Xiaobin Hu

Xiaobin Hu, Shuo Wang, Xuebin Qin et al.

Spotting camouflaged objects that are visually assimilated into the background is tricky for both object detection algorithms and humans who are usually confused or cheated by the perfectly intrinsic similarities between the foreground objects and the background surroundings. To tackle this challenge, we aim to extract the high-resolution texture details to avoid the detail degradation that causes blurred vision in edges and boundaries. We introduce a novel HitNet to refine the low-resolution representations by high-resolution features in an iterative feedback manner, essentially a global loop-based connection among the multi-scale resolutions. In addition, an iterative feedback loss is proposed to impose more constraints on each feedback connection. Extensive experiments on four challenging datasets demonstrate that our \ourmodel~breaks the performance bottleneck and achieves significant improvements compared with 29 state-of-the-art methods. To address the data scarcity in camouflaged scenarios, we provide an application example by employing cross-domain learning to extract the features that can reflect the camouflaged object properties and embed the features into salient objects, thereby generating more camouflaged training samples from the diverse salient object datasets The code will be available at https://github.com/HUuxiaobin/HitNet.

Chengwen Liu, Xiaomin Yu, Zhuoyue Chang et al.

In real-world video question answering scenarios, videos often provide only localized visual cues, while verifiable answers are distributed across the open web; models therefore need to jointly perform cross-frame clue extraction, iterative retrieval, and multi-hop reasoning-based verification. To bridge this gap, we construct the first video deep research benchmark, VideoDR. VideoDR centers on video-conditioned open-domain video question answering, requiring cross-frame visual anchor extraction, interactive web retrieval, and multi-hop reasoning over joint video-web evidence; through rigorous human annotation and quality control, we obtain high-quality video deep research samples spanning six semantic domains. We evaluate multiple closed-source and open-source multimodal large language models under both the Workflow and Agentic paradigms, and the results show that Agentic is not consistently superior to Workflow: its gains depend on a model's ability to maintain the initial video anchors over long retrieval chains. Further analysis indicates that goal drift and long-horizon consistency are the core bottlenecks. In sum, VideoDR provides a systematic benchmark for studying video agents in open-web settings and reveals the key challenges for next-generation video deep research agents.

Yujie Liang, Xiaobin Hu, Boyuan Jiang et al. · tencent-ai

Although diffusion-based image virtual try-on has made considerable progress, emerging approaches still struggle to effectively address the issue of hand occlusion (i.e., clothing regions occluded by the hand part), leading to a notable degradation of the try-on performance. To tackle this issue widely existing in real-world scenarios, we propose VTON-HandFit, leveraging the power of hand priors to reconstruct the appearance and structure for hand occlusion cases. Firstly, we tailor a Handpose Aggregation Net using the ControlNet-based structure explicitly and adaptively encoding the global hand and pose priors. Besides, to fully exploit the hand-related structure and appearance information, we propose Hand-feature Disentanglement Embedding module to disentangle the hand priors into the hand structure-parametric and visual-appearance features, and customize a masked cross attention for further decoupled feature embedding. Lastly, we customize a hand-canny constraint loss to better learn the structure edge knowledge from the hand template of model image. VTON-HandFit outperforms the baselines in qualitative and quantitative evaluations on the public dataset and our self-collected hand-occlusion Handfit-3K dataset particularly for the arbitrary hand pose occlusion cases in real-world scenarios. The Code and dataset will be available at \url{https://github.com/VTON-HandFit/VTON-HandFit}.

Xuebin Qin, Hang Dai, Xiaobin Hu et al.

