Donglin Di

He Feng, Yongjia Ma, Donglin Di et al.

Portrait animation methods have achieved substantial visual quality and lip synchronization, but fine-grained manipulation of the eye region still faces a trade-off between input granularity and motion accuracy. Existing methods using emotion labels or coarse text prompts are insufficient for describing subtle ocular dynamics, whereas approaches based on Action Units or driving videos provide higher fidelity at the cost of a heavier input burden. These limitations are still restrictive for beyond-emotion states (e.g., thinking) and drowsiness. In light of the above, we propose CogPortrait, a two-stage framework that generates portrait animations from high-level labels. In the first stage, three chain-of-thought Multimodal Large Language Models (MLLMs) agents compile high-level labels into facial keypoints through temporal event planning, prototype retrieval, and composition from a real-behavior library, and semantic-physiological constraint enforcement. In the second stage, a DiT-based video generation backbone synthesizes the final animation conditioned on the keypoints, reference portrait, audio, and text prompt, enhanced by a dynamic classifier-free guidance strategy with eye-region-aware reweighting and KTO-based refinement for boundary cases. We further introduce the EMH benchmark covering diverse emotions and beyond-emotion categories with two AU-level metrics for evaluating fine-grained eye-region and head-motion control. Extensive experiments on HDTF and the EMH benchmark demonstrate that CogPortrait achieves more precise eye-region control than existing methods while maintaining supe- rior visual quality and identity consistency

Xiangchen Yin, Donglin Di, Lei Fan et al.

Recent methods using diffusion models have made significant progress in human image generation with various control signals such as pose priors. However, existing efforts are still struggling to generate high-quality images with consistent pose alignment, resulting in unsatisfactory output. In this paper, we propose a framework that delves into the graph relations of pose priors to provide control information for human image generation. The main idea is to establish a graph topological structure between the pose priors and latent representation of diffusion models to capture the intrinsic associations between different pose parts. A Progressive Graph Integrator (PGI) is designed to learn the spatial relationships of the pose priors with the graph structure, adopting a hierarchical strategy within an Adapter to gradually propagate information across different pose parts. Besides, a pose perception loss is introduced based on a pretrained pose estimation network to minimize the pose differences. Extensive qualitative and quantitative experiments conducted on the Human-Art and LAION-Human datasets clearly demonstrate that our model can achieve significant performance improvement over the latest benchmark models. The code is available at \url{https://xiangchenyin.github.io/GRPose/}.

Minghui Wu, Chenxu Zhao, Anyang Su et al.

Understanding of video creativity and content often varies among individuals, with differences in focal points and cognitive levels across different ages, experiences, and genders. There is currently a lack of research in this area, and most existing benchmarks suffer from several drawbacks: 1) a limited number of modalities and answers with restrictive length; 2) the content and scenarios within the videos are excessively monotonous, transmitting allegories and emotions that are overly simplistic. To bridge the gap to real-world applications, we introduce a large-scale Subjective Response Indicators for Advertisement Videos dataset, namely SRI-ADV. Specifically, we collected real changes in Electroencephalographic (EEG) and eye-tracking regions from different demographics while they viewed identical video content. Utilizing this multi-modal dataset, we developed tasks and protocols to analyze and evaluate the extent of cognitive understanding of video content among different users. Along with the dataset, we designed a Hypergraph Multi-modal Large Language Model (HMLLM) to explore the associations among different demographics, video elements, EEG, and eye-tracking indicators. HMLLM could bridge semantic gaps across rich modalities and integrate information beyond different modalities to perform logical reasoning. Extensive experimental evaluations on SRI-ADV and other additional video-based generative performance benchmarks demonstrate the effectiveness of our method. The codes and dataset will be released at https://github.com/mininglamp-MLLM/HMLLM.

Tonghua Su, Fuxiang Yang, Xiang Zhou et al.

In this work, we propose a task called "Scene Style Text Editing (SSTE)", changing the text content as well as the text style of the source image while keeping the original text scene. Existing methods neglect to fine-grained adjust the style of the foreground text, such as its rotation angle, color, and font type. To tackle this task, we propose a quadruple framework named "QuadNet" to embed and adjust foreground text styles in the latent feature space. Specifically, QuadNet consists of four parts, namely background inpainting, style encoder, content encoder, and fusion generator. The background inpainting erases the source text content and recovers the appropriate background with a highly authentic texture. The style encoder extracts the style embedding of the foreground text. The content encoder provides target text representations in the latent feature space to implement the content edits. The fusion generator combines the information yielded from the mentioned parts and generates the rendered text images. Practically, our method is capable of performing promisingly on real-world datasets with merely string-level annotation. To the best of our knowledge, our work is the first to finely manipulate the foreground text content and style by deeply semantic editing in the latent feature space. Extensive experiments demonstrate that QuadNet has the ability to generate photo-realistic foreground text and avoid source text shadows in real-world scenes when editing text content.

Moule Lin, Weipeng Jing, Donglin Di et al.

Hyperspectral images have significant applications in various domains, since they register numerous semantic and spatial information in the spectral band with spatial variability of spectral signatures. Two critical challenges in identifying pixels of the hyperspectral image are respectively representing the correlated information among the local and global, as well as the abundant parameters of the model. To tackle this challenge, we propose a Multi-Scale U-shape Multi-Layer Perceptron (MUMLP) a model consisting of the designed MSC (Multi-Scale Channel) block and the UMLP (U-shape Multi-Layer Perceptron) structure. MSC transforms the channel dimension and mixes spectral band feature to embed the deep-level representation adequately. UMLP is designed by the encoder-decoder structure with multi-layer perceptron layers, which is capable of compressing the large-scale parameters. Extensive experiments are conducted to demonstrate our model can outperform state-of-the-art methods across-the-board on three wide-adopted public datasets, namely Pavia University, Houston 2013 and Houston 2018

Lei Fan, Yiwen Ding, Dongdong Fan et al.

Cereal grains are a vital part of human diets and are important commodities for people's livelihood and international trade. Grain Appearance Inspection (GAI) serves as one of the crucial steps for the determination of grain quality and grain stratification for proper circulation, storage and food processing, etc. GAI is routinely performed manually by qualified inspectors with the aid of some hand tools. Automated GAI has the benefit of greatly assisting inspectors with their jobs but has been limited due to the lack of datasets and clear definitions of the tasks. In this paper we formulate GAI as three ubiquitous computer vision tasks: fine-grained recognition, domain adaptation and out-of-distribution recognition. We present a large-scale and publicly available cereal grains dataset called GrainSpace. Specifically, we construct three types of device prototypes for data acquisition, and a total of 5.25 million images determined by professional inspectors. The grain samples including wheat, maize and rice are collected from five countries and more than 30 regions. We also develop a comprehensive benchmark based on semi-supervised learning and self-supervised learning techniques. To the best of our knowledge, GrainSpace is the first publicly released dataset for cereal grain inspection.

Chaofan Luo, Donglin Di, Xun Yang et al.

