in original base models.

cs.CV↗

arXiv2025

Don't Let Your Robot be Harmful: Responsible Robotic Manipulation via Safety-as-Policy

Unthinking execution of human instructions in robotic manipulation can lead to severe safety risks, such as poisonings, fires, and even explosions. In this paper, we present responsible robotic manipulation, which requires robots to consider potential hazards in the real-world environment while completing instructions and performing complex operations safely and efficiently. However, such scenarios in real world are variable and risky for training. To address this challenge, we propose Safety-as-policy, which includes (i) a world model to automatically generate scenarios containing safety risks and conduct virtual interactions, and (ii) a mental model to infer consequences with reflections and gradually develop the cognition of safety, allowing robots to accomplish tasks while avoiding dangers. Additionally, we create the SafeBox synthetic dataset, which includes one hundred responsible robotic manipulation tasks with different safety risk scenarios and instructions, effectively reducing the risks associated with real-world experiments. Experiments demonstrate that Safety-as-policy can avoid risks and efficiently complete tasks in both synthetic dataset and real-world experiments, significantly outperforming baseline methods. Our SafeBox dataset shows consistent evaluation results with real-world scenarios, serving as a safe and effective benchmark for future research.

cs.RO↗