We present a systematic study on a new task called dichotomous image segmentation (DIS) , which aims to segment highly accurate objects from natural images. To this end, we collected the first large-scale DIS dataset, called DIS5K, which contains 5,470 high-resolution (e.g., 2K, 4K or larger) images covering camouflaged, salient, or meticulous objects in various backgrounds. DIS is annotated with extremely fine-grained labels. Besides, we introduce a simple intermediate supervision baseline (IS-Net) using both feature-level and mask-level guidance for DIS model training. IS-Net outperforms various cutting-edge baselines on the proposed DIS5K, making it a general self-learned supervision network that can facilitate future research in DIS. Further, we design a new metric called human correction efforts (HCE) which approximates the number of mouse clicking operations required to correct the false positives and false negatives. HCE is utilized to measure the gap between models and real-world applications and thus can complement existing metrics. Finally, we conduct the largest-scale benchmark, evaluating 16 representative segmentation models, providing a more insightful discussion regarding object complexities, and showing several potential applications (e.g., background removal, art design, 3D reconstruction). Hoping these efforts can open up promising directions for both academic and industries. Project page: https://xuebinqin.github.io/dis/index.html.

Zhifei Xie, Zihang Liu, Ze An et al.

Audio is an inherently interactive modality, yet today's Large Audio Language Models (LALMs) are offline, and streaming audio models each handle only a single task such as streaming ASR or voice chatting. It is time to unify them into one online LALM: a model that, through an always-on perceive-decide-respond loop, listens to sound, environment, and instructions in real time and reacts on the fly. We formalize this regime as the Audio Interaction Model, and realize it with Audio-Interaction, a unified streaming model that retains offline task execution while adding online general audio instruction following, from dialogue to full voice chatting, deciding when to respond from the semantics of the stream. To enable this, we propose SoundFlow, a framework that instantiates the perceive-decide-respond loop end to end, from data to training to deployment, through streaming-native data construction, comprehension-aware training, and asynchronous low-latency inference for stable real-time interaction. We further construct StreamAudio-2M, a 2.6M-item streaming corpus spanning 7 fundamental abilities and 28 sub-tasks, and Proactive-Sound-Bench for evaluating proactive audio intervention. Across 8 benchmarks, Audio-Interaction preserves competitive performance on mainstream audio tasks while unlocking capabilities inaccessible to offline LALMs, including real-time ASR, streaming audio instruction following, and proactive help.

Yinan Chen, Chuming Lin, Zhennan Chen et al.

While instruction-based video editing has seen significant progress, joint audio-visual editing remains constrained by the absence of dedicated datasets and benchmarks. To bridge this gap, we present JAVEdit-100k, the first large-scale, high-quality dataset tailored for instruction-guided joint audio-visual editing. Focusing on human-centric videos, JAVEdit-100k comprises approximately 100K editing triplets spanning five distinct categories, including subject editing and speech editing. This dataset is rigorously constructed via four meticulously designed generation pipelines, seamlessly paired with an agent-in-the-loop quality control mechanism. Furthermore, to address the lack of standardized evaluation within the field, we introduce JAVEditBench, a comprehensive benchmark featuring curated source videos and human-aligned instructions across all editing categories. Finally, we propose JAVEdit, a pioneering baseline model for instruction-guided joint audio-visual editing. Experiments show that \model\ outperforms all baselines on five of six evaluation metrics.

Xiaobin Lu, Xiaobin Hu, Jun Luo et al.

Blind face restoration endeavors to restore a clear face image from a degraded counterpart. Recent approaches employing Generative Adversarial Networks (GANs) as priors have demonstrated remarkable success in this field. However, these methods encounter challenges in achieving a balance between realism and fidelity, particularly in complex degradation scenarios. To inherit the exceptional realism generative ability of the diffusion model and also constrained by the identity-aware fidelity, we propose a novel diffusion-based framework by embedding the 3D facial priors as structure and identity constraints into a denoising diffusion process. Specifically, in order to obtain more accurate 3D prior representations, the 3D facial image is reconstructed by a 3D Morphable Model (3DMM) using an initial restored face image that has been processed by a pretrained restoration network. A customized multi-level feature extraction method is employed to exploit both structural and identity information of 3D facial images, which are then mapped into the noise estimation process. In order to enhance the fusion of identity information into the noise estimation, we propose a Time-Aware Fusion Block (TAFB). This module offers a more efficient and adaptive fusion of weights for denoising, considering the dynamic nature of the denoising process in the diffusion model, which involves initial structure refinement followed by texture detail enhancement. Extensive experiments demonstrate that our network performs favorably against state-of-the-art algorithms on synthetic and real-world datasets for blind face restoration. The Code is released on our project page at https://github.com/838143396/3Diffusion.