Despite significant strides in the field of 3D scene editing, current methods encounter substantial challenge, particularly in preserving 3D consistency in multi-view editing process. To tackle this challenge, we propose a progressive 3D editing strategy that ensures multi-view consistency via a Trajectory-Anchored Scheme (TAS) with a dual-branch editing mechanism. Specifically, TAS facilitates a tightly coupled iterative process between 2D view editing and 3D updating, preventing error accumulation yielded from text-to-image process. Additionally, we explore the relationship between optimization-based methods and reconstruction-based methods, offering a unified perspective for selecting superior design choice, supporting the rationale behind the designed TAS. We further present a tuning-free View-Consistent Attention Control (VCAC) module that leverages cross-view semantic and geometric reference from the source branch to yield aligned views from the target branch during the editing of 2D views. To validate the effectiveness of our method, we analyze 2D examples to demonstrate the improved consistency with the VCAC module. Further extensive quantitative and qualitative results in text-guided 3D scene editing indicate that our method achieves superior editing quality compared to state-of-the-art methods. We will make the complete codebase publicly available following the conclusion of the review process.

Yizhi Zhang, Lei Fan, Zhulin Tao et al.

Universal, transferable whole-slide image (WSI) representations are central to computational pathology. Incorporating multiple markers (e.g., immunohistochemistry, IHC) alongside H&E enriches H&E-based features with diverse, biologically meaningful information. However, progress is limited by the scarcity of well-aligned multi-stain datasets. Inter-stain misalignment shifts corresponding tissue across slides, hindering consistent patch-level features and degrading slide-level embeddings. To address this, we curated a slide-level aligned, five-stain dataset (H&E, HER2, KI67, ER, PGR) to enable paired H&E-IHC learning and robust cross-stain representation. Leveraging this dataset, we propose Cross-Stain Contrastive Learning (CSCL), a two-stage pretraining framework with a lightweight adapter trained using patch-wise contrastive alignment to improve the compatibility of H&E features with corresponding IHC-derived contextual cues, and slide-level representation learning with Multiple Instance Learning (MIL), which uses a cross-stain attention fusion module to integrate stain-specific patch features and a cross-stain global alignment module to enforce consistency among slide-level embeddings across different stains. Experiments on cancer subtype classification, IHC biomarker status classification, and survival prediction show consistent gains, yielding high-quality, transferable H&E slide-level representations. The code and data are available at https://github.com/lily-zyz/CSCL.

Baorui Ma, Jiahui Yang, Donglin Di et al.

Scaling has powered recent advances in vision foundation models, yet extending this paradigm to metric depth estimation remains challenging due to heterogeneous sensor noise, camera-dependent biases, and metric ambiguity in noisy cross-source 3D data. We introduce Metric Anything, a simple and scalable pretraining framework that learns metric depth from noisy, diverse 3D sources without manually engineered prompts, camera-specific modeling, or task-specific architectures. Central to our approach is the Sparse Metric Prompt, created by randomly masking depth maps, which serves as a universal interface that decouples spatial reasoning from sensor and camera biases. Using about 20M image-depth pairs spanning reconstructed, captured, and rendered 3D data across 10000 camera models, we demonstrate-for the first time-a clear scaling trend in the metric depth track. The pretrained model excels at prompt-driven tasks such as depth completion, super-resolution and Radar-camera fusion, while its distilled prompt-free student achieves state-of-the-art results on monocular depth estimation, camera intrinsics recovery, single/multi-view metric 3D reconstruction, and VLA planning. We also show that using pretrained ViT of Metric Anything as a visual encoder significantly boosts Multimodal Large Language Model capabilities in spatial intelligence. These results show that metric depth estimation can benefit from the same scaling laws that drive modern foundation models, establishing a new path toward scalable and efficient real-world metric perception. We open-source MetricAnything at http://metric-anything.github.io/metric-anything-io/ to support community research.

Donglin Di, He Feng, Wenzhang Sun et al.

Human-centric generative models are becoming increasingly popular, giving rise to various innovative tools and applications, such as talking face videos conditioned on text or audio prompts. The core of these capabilities lies in powerful pre-trained foundation models, trained on large-scale, high-quality datasets. However, many advanced methods rely on in-house data subject to various constraints, and other current studies fail to generate high-resolution face videos, which is mainly attributed to the significant lack of large-scale, high-quality face video datasets. In this paper, we introduce a human face video dataset, \textbf{DH-FaceVid-1K}. Our collection spans 1,200 hours in total, encompassing 270,043 video clips from over 20,000 individuals. Each sample includes corresponding speech audio, facial keypoints, and text annotations. Compared to other publicly available datasets, ours distinguishes itself through its multi-ethnic coverage and high-quality, comprehensive individual attributes. We establish multiple face video generation models supporting tasks such as text-to-video and image-to-video generation. In addition, we develop comprehensive benchmarks to validate the scaling law when using different proportions of proposed dataset. Our primary aim is to contribute a face video dataset, particularly addressing the underrepresentation of Asian faces in existing curated datasets and thereby enriching the global spectrum of face-centric data and mitigating demographic biases. \textbf{Project Page:} https://luna-ai-lab.github.io/DH-FaceVid-1K/

Yun Liang, Zhiguang Hu, Junjie Huang et al.

Current unsupervised anomaly detection approaches perform well on public datasets but struggle with specific anomaly types due to the domain gap between pre-trained feature extractors and target-specific domains. To tackle this issue, this paper presents a two-stage training strategy, called \textbf{ToCoAD}. In the first stage, a discriminative network is trained by using synthetic anomalies in a self-supervised learning manner. This network is then utilized in the second stage to provide a negative feature guide, aiding in the training of the feature extractor through bootstrap contrastive learning. This approach enables the model to progressively learn the distribution of anomalies specific to industrial datasets, effectively enhancing its generalizability to various types of anomalies. Extensive experiments are conducted to demonstrate the effectiveness of our proposed two-stage training strategy, and our model produces competitive performance, achieving pixel-level AUROC scores of 98.21\%, 98.43\% and 97.70\% on MVTec AD, VisA and BTAD respectively.

Fuxiang Yang, Donglin Di, Lulu Tang et al.

Vision-Language-Action (VLA) models are a promising path toward embodied intelligence, yet they often overlook the predictive and temporal-causal structure underlying visual dynamics. World-model VLAs address this by predicting future frames, but waste capacity reconstructing redundant backgrounds. Latent-action VLAs encode frame-to-frame transitions compactly, but lack temporally continuous dynamic modeling and world knowledge. To overcome these limitations, we introduce CoWVLA (Chain-of-World VLA), a new "Chain of World" paradigm that unifies world-model temporal reasoning with a disentangled latent motion representation. First, a pretrained video VAE serves as a latent motion extractor, explicitly factorizing video segments into structure and motion latents. Then, during pre-training, the VLA learns from an instruction and an initial frame to infer a continuous latent motion chain and predict the segment's terminal frame. Finally, during co-fine-tuning, this latent dynamic is aligned with discrete action prediction by jointly modeling sparse keyframes and action sequences in a unified autoregressive decoder. This design preserves the world-model benefits of temporal reasoning and world knowledge while retaining the compactness and interpretability of latent actions, enabling efficient visuomotor learning. Extensive experiments on robotic simulation benchmarks show that CoWVLA outperforms existing world-model and latent-action approaches and achieves moderate computational efficiency, highlighting its potential as a more effective VLA pretraining paradigm. The project website can be found at https://fx-hit.github.io/cowvla-io.