Tong Zhang, Jiangning Zhang, Zhucun Xue et al.

Balancing convergence speed, generalization capability, and computational efficiency remains a core challenge in deep learning optimization. First-order gradient descent methods, epitomized by stochastic gradient descent (SGD) and Adam, serve as the cornerstone of modern training pipelines. However, large-scale model training, stringent differential privacy requirements, and distributed learning paradigms expose critical limitations in these conventional approaches regarding privacy protection and memory efficiency. To mitigate these bottlenecks, researchers explore second-order optimization techniques to surpass first-order performance ceilings, while zeroth-order methods reemerge to alleviate memory constraints inherent to large-scale training. Despite this proliferation of methodologies, the field lacks a cohesive framework that unifies underlying principles and delineates application scenarios for these disparate approaches. In this work, we retrospectively analyze the evolutionary trajectory of deep learning optimization algorithms and present a comprehensive empirical evaluation of mainstream optimizers across diverse model architectures and training scenarios. We distill key emerging trends and fundamental design trade-offs, pinpointing promising directions for future research. By synthesizing theoretical insights with extensive empirical evidence, we provide actionable guidance for designing next-generation highly efficient, robust, and trustworthy optimization methods. The code is available at https://github.com/APRIL-AIGC/Awesome-Optimizer.

Jiangning Zhang, Junwei Zhu, Zhenye Gan et al.

We propose a multimodal-driven framework for high-fidelity long-term digital human animation termed $\textbf{Soul}$, which generates semantically coherent videos from a single-frame portrait image, text prompts, and audio, achieving precise lip synchronization, vivid facial expressions, and robust identity preservation. We construct Soul-1M, containing 1 million finely annotated samples with a precise automated annotation pipeline (covering portrait, upper-body, full-body, and multi-person scenes) to mitigate data scarcity, and we carefully curate Soul-Bench for comprehensive and fair evaluation of audio-/text-guided animation methods. The model is built on the Wan2.2-5B backbone, integrating audio-injection layers and multiple training strategies together with threshold-aware codebook replacement to ensure long-term generation consistency. Meanwhile, step/CFG distillation and a lightweight VAE are used to optimize inference efficiency, achieving an 11.4$\times$ speedup with negligible quality loss. Extensive experiments show that Soul significantly outperforms current leading open-source and commercial models on video quality, video-text alignment, identity preservation, and lip-synchronization accuracy, demonstrating broad applicability in real-world scenarios such as virtual anchors and film production. Project page at https://zhangzjn.github.io/projects/Soul/

Haozhen Gong, Xiaozhong Ji, Yuansen Liu et al.

MLLMs MLLMs are beginning to appear in clinical workflows, but their ability to perform complex medical reasoning remains unclear. We present Med-CMR, a fine-grained Medical Complex Multimodal Reasoning benchmark. Med-CMR distinguishes from existing counterparts by three core features: 1) Systematic capability decomposition, splitting medical multimodal reasoning into fine-grained visual understanding and multi-step reasoning to enable targeted evaluation; 2) Challenging task design, with visual understanding across three key dimensions (small-object detection, fine-detail discrimination, spatial understanding) and reasoning covering four clinically relevant scenarios (temporal prediction, causal reasoning, long-tail generalization, multi-source integration); 3) Broad, high-quality data coverage, comprising 20,653 Visual Question Answering (VQA) pairs spanning 11 organ systems and 12 imaging modalities, validated via a rigorous two-stage (human expert + model-assisted) review to ensure clinical authenticity. We evaluate 18 state-of-the-art MLLMs with Med-CMR, revealing GPT-5 as the top-performing commercial model: 57.81 accuracy on multiple-choice questions (MCQs) and a 48.70 open-ended score, outperforming Gemini 2.5 Pro (49.87 MCQ accuracy, 45.98 open-ended score) and leading open-source model Qwen3-VL-235B-A22B (49.34 MCQ accuracy, 42.62 open-ended score). However, specialized medical MLLMs do not reliably outperform strong general models, and long-tail generalization emerges as the dominant failure mode. Med-CMR thus provides a stress test for visual-reasoning integration and rare-case robustness in medical MLLMs, and a rigorous yardstick for future clinical systems.