Jintao Sun, Hu Zhang, Donglin Di et al.

Vision-Language models (VLMs) have demonstrated remarkable capability in ground-view visual understanding but often fracture when deployed on high-altitude Unmanned Aerial Vehicles (UAVs). The failure largely stems from a pronounced domain shift, characterized by tiny and densely packed objects, repetitive textures, and ambiguous top-down orientations. These factors severely disrupt semantic grounding and hinder both spatial reasoning and controllable generation. To bridge this critical gap, we introduce UAVReason, the first unified large-scale multi-modal benchmark dedicated to nadir-view UAV scenarios, derived from a high-fidelity UAV simulation platform. In contrast to existing UAV benchmarks, which are largely siloed and focus on single tasks like object detection or segmentation, UAVReason uniquely consolidates over 273K Visual Question Answering (VQA) pairs, including 23.6K single frames with detailed captions, 68.2K 2-frame temporal sequences, and 188.8K cross-modal generation samples. The benchmark probes 22 diverse reasoning types across spatial and temporal axes while simultaneously evaluating high-fidelity generation across RGB, depth, and segmentation modalities. We further establish a strong, unified baseline model via multi-task learning. Extensive experiments validate the efficacy of our unified approach across diverse metrics, such as EM/F1 for VQA, mIoU for segmentation, and CLIP Score for generation. These results indicate limitations of general-domain vision-language models and show that unified multi-task learning substantially improves UAV-native performance. All data, code, and evaluation tools will be publicly released to advance UAV multimodal research.

He Feng, Donglin Di, Yongjia Ma et al.

The objective of face animation is to generate dynamic and expressive talking head videos from a single reference face, utilizing driving conditions derived from either video or audio inputs. Current approaches often require fine-tuning for specific identities and frequently fail to produce expressive videos due to the limited effectiveness of Wav2Pose modules. To facilitate the generation of one-shot and more consecutive talking head videos, we refine an existing Image2Video model by integrating a Face Locator and Motion Frame mechanism. We subsequently optimize the model using extensive human face video datasets, significantly enhancing its ability to produce high-quality and expressive talking head videos. Additionally, we develop a demo platform using the Gradio framework, which streamlines the process, enabling users to quickly create customized talking head videos.

Haodong Yu, Yabo Zhang, Donglin Di et al.

While diffusion models excel at generating images with conventional dimensions, pushing them to synthesize ultra-high-resolution imagery at extreme aspect ratios (EAR) often triggers catastrophic structural failures, such as object repetition and spatial fragmentation. This limitation fundamentally stems from a lack of robust spatial priors, as static text-to-image models are primarily trained on image distributions with conventional dimensions. To overcome this bottleneck, we present ScrollScape, a novel framework that reformulates EAR image synthesis into a continuous video generation process through two core innovations. By mapping the spatial expansion of a massive canvas to the temporal evolution of video frames, ScrollScape leverages the inherent temporal consistency of video models as a powerful global constraint to ensure long-range structural integrity. Specifically, Scanning Positional Encoding (ScanPE) distributes global coordinates across frames to act as a flexible moving camera, while Scrolling Super-Resolution (ScrollSR) leverages video super-resolution priors to circumvent memory bottlenecks, efficiently scaling outputs to an unprecedented 32K resolution. Fine-tuned on a curated 3K multi-ratio image dataset, ScrollScape effectively aligns pre-trained video priors with the EAR generation task. Extensive evaluations demonstrate that it significantly outperforms existing image-diffusion baselines by eliminating severe localized artifacts. Consequently, our method overcomes inherent structural bottlenecks to ensure exceptional global coherence and visual fidelity across diverse domains at extreme scales.

Xiaokai Chen, Xuan Liu, Donglin Di et al.

In recent years, facial video generation models have gained popularity. However, these models often lack expressive power when dealing with exaggerated anime-style faces due to the absence of high-quality anime-style face training sets. We propose a facial animation platform that enables real-time conversion from real human faces to cartoon-style faces, supporting multiple models. Built on the Gradio framework, our platform ensures excellent interactivity and user-friendliness. Users can input a real face video or image and select their desired cartoon style. The system will then automatically analyze facial features, execute necessary preprocessing, and invoke appropriate models to generate expressive anime-style faces. We employ a variety of models within our system to process the HDTF dataset, thereby creating an animated facial video dataset.

Kaiwen Yan, Hongcheng Guo, Xuanqing Shi et al.

With the rapid advancement of Large Language Models (LLMs), the demand for robust instruction-following capabilities in code generation tasks has grown significantly. Code generation not only facilitates faster prototyping and automated testing, but also augments developer efficiency through improved maintainability and reusability of code. In this paper, we introduce CodeIF, the first benchmark specifically designed to assess the abilities of LLMs to adhere to task-oriented instructions within diverse code generation scenarios. CodeIF encompasses a broad range of tasks, including function synthesis, error debugging, algorithmic refactoring, and code explanation, thereby providing a comprehensive suite to evaluate model performance across varying complexity levels and programming domains. We conduct extensive experiments with LLMs, analyzing their strengths and limitations in meeting the demands of these tasks. The experimental results offer valuable insights into how well current models align with human instructions, as well as the extent to which they can generate consistent, maintainable, and contextually relevant code. Our findings not only underscore the critical role that instruction-following LLMs can play in modern software development, but also illuminate pathways for future research aimed at enhancing their adaptability, reliability, and overall effectiveness in automated code generation. CodeIF data and code are publicly available: https://github.com/lin-rany/codeIF

Hao Sun, Lei Fan, Donglin Di et al.

Articulated object generation has seen increasing advancements, yet existing models often lack the ability to be conditioned on text prompts. To address the significant gap between textual descriptions and 3D articulated object representations, we propose GAOT, a three-phase framework that generates articulated objects from text prompts, leveraging diffusion models and hypergraph learning in a three-step process. First, we fine-tune a point cloud generation model to produce a coarse representation of objects from text prompts. Given the inherent connection between articulated objects and graph structures, we design a hypergraph-based learning method to refine these coarse representations, representing object parts as graph vertices. Finally, leveraging a diffusion model, the joints of articulated objects-represented as graph edges-are generated based on the object parts. Extensive qualitative and quantitative experiments on the PartNet-Mobility dataset demonstrate the effectiveness of our approach, achieving superior performance over previous methods.

Fuxiang Yang, Tonghua Su, Donglin Di et al.

Scene Text Editing (STE) involves replacing text in a scene image with new target text while preserving both the original text style and background texture. Existing methods suffer from two major challenges: inconsistency and length-insensitivity. They often fail to maintain coherence between the edited local patch and the surrounding area, and they struggle to handle significant differences in text length before and after editing. To tackle these challenges, we propose an end-to-end framework called Global-Local Aware Scene Text Editing (GLASTE), which simultaneously incorporates high-level global contextual information along with delicate local features. Specifically, we design a global-local combination structure, joint global and local losses, and enhance text image features to ensure consistency in text style within local patches while maintaining harmony between local and global areas. Additionally, we express the text style as a vector independent of the image size, which can be transferred to target text images of various sizes. We use an affine fusion to fill target text images into the editing patch while maintaining their aspect ratio unchanged. Extensive experiments on real-world datasets validate that our GLASTE model outperforms previous methods in both quantitative metrics and qualitative results and effectively mitigates the two challenges.

Junze Wang, Lei Fan, Dezheng Zhang et al.

Visual Prompt Tuning (VPT) adapts a frozen Vision Transformer (ViT) to downstream tasks by inserting a small number of learnable prompt tokens into the token sequence at each layer. However, we observe that existing VPT variants often suffer from unstable training dynamics, characterized by gradient oscillations. A layer-wise analysis reveals that shallow-layer prompts tend to stagnate early, while deeper-layer prompts exhibit high-variance oscillations, leading to cross-layer mismatch. These issues slow convergence and degrade final performance. To address these challenges, we propose Prompt-Agnostic Evolution ($\mathtt{PAE}$), which strengthens vision prompt tuning by explicitly modeling prompt dynamics. From a frequency-domain perspective, we initialize prompts in a task-aware direction by uncovering and propagating frequency shortcut patterns that the backbone inherently exploits for recognition. To ensure coherent evolution across layers, we employ a shared Koopman operator that imposes a global linear transformation instead of uncoordinated, layer-specific updates. Finally, inspired by Lyapunov stability theory, we introduce a regularizer that constrains error amplification during evolution. Extensive experiments show that $\mathtt{PAE}$ accelerates convergence with an average $1.41\times$ speedup and improves accuracy by 1--3% on 25 datasets across multiple downstream tasks. Beyond performance, $\mathtt{PAE}$ is prompt-agnostic and lightweight, and it integrates seamlessly with diverse VPT variants without backbone modification or inference-time changes.

Shibo Sun, Xue Li, Donglin Di et al.

While large language models (LLMs) have advanced procedural planning for embodied AI systems through strong reasoning abilities, the integration of multimodal inputs and counterfactual reasoning remains underexplored. To tackle these challenges, we introduce LLaPa, a vision-language model framework designed for multimodal procedural planning. LLaPa generates executable action sequences from textual task descriptions and visual environmental images using vision-language models (VLMs). Furthermore, we enhance LLaPa with two auxiliary modules to improve procedural planning. The first module, the Task-Environment Reranker (TER), leverages task-oriented segmentation to create a task-sensitive feature space, aligning textual descriptions with visual environments and emphasizing critical regions for procedural execution. The second module, the Counterfactual Activities Retriever (CAR), identifies and emphasizes potential counterfactual conditions, enhancing the model's reasoning capability in counterfactual scenarios. Extensive experiments on ActPlan-1K and ALFRED benchmarks demonstrate that LLaPa generates higher-quality plans with superior LCS and correctness, outperforming advanced models. The code and models are available https://github.com/sunshibo1234/LLaPa.

Haotian Ling, Zequn Chen, Qiuying Chen et al.

Consistent pose-driven character animation has achieved remarkable progress in single-character scenarios. However, extending these advances to multi-character settings is non-trivial, especially when position swap is involved. Beyond mere scaling, the core challenge lies in enforcing correct Identity Correspondence (IC) between characters in reference and generated frames. To address this, we introduce EverybodyDance, a systematic solution targeting IC correctness in multi-character animation. EverybodyDance is built around the Identity Matching Graph (IMG), which models characters in the generated and reference frames as two node sets in a weighted complete bipartite graph. Edge weights, computed via our proposed Mask-Query Attention (MQA), quantify the affinity between each pair of characters. Our key insight is to formalize IC correctness as a graph structural metric and to optimize it during training. We also propose a series of targeted strategies tailored for multi-character animation, including identity-embedded guidance, a multi-scale matching strategy, and pre-classified sampling, which work synergistically. Finally, to evaluate IC performance, we curate the Identity Correspondence Evaluation benchmark, dedicated to multi-character IC correctness. Extensive experiments demonstrate that EverybodyDance substantially outperforms state-of-the-art baselines in both IC and visual fidelity.

Guoxin Zang, Xue Li, Donglin Di et al.

While Vision-Language Models (VLMs) have shown promising progress in general multimodal tasks, they often struggle in industrial anomaly detection and reasoning, particularly in delivering interpretable explanations and generalizing to unseen categories. This limitation stems from the inherently domain-specific nature of anomaly detection, which hinders the applicability of existing VLMs in industrial scenarios that require precise, structured, and context-aware analysis. To address these challenges, we propose SAGE, a VLM-based framework that enhances anomaly reasoning through Self-Guided Fact Enhancement (SFE) and Entropy-aware Direct Preference Optimization (E-DPO). SFE integrates domain-specific knowledge into visual reasoning via fact extraction and fusion, while E-DPO aligns model outputs with expert preferences using entropy-aware optimization. Additionally, we introduce AD-PL, a preference-optimized dataset tailored for industrial anomaly reasoning, consisting of 28,415 question-answering instances with expert-ranked responses. To evaluate anomaly reasoning models, we develop Multiscale Logical Evaluation (MLE), a quantitative framework analyzing model logic and consistency. SAGE demonstrates superior performance on industrial anomaly datasets under zero-shot and one-shot settings. The code, model and dataset are available at https://github.com/amoreZgx1n/SAGE.

Mingcheng Qu, Yuncong Wu, Donglin Di et al.

Spatial transcriptomics (ST) provides crucial insights into tissue micro-environments, but is limited to its high cost and complexity. As an alternative, predicting gene expression from pathology whole slide images (WSI) is gaining increasing attention. However, existing methods typically rely on single patches or a single pathology modality, neglecting the complex spatial and molecular interactions between target and neighboring information (e.g., gene co-expression). This leads to a failure in establishing connections among adjacent regions and capturing intricate cross-modal relationships. To address these issues, we propose NH2ST, a framework that integrates spatial context and both pathology and gene modalities for gene expression prediction. Our model comprises a query branch and a neighbor branch to process paired target patch and gene data and their neighboring regions, where cross-attention and contrastive learning are employed to capture intrinsic associations and ensure alignments between pathology and gene expression. Extensive experiments on six datasets demonstrate that our model consistently outperforms existing methods, achieving over 20% in PCC metrics. Codes are available at https://github.com/MCPathology/NH2ST

Mingcheng Qu, Guang Yang, Donglin Di et al.