Haoxuan Xu. Yi Liu, Boyuan Jiang, Jinlong Peng et al. · tencent-ai

Recent advances in post-training paradigms have achieved remarkable success in high-level generation tasks, yet their potential for low-level vision remains rarely explored. Existing image restoration (IR) methods rely on pixel-level hard-fitting to ground-truth images, struggling with over-smoothing and poor generalization. To address these limitations, we propose IRPO, a low-level GRPO-based post-training paradigm that systematically explores both data formulation and reward modeling. We first explore a data formulation principle for low-level post-training paradigm, in which selecting underperforming samples from the pre-training stage yields optimal performance and improved efficiency. Furthermore, we model a reward-level criteria system that balances objective accuracy and human perceptual preference through three complementary components: a General Reward for structural fidelity, an Expert Reward leveraging Qwen-VL for perceptual alignment, and a Restoration Reward for task-specific low-level quality. Comprehensive experiments on six in-domain and five out-of-domain (OOD) low-level benchmarks demonstrate that IRPO achieves state-of-the-art results across diverse degradation types, surpassing the AdaIR baseline by 0.83 dB on in-domain tasks and 3.43 dB on OOD settings. Our code can be shown in https://github.com/HaoxuanXU1024/IRPO.

Xinlei Yu, Chengming Xu, Guibin Zhang et al.

Despite the remarkable success of Vision-Language Models (VLMs), their performance on a range of complex visual tasks is often hindered by a "visual processing bottleneck": a propensity to lose grounding in visual evidence and exhibit a deficit in contextualized visual experience during prolonged generation. Drawing inspiration from human cognitive memory theory, which distinguishes short-term visually-dominant memory and long-term semantically-dominant memory, we propose VisMem, a cognitively-aligned framework that equips VLMs with dynamic latent vision memories, a short-term module for fine-grained perceptual retention and a long-term module for abstract semantic consolidation. These memories are seamlessly invoked during inference, allowing VLMs to maintain both perceptual fidelity and semantic consistency across thinking and generation. Extensive experiments across diverse visual benchmarks for understanding, reasoning, and generation reveal that VisMem delivers a significant average performance boost of 11.8% relative to the vanilla model and outperforms all counterparts, establishing a new paradigm for latent-space memory enhancement. The code will be available: https://github.com/YU-deep/VisMem.git.

Haojun Chen, Haoyang He, Chengming Xu et al.

Text-to-Image (T2I) models have recently seen notable progress around 1K and 2K resolution. With the extreme desire for better visual experience and the rapid development of imaging technology, the demand for Ultra-High-Resolution (UHR) image generation has grown significantly. However, UHR image generation poses great challenges due to the scarcity and complexity of high-resolution content. In this paper, we first introduce PixVerve-95K, a high-quality, open-source UHR T2I dataset curated with a carefully designed data pipeline, which contains 95K images across diverse scenarios (each image has a minimum pixel-count of 100M) and seven-dimensional annotations. Based on our large-scale image-text dataset, we take a pioneering step to extend various T2I foundation models to native 100MP generation with three training schemes. Finally, leveraging both conventional metrics and multimodal large language model-based assessments, our proposed PixVerve-Bench benchmark establishes a comprehensive evaluation protocol for UHR images encompassing visual quality and semantic alignment. Extensive experimental results on our benchmark and the constructive exploration of training strategies collaboratively provide valuable insights for future breakthroughs.

Zheng Nie, Ruolin Shen, Xinlei Yu et al.