Multimodal pathology-genomic analysis is critical for cancer survival prediction. However, existing approaches predominantly integrate formalin-fixed paraffin-embedded (FFPE) slides with genomic data, while neglecting the availability of other preservation slides, such as Fresh Froze (FF) slides. Moreover, as the high-resolution spatial nature of pathology data tends to dominate the cross-modality fusion process, it hinders effective multimodal fusion and leads to modality imbalance challenges between pathology and genomics. These methods also typically require complete data modalities, limiting their clinical applicability with incomplete modalities, such as missing either pathology or genomic data. In this paper, we propose a multimodal survival prediction framework that leverages hypergraph learning to effectively integrate multi-WSI information and cross-modality interactions between pathology slides and genomics data while addressing modality imbalance. In addition, we introduce a memory mechanism that stores previously learned paired pathology-genomic features and dynamically compensates for incomplete modalities. Experiments on five TCGA datasets demonstrate that our model outperforms advanced methods by over 2.3% in C-Index. Under incomplete modality scenarios, our approach surpasses pathology-only (3.3%) and gene-only models (7.9%). Code: https://github.com/MCPathology/M2Surv

Junze Wang, Lei Fan, Weipeng Jing et al.

Existing methods for multimodal MRI segmentation with missing modalities typically assume that all MRI modalities are available during training. However, in clinical practice, some modalities may be missing due to the sequential nature of MRI acquisition, leading to performance degradation. Furthermore, retraining models to accommodate newly available modalities can be inefficient and may cause overfitting, potentially compromising previously learned knowledge. To address these challenges, we propose Replay-based Hypergraph Domain Incremental Learning (ReHyDIL) for brain tumor segmentation with missing modalities. ReHyDIL leverages Domain Incremental Learning (DIL) to enable the segmentation model to learn from newly acquired MRI modalities without forgetting previously learned information. To enhance segmentation performance across diverse patient scenarios, we introduce the Cross-Patient Hypergraph Segmentation Network (CHSNet), which utilizes hypergraphs to capture high-order associations between patients. Additionally, we incorporate Tversky-Aware Contrastive (TAC) loss to effectively mitigate information imbalance both across and within different modalities. Extensive experiments on the BraTS2019 dataset demonstrate that ReHyDIL outperforms state-of-the-art methods, achieving an improvement of over 2% in the Dice Similarity Coefficient across various tumor regions. Our code is available at https://github.com/reeive/ReHyDIL.

Donglin Di, Jiahui Yang, Chaofan Luo et al.

Text-to-3D generation represents an exciting field that has seen rapid advancements, facilitating the transformation of textual descriptions into detailed 3D models. However, current progress often neglects the intricate high-order correlation of geometry and texture within 3D objects, leading to challenges such as over-smoothness, over-saturation and the Janus problem. In this work, we propose a method named ``3D Gaussian Generation via Hypergraph (Hyper-3DG)'', designed to capture the sophisticated high-order correlations present within 3D objects. Our framework is anchored by a well-established mainflow and an essential module, named ``Geometry and Texture Hypergraph Refiner (HGRefiner)''. This module not only refines the representation of 3D Gaussians but also accelerates the update process of these 3D Gaussians by conducting the Patch-3DGS Hypergraph Learning on both explicit attributes and latent visual features. Our framework allows for the production of finely generated 3D objects within a cohesive optimization, effectively circumventing degradation. Extensive experimentation has shown that our proposed method significantly enhances the quality of 3D generation while incurring no additional computational overhead for the underlying framework. (Project code: https://github.com/yjhboy/Hyper3DG)

Haoyu Wang, Zhilu Zhang, Donglin Di et al.

The goal of image-based virtual try-on is to generate an image of the target person naturally wearing the given clothing. However, existing methods solely focus on the frontal try-on using the frontal clothing. When the views of the clothing and person are significantly inconsistent, particularly when the person's view is non-frontal, the results are unsatisfactory. To address this challenge, we introduce Multi-View Virtual Try-ON (MV-VTON), which aims to reconstruct the dressing results from multiple views using the given clothes. Given that single-view clothes provide insufficient information for MV-VTON, we instead employ two images, i.e., the frontal and back views of the clothing, to encompass the complete view as much as possible. Moreover, we adopt diffusion models that have demonstrated superior abilities to perform our MV-VTON. In particular, we propose a view-adaptive selection method where hard-selection and soft-selection are applied to the global and local clothing feature extraction, respectively. This ensures that the clothing features are roughly fit to the person's view. Subsequently, we suggest joint attention blocks to align and fuse clothing features with person features. Additionally, we collect a MV-VTON dataset MVG, in which each person has multiple photos with diverse views and poses. Experiments show that the proposed method not only achieves state-of-the-art results on MV-VTON task using our MVG dataset, but also has superiority on frontal-view virtual try-on task using VITON-HD and DressCode datasets.

Lei Fan, Dongdong Fan, Zhiguang Hu et al.

We present MANTA, a visual-text anomaly detection dataset for tiny objects. The visual component comprises over 137.3K images across 38 object categories spanning five typical domains, of which 8.6K images are labeled as anomalous with pixel-level annotations. Each image is captured from five distinct viewpoints to ensure comprehensive object coverage. The text component consists of two subsets: Declarative Knowledge, including 875 words that describe common anomalies across various domains and specific categories, with detailed explanations for < what, why, how>, including causes and visual characteristics; and Constructivist Learning, providing 2K multiple-choice questions with varying levels of difficulty, each paired with images and corresponded answer explanations. We also propose a baseline for visual-text tasks and conduct extensive benchmarking experiments to evaluate advanced methods across different settings, highlighting the challenges and efficacy of our dataset.

Huaize Liu, Wenzhang Sun, Donglin Di et al.

The generation of talking avatars has achieved significant advancements in precise audio synchronization. However, crafting lifelike talking head videos requires capturing a broad spectrum of emotions and subtle facial expressions. Current methods face fundamental challenges: a) the absence of frameworks for modeling single basic emotional expressions, which restricts the generation of complex emotions such as compound emotions; b) the lack of comprehensive datasets rich in human emotional expressions, which limits the potential of models. To address these challenges, we propose the following innovations: 1) the Mixture of Emotion Experts (MoEE) model, which decouples six fundamental emotions to enable the precise synthesis of both singular and compound emotional states; 2) the DH-FaceEmoVid-150 dataset, specifically curated to include six prevalent human emotional expressions as well as four types of compound emotions, thereby expanding the training potential of emotion-driven models. Furthermore, to enhance the flexibility of emotion control, we propose an emotion-to-latents module that leverages multimodal inputs, aligning diverse control signals-such as audio, text, and labels-to ensure more varied control inputs as well as the ability to control emotions using audio alone. Through extensive quantitative and qualitative evaluations, we demonstrate that the MoEE framework, in conjunction with the DH-FaceEmoVid-150 dataset, excels in generating complex emotional expressions and nuanced facial details, setting a new benchmark in the field. These datasets will be publicly released.

Wenzhang Sun, Qirui Hou, Donglin Di et al.

Diffusion Transformers (DiT) excel in video generation but encounter significant computational challenges due to the quadratic complexity of attention. Notably, attention differences between adjacent diffusion steps follow a U-shaped pattern. Current methods leverage this property by caching attention blocks, however, they still struggle with sudden error spikes and large discrepancies. To address these issues, we propose UniCP a unified caching and pruning framework for efficient video generation. UniCP optimizes both temporal and spatial dimensions through. Error Aware Dynamic Cache Window (EDCW): Dynamically adjusts cache window sizes for different blocks at various timesteps, adapting to abrupt error changes. PCA based Slicing (PCAS) and Dynamic Weight Shift (DWS): PCAS prunes redundant attention components, and DWS integrates caching and pruning by enabling dynamic switching between pruned and cached outputs. By adjusting cache windows and pruning redundant components, UniCP enhances computational efficiency and maintains video detail fidelity. Experimental results show that UniCP outperforms existing methods in both performance and efficiency.

Wenzhang Sun, Xiang Li, Donglin Di et al.

Recently, animating portrait images using audio input is a popular task. Creating lifelike talking head videos requires flexible and natural movements, including facial and head dynamics, camera motion, realistic light and shadow effects. Existing methods struggle to offer comprehensive, multifaceted control over these aspects. In this work, we introduce UniAvatar, a designed method that provides extensive control over a wide range of motion and illumination conditions. Specifically, we use the FLAME model to render all motion information onto a single image, maintaining the integrity of 3D motion details while enabling fine-grained, pixel-level control. Beyond motion, this approach also allows for comprehensive global illumination control. We design independent modules to manage both 3D motion and illumination, permitting separate and combined control. Extensive experiments demonstrate that our method outperforms others in both broad-range motion control and lighting control. Additionally, to enhance the diversity of motion and environmental contexts in current datasets, we collect and plan to publicly release two datasets, DH-FaceDrasMvVid-100 and DH-FaceReliVid-200, which capture significant head movements during speech and various lighting scenarios.

Mingcheng Qu, Yuncong Wu, Donglin Di et al.

Spatial transcriptomics (ST) has emerged as an advanced technology that provides spatial context to gene expression. Recently, deep learning-based methods have shown the capability to predict gene expression from WSI data using ST data. Existing approaches typically extract features from images and the neighboring regions using pretrained models, and then develop methods to fuse this information to generate the final output. However, these methods often fail to account for the cellular structure similarity, cellular density and the interactions within the microenvironment. In this paper, we propose a framework named BG-TRIPLEX, which leverages boundary information extracted from pathological images as guiding features to enhance gene expression prediction from WSIs. Specifically, our model consists of three branches: the spot, in-context and global branches. In the spot and in-context branches, boundary information, including edge and nuclei characteristics, is extracted using pretrained models. These boundary features guide the learning of cellular morphology and the characteristics of microenvironment through Multi-Head Cross-Attention. Finally, these features are integrated with global features to predict the final output. Extensive experiments were conducted on three public ST datasets. The results demonstrate that our BG-TRIPLEX consistently outperforms existing methods in terms of Pearson Correlation Coefficient (PCC). This method highlights the crucial role of boundary features in understanding the complex interactions between WSI and gene expression, offering a promising direction for future research.

Yongjia Ma, Donglin Di, Xuan Liu et al.

Rectified flow models have achieved remarkable performance in image and video generation tasks. However, existing numerical solvers face a trade-off between fast sampling and high-accuracy solutions, limiting their effectiveness in downstream applications such as reconstruction and editing. To address this challenge, we propose leveraging the Adams-Bashforth-Moulton (ABM) predictor-corrector method to enhance the accuracy of ODE solving in rectified flow models. Specifically, we introduce ABM-Solver, which integrates a multi step predictor corrector approach to reduce local truncation errors and employs Adaptive Step Size Adjustment to improve sampling speed. Furthermore, to effectively preserve non edited regions while facilitating semantic modifications, we introduce a Mask Guided Feature Injection module. We estimate self-similarity to generate a spatial mask that differentiates preserved regions from those available for editing. Extensive experiments on multiple high-resolution image datasets validate that ABM-Solver significantly improves inversion precision and editing quality, outperforming existing solvers without requiring additional training or optimization.

Jiahui Yang, Donglin Di, Baorui Ma et al.

In recent years, advancements in generative models have significantly expanded the capabilities of text-to-3D generation. Many approaches rely on Score Distillation Sampling (SDS) technology. However, SDS struggles to accommodate multi-condition inputs, such as text and visual prompts, in customized generation tasks. To explore the core reasons, we decompose SDS into a difference term and a classifier-free guidance term. Our analysis identifies the core issue as arising from the difference term and the random noise addition during the optimization process, both contributing to deviations from the target mode during distillation. To address this, we propose a novel algorithm, Classifier Score Matching (CSM), which removes the difference term in SDS and uses a deterministic noise addition process to reduce noise during optimization, effectively overcoming the low-quality limitations of SDS in our customized generation framework. Based on CSM, we integrate visual prompt information with an attention fusion mechanism and sampling guidance techniques, forming the Visual Prompt CSM (VPCSM) algorithm. Furthermore, we introduce a Semantic-Geometry Calibration (SGC) module to enhance quality through improved textual information integration. We present our approach as TV-3DG, with extensive experiments demonstrating its capability to achieve stable, high-quality, customized 3D generation. Project page: \url{https://yjhboy.github.io/TV-3DG}

Juntong Fang, Zequn Chen, Weiqi Zhang et al.

Reconstructing dynamic 4D scenes remains challenging due to the presence of moving objects that corrupt camera pose estimation. Existing optimization methods alleviate this issue with additional supervision, but they are mostly computationally expensive and impractical in real-time applications. To address these limitations, we propose MoRe, a feedforward 4D reconstruction network that efficiently recovers dynamic 3D scenes from monocular videos. Built upon a strong static reconstruction backbone, MoRe employs an attention-forcing strategy to disentangle dynamic motion from static structure. To further enhance robustness, we fine-tune the model on large-scale, diverse datasets encompassing both dynamic and static scenes. Moreover, our grouped causal attention captures temporal dependencies and adapts to varying token lengths across frames, ensuring temporally coherent geometry reconstruction. Extensive experiments on multiple benchmarks demonstrate that MoRe achieves high-quality dynamic reconstructions with exceptional efficiency.

Yongjia Ma, Junlin Chen, Donglin Di et al.

Creating high-fidelity, coherent long videos is a sought-after aspiration. While recent video diffusion models have shown promising potential, they still grapple with spatiotemporal inconsistencies and high computational resource demands. We propose GLC-Diffusion, a tuning-free method for long video generation. It models the long video denoising process by establishing denoising trajectories through Global-Local Collaborative Denoising to ensure overall content consistency and temporal coherence between frames. Additionally, we introduce a Noise Reinitialization strategy which combines local noise shuffling with frequency fusion to improve global content consistency and visual diversity. Further, we propose a Video Motion Consistency Refinement (VMCR) module that computes the gradient of pixel-wise and frequency-wise losses to enhance visual consistency and temporal smoothness. Extensive experiments, including quantitative and qualitative evaluations on videos of varying lengths (\textit{e.g.}, 3\times and 6\times longer), demonstrate that our method effectively integrates with existing video diffusion models, producing coherent, high-fidelity long videos superior to previous approaches.

Mingcheng Qu, Guang Yang, Donglin Di et al.