Scaling vision-language models into Visual Multiagent Systems (VMAS) is hindered by two coupled issues. First, communication topologies are fixed before inference, leaving them blind to visual content and query context; second, agent reasoning abilities remain static during deployment. These issues reinforce each other: a rigid topology fails to leverage richer agent expertise, while static agents lack incentives to specialize for a given query. We address this with SkillGraph, a joint framework that evolves both agent expertise and communication topology. Within this framework, a Multimodal Graph Transformer (MMGT) encodes visual tokens, instruction semantics and active skill embeddings to predict a query-conditioned collaboration graph, replacing hand-crafted routing with dynamic, content-aware information flow. Complementing this, a Skill Designer distills and refines reasoning heuristics from failure cases, constructing a self-evolving multimodal Skill Bank. Crucially, updated skill embeddings are fed back into the MMGT, enabling the topology to adapt alongside capability growth. Experiments show that SkillGraph achieves consistent improvements across four benchmarks, five common MAS structures and four base models. Code is available at https://github.com/niez233/skillgraph.

Yicheng Xu, Jiangning Zhang, Zhucun Xue et al.

In-context Learning enables training-free adaptation via demonstrations but remains highly sensitive to example selection and formatting. In unified multimodal models spanning understanding and generation, this sensitivity is exacerbated by cross-modal interference and varying cognitive demands. Consequently, In-context Learning efficacy is often non-monotonic and highly task-dependent. To diagnose these behaviors, we introduce a six-level capability-oriented taxonomy that categorizes the functional role of demonstrations from basic perception to high-order discernment. Guided by this cognitive framework, we construct UniICL-760K, a large-scale corpus featuring curated 8-shot In-context Learning episodes across 15 subtasks, alongside UniICL-Bench for rigorous, controlled evaluation. As an architectural intervention to stabilize few-shot adaptation, we propose the Context-Adaptive Prototype Modulator, a lightweight, plug-and-play module. Evaluations on UniICL-Bench show that our approach yields highly competitive unified results, outperforming larger-parameter multimodal large language model baselines on most understanding In-context Learning tasks. Data and code will be available soon at https://github.com/xuyicheng-zju/UniICL.

Matteo Ferrante, Lisa Rinaldi, Francesca Botta et al.

Lesion segmentation is a crucial step of the radiomic workflow. Manual segmentation requires long execution time and is prone to variability, impairing the realisation of radiomic studies and their robustness. In this study, a deep-learning automatic segmentation method was applied on computed tomography images of non-small-cell lung cancer patients. The use of manual vs automatic segmentation in the performance of survival radiomic models was assessed, as well. METHODS A total of 899 NSCLC patients were included (2 proprietary: A and B, 1 public datasets: C). Automatic segmentation of lung lesions was performed by training a previously developed architecture, the nnU-Net, including 2D, 3D and cascade approaches. The quality of automatic segmentation was evaluated with DICE coefficient, considering manual contours as reference. The impact of automatic segmentation on the performance of a radiomic model for patient survival was explored by extracting radiomic hand-crafted and deep-learning features from manual and automatic contours of dataset A, and feeding different machine learning algorithms to classify survival above/below median. Models' accuracies were assessed and compared. RESULTS The best agreement between automatic and manual contours with DICE=0.78 +(0.12) was achieved by averaging predictions from 2D and 3D models, and applying a post-processing technique to extract the maximum connected component. No statistical differences were observed in the performances of survival models when using manual or automatic contours, hand-crafted, or deep features. The best classifier showed an accuracy between 0.65 and 0.78. CONCLUSION The promising role of nnU-Net for automatic segmentation of lung lesions was confirmed, dramatically reducing the time-consuming physicians' workload without impairing the accuracy of survival predictive models based on radiomics.

Zhucun Xue, Jiangning Zhang, Juntao Jiang et al.