Multimodal pathology-genomic analysis has become increasingly prominent in cancer survival prediction. However, existing studies mainly utilize multi-instance learning to aggregate patch-level features, neglecting the information loss of contextual and hierarchical details within pathology images. Furthermore, the disparity in data granularity and dimensionality between pathology and genomics leads to a significant modality imbalance. The high spatial resolution inherent in pathology data renders it a dominant role while overshadowing genomics in multimodal integration. In this paper, we propose a multimodal survival prediction framework that incorporates hypergraph learning to effectively capture both contextual and hierarchical details from pathology images. Moreover, it employs a modality rebalance mechanism and an interactive alignment fusion strategy to dynamically reweight the contributions of the two modalities, thereby mitigating the pathology-genomics imbalance. Quantitative and qualitative experiments are conducted on five TCGA datasets, demonstrating that our model outperforms advanced methods by over 3.4\% in C-Index performance.

Lei Fan, Junjie Huang, Donglin Di et al.

For anomaly detection (AD), early approaches often train separate models for individual classes, yielding high performance but posing challenges in scalability and resource management. Recent efforts have shifted toward training a single model capable of handling multiple classes. However, directly extending early AD methods to multi-class settings often results in degraded performance. In this paper, we investigate this performance degradation observed in reconstruction-based methods, identifying the key issue: inter-class confusion. This confusion emerges when a model trained in multi-class scenarios incorrectly reconstructs samples from one class as those of another, thereby exacerbating reconstruction errors. To this end, we propose a simple yet effective modification, called class-aware contrastive learning (CCL). By explicitly leveraging raw object category information (\eg carpet or wood) as supervised signals, we introduce local CL to refine multiscale dense features, and global CL to obtain more compact feature representations of normal patterns, thereby effectively adapting the models to multi-class settings. Experiments across five datasets validate the effectiveness of our approach, demonstrating significant improvements and superior performance compared to state-of-the-art methods. Notably, ablation studies indicate that pseudo-class labels can achieve comparable performance.

Hongcheng Guo, Juntao Yao, Boyang Wang et al.

Mixture-of-Experts (MoE) architectures have emerged as a promising paradigm for scaling large language models (LLMs) with sparse activation of task-specific experts. Despite their computational efficiency during inference, the massive overall parameter footprint of MoE models (e.g., GPT-4) introduces critical challenges for practical deployment. Current pruning approaches often fail to address two inherent characteristics of MoE systems: 1).intra-layer expert homogeneity where experts within the same MoE layer exhibit functional redundancy, and 2). inter-layer similarity patterns where deeper layers tend to contain progressively more homogeneous experts. To tackle these issues, we propose Cluster-driven Expert Pruning (C-Prune), a novel two-stage framework for adaptive task-specific compression of MoE LLMs. C-Prune operates through layer-wise expert clustering, which groups functionally similar experts within each MoE layer using parameter similarity metrics, followed by global cluster pruning, which eliminates redundant clusters across all layers through a unified importance scoring mechanism that accounts for cross-layer homogeneity. We validate C-Prune through extensive experiments on multiple MoE models and benchmarks. The results demonstrate that C-Prune effectively reduces model size while outperforming existing MoE pruning methods.

Qi Xie, Yongjia Ma, Donglin Di et al.

Achieving ID-preserving text-to-video (T2V) generation remains challenging despite recent advances in diffusion-based models. Existing approaches often fail to capture fine-grained facial dynamics or maintain temporal identity coherence. To address these limitations, we propose MoCA, a novel Video Diffusion Model built on a Diffusion Transformer (DiT) backbone, incorporating a Mixture of Cross-Attention mechanism inspired by the Mixture-of-Experts paradigm. Our framework improves inter-frame identity consistency by embedding MoCA layers into each DiT block, where Hierarchical Temporal Pooling captures identity features over varying timescales, and Temporal-Aware Cross-Attention Experts dynamically model spatiotemporal relationships. We further incorporate a Latent Video Perceptual Loss to enhance identity coherence and fine-grained details across video frames. To train this model, we collect CelebIPVid, a dataset of 10,000 high-resolution videos from 1,000 diverse individuals, promoting cross-ethnicity generalization. Extensive experiments on CelebIPVid show that MoCA outperforms existing T2V methods by over 5% across Face similarity.

Jiahui Yang, Yongjia Ma, Donglin Di et al.

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/.

Qiyuan Zhang, Chenyu Wu, Wenzhang Sun et al.

Recent advancements in portrait video generation have been noteworthy. However, existing methods rely heavily on human priors and pre-trained generative models, Motion representations based on human priors may introduce unrealistic motion, while methods relying on pre-trained generative models often suffer from inefficient inference. To address these challenges, we propose Semantic Latent Motion (SeMo), a compact and expressive motion representation. Leveraging this representation, our approach achieve both high-quality visual results and efficient inference. SeMo follows an effective three-step framework: Abstraction, Reasoning, and Generation. First, in the Abstraction step, we use a carefully designed Masked Motion Encoder, which leverages a self-supervised learning paradigm to compress the subject's motion state into a compact and abstract latent motion (1D token). Second, in the Reasoning step, we efficiently generate motion sequences based on the driving audio signal. Finally, in the Generation step, the motion dynamics serve as conditional information to guide the motion decoder in synthesizing realistic transitions from reference frame to target video. Thanks to the compact and expressive nature of Semantic Latent Motion, our method achieves efficient motion representation and high-quality video generation. User studies demonstrate that our approach surpasses state-of-the-art models with an 81% win rate in realism. Extensive experiments further highlight its strong compression capability, reconstruction quality, and generative potential.

Donglin Di, Weinan Zhang, Yue Zhang et al.

Making use of off-the-shelf resources of resource-rich languages to transfer knowledge for low-resource languages raises much attention recently. The requirements of enabling the model to reach the reliable performance lack well guided, such as the scale of required annotated data or the effective framework. To investigate the first question, we empirically investigate the cost-effectiveness of several methods to train the intent classification and slot-filling models for Indonesia (ID) from scratch by utilizing the English data. Confronting the second challenge, we propose a Bi-Confidence-Frequency Cross-Lingual transfer framework (BiCF), composed by ``BiCF Mixing'', ``Latent Space Refinement'' and ``Joint Decoder'', respectively, to tackle the obstacle of lacking low-resource language dialogue data. Extensive experiments demonstrate our framework performs reliably and cost-efficiently on different scales of manually annotated Indonesian data. We release a large-scale fine-labeled dialogue dataset (ID-WOZ) and ID-BERT of Indonesian for further research.

Zhou Liu, Tonghua Su, Hongshi Zhang et al.

Advances in image tampering techniques, particularly generative models, pose significant challenges to media verification, digital forensics, and public trust. Existing image forgery detection and localization (IFDL) methods suffer from two key limitations: over-reliance on semantic content while neglecting textural cues, and limited interpretability of subtle low-level tampering traces. To address these issues, we propose FOCA, a multimodal large language model-based framework that integrates discriminative features from both the RGB spatial and frequency domains via a cross-attention fusion module. This design enables accurate forgery detection and localization while providing explicit, human-interpretable cross-domain explanations. We further introduce FSE-Set, a large-scale dataset with diverse authentic and tampered images, pixel-level masks, and dual-domain annotations. Extensive experiments show that FOCA outperforms state-of-the-art methods in detection performance and interpretability across both spatial and frequency domains.

Hai Ling, Jia Guo, Zhulin Tao et al.

Anomaly detection (AD) aims to identify defects using normal-only training data. Existing anomaly detection benchmarks (e.g., MVTec-AD with 15 categories) cover only a narrow range of categories, limiting the evaluation of cross-context generalization and scalability. We introduce ADNet, a large-scale, multi-domain benchmark comprising 380 categories aggregated from 49 publicly available datasets across Electronics, Industry, Agrifood, Infrastructure, and Medical domains. The benchmark includes a total of 196,294 RGB images, consisting of 116,192 normal samples for training and 80,102 test images, of which 60,311 are anomalous. All images are standardized with MVTec-style pixel-level annotations and structured text descriptions spanning both spatial and visual attributes, enabling multimodal anomaly detection tasks. Extensive experiments reveal a clear scalability challenge: existing state-of-the-art methods achieve 90.6% I-AUROC in one-for-one settings but drop to 78.5% when scaling to all 380 categories in a multi-class setting. To address this, we propose Dinomaly-m, a context-guided Mixture-of-Experts extension of Dinomaly that expands decoder capacity without increasing inference cost. It achieves 83.2% I-AUROC and 93.1% P-AUROC, demonstrating superior performance over existing approaches. ADNet is designed as a standardized and extensible benchmark, supporting the community in expanding anomaly detection datasets across diverse domains and providing a scalable foundation for future anomaly detection foundation models. Dataset: https://grainnet.github.io/ADNet

Jia Guo, Shuai Lu, Lei Fan et al.

Unsupervised anomaly detection (UAD) has evolved from building specialized single-class models to unified multi-class models, yet existing multi-class models significantly underperform the most advanced one-for-one counterparts. Moreover, the field has fragmented into specialized methods tailored to specific scenarios (multi-class, 3D, few-shot, etc.), creating deployment barriers and highlighting the need for a unified solution. In this paper, we present Dinomaly2, the first unified framework for full-spectrum image UAD, which bridges the performance gap in multi-class models while seamlessly extending across diverse data modalities and task settings. Guided by the "less is more" philosophy, we demonstrate that the orchestration of five simple element achieves superior performance in a standard reconstruction-based framework. This methodological minimalism enables natural extension across diverse tasks without modification, establishing that simplicity is the foundation of true universality. Extensive experiments on 12 UAD benchmarks demonstrate Dinomaly2's full-spectrum superiority across multiple modalities (2D, multi-view, RGB-3D, RGB-IR), task settings (single-class, multi-class, inference-unified multi-class, few-shot) and application domains (industrial, biological, outdoor). For example, our multi-class model achieves unprecedented 99.9% and 99.3% image-level (I-) AUROC on MVTec-AD and VisA respectively. For multi-view and multi-modal inspection, Dinomaly2 demonstrates state-of-the-art performance with minimum adaptations. Moreover, using only 8 normal examples per class, our method surpasses previous full-shot models, achieving 98.7% and 97.4% I-AUROC on MVTec-AD and VisA. The combination of minimalistic design, computational scalability, and universal applicability positions Dinomaly2 as a unified solution for the full spectrum of real-world anomaly detection applications.

Haoyu Wang, Hao Tang, Donglin Di et al.

Existing video generation models predominantly emphasize appearance fidelity while exhibiting limited ability to synthesize complex human motions, such as whole-body movements, long-range dynamics, and fine-grained human-environment interactions. This often leads to unrealistic or physically implausible movements with inadequate structural coherence. To conquer these challenges, we propose MoSA, which decouples the process of human video generation into two components, i.e., structure generation and appearance generation. MoSA first employs a 3D structure transformer to generate a human motion sequence from the text prompt. The remaining video appearance is then synthesized under the guidance of this structural sequence. We achieve fine-grained control over the sparse human structures by introducing Human-Aware Dynamic Control modules with a dense tracking constraint during training. The modeling of human-environment interactions is improved through the proposed contact constraint. Those two components work comprehensively to ensure the structural and appearance fidelity across the generated videos. This paper also contributes a large-scale human video dataset, which features more complex and diverse motions than existing human video datasets. We conduct comprehensive comparisons between MoSA and a variety of approaches, including general video generation models, human video generation models, and human animation models. Experiments demonstrate that MoSA substantially outperforms existing approaches across the majority of evaluation metrics.

Chunji Lv, Zequn Chen, Donglin Di et al.

While physics-grounded 3D motion synthesis has seen significant progress, current methods face critical limitations. They typically rely on pre-reconstructed 3D Gaussian Splatting (3DGS) representations, while physics integration depends on either inflexible, manually defined physical attributes or unstable, optimization-heavy guidance from video models. To overcome these challenges, we introduce PhysGM, a feed-forward framework that jointly predicts a 3D Gaussian representation and its physical properties from a single image, enabling immediate, physical simulation and high-fidelity 4D rendering. We first establish a base model by jointly optimizing for Gaussian reconstruction and probabilistic physics prediction. The model is then refined with physically plausible reference videos to enhance both rendering fidelity and physics prediction accuracy. We adopt the Direct Preference Optimization (DPO) to align its simulations with reference videos, circumventing Score Distillation Sampling (SDS) optimization which needs back-propagating gradients through the complex differentiable simulation and rasterization. To facilitate the training, we introduce a new dataset PhysAssets of over 24,000 3D assets, annotated with physical properties and corresponding guiding videos. Experimental results demonstrate that our method effectively generates high-fidelity 4D simulations from a single image in one minute. This represents a significant speedup over prior works while delivering realistic rendering results. Our project page is at:https://hihixiaolv.github.io/PhysGM.github.io/

He Feng, Yongjia Ma, Donglin Di et al.

Portrait animation aims to synthesize talking videos from a static reference face, conditioned on audio and style frame cues (e.g., emotion and head poses), while ensuring precise lip synchronization and faithful reproduction of speaking styles. Existing diffusion-based portrait animation methods primarily focus on lip synchronization or static emotion transformation, often overlooking dynamic styles such as head movements. Moreover, most of these methods rely on a dual U-Net architecture, which preserves identity consistency but incurs additional computational overhead. To this end, we propose DiTalker, a unified DiT-based framework for speaking style-controllable portrait animation. We design a Style-Emotion Encoding Module that employs two separate branches: a style branch extracting identity-specific style information (e.g., head poses and movements), and an emotion branch extracting identity-agnostic emotion features. We further introduce an Audio-Style Fusion Module that decouples audio and speaking styles via two parallel cross-attention layers, using these features to guide the animation process. To enhance the quality of results, we adopt and modify two optimization constraints: one to improve lip synchronization and the other to preserve fine-grained identity and background details. Extensive experiments demonstrate the superiority of DiTalker in terms of lip synchronization and speaking style controllability. Project Page: https://thenameishope.github.io/DiTalker/