The rapid expansion of research across machine learning, vision, and language has produced a volume of publications that is increasingly difficult to synthesize. Traditional bibliometric tools rely mainly on metadata and offer limited visibility into the semantic content of papers, making it hard to track how research themes evolve over time or how different areas influence one another. To obtain a clearer picture of recent developments, we compile a unified corpus of more than 100,000 papers from 22 major conferences between 2020 and 2025 and construct a multidimensional profiling pipeline to organize and analyze their textual content. By combining topic clustering, LLM-assisted parsing, and structured retrieval, we derive a comprehensive representation of research activity that supports the study of topic lifecycles, methodological transitions, dataset and model usage patterns, and institutional research directions. Our analysis highlights several notable shifts, including the growth of safety, multimodal reasoning, and agent-oriented studies, as well as the gradual stabilization of areas such as neural machine translation and graph-based methods. These findings provide an evidence-based view of how AI research is evolving and offer a resource for understanding broader trends and identifying emerging directions. Code and dataset: https://github.com/xzc-zju/Profiling_Scientific_Literature

Jiayi Luo, Qiyan Liu, Tengyang Wang et al.

Autoregressive (AR) video generation has emerged as a promising paradigm for long-horizon video synthesis, where each frame is generated conditioned on previously generated tokens. To accelerate inference, the KV cache is used to avoid redundant recomputation across generation steps. Nevertheless, its growth with generation length introduces increasing memory and error accumulation, limiting the scalability of AR models to even longer sequences. Existing KV cache compression methods mitigate this issue by selectively retaining only video tokens deemed important. However, most existing methods assess token importance using short-horizon signals derived from the current or historical generation context, making these methods prone to overlooking tokens that appear unimportant at early steps but later become critical for future frames. In this work, we identify an important property of trained AR video models: although RoPE-modulated queries evolve across autoregressive steps, the underlying canonical pre-RoPE query distribution remains remarkably stable throughout the video generation process. This approximate stationarity implies that future query distributions are estimable from historical statistics, enabling principled future-aware cache decisions without any additional training. Building on this insight, we propose Future Forcing, a training-free future-aware KV cache policy for AR video generation. Specifically, Future Forcing first constructs a future query proxy from historical statistics, then scores KV cache tokens by their importance under this proxy, and finally merges redundant token pairs within the affine subspace induced by the future query. Extensive experiments show that Future Forcing improves long-horizon consistency under limited KV caches, achieving up to 1.49 improvement in subject consistency on VBench-Long for 60s generation over existing AR video KV cache policies.

Xiaomin Yu, Yi Xin, Yuhui Zhang et al.

Despite the success of multimodal contrastive learning in aligning visual and linguistic representations, a persistent geometric anomaly, the Modality Gap, remains: embeddings of distinct modalities expressing identical semantics occupy systematically offset regions. Prior approaches to bridge this gap are largely limited by oversimplified isotropic assumptions, hindering their application in large-scale scenarios. In this paper, we address these limitations by precisely characterizing the geometric shape of the modality gap and leveraging it for efficient model scaling. First, we propose the Fixed-frame Modality Gap Theory, which decomposes the modality gap within a frozen reference frame into stable biases and anisotropic residuals. Guided by this precise modeling, we introduce ReAlign, a training-free modality alignment strategy. Utilizing statistics from massive unpaired data, ReAlign aligns text representation into the image representation distribution via a three-step process comprising Anchor, Trace, and Centroid Alignment, thereby explicitly rectifying geometric misalignment. Building on ReAlign, we propose ReVision, a scalable training paradigm for Multimodal Large Language Models~(MLLMs). ReVision integrates ReAlign into the pretraining stage, enabling the model to learn the distribution of visual representations from unpaired text before visual instruction tuning, without the need for large-scale, high-quality image-text pairs. Our framework demonstrates that statistically aligned unpaired data can effectively substitute for expensive image-text pairs, offering a robust path for the efficient scaling of MLLMs.

Yunhang Qian, Xiaobin Hu, Jiaquan Yu et al.

While Multi-Agent Systems (MAS) show potential for complex clinical decision support, the field remains hindered by architectural fragmentation and the lack of standardized multimodal integration. Current medical MAS research suffers from non-uniform data ingestion pipelines, inconsistent visual-reasoning evaluation, and a lack of cross-specialty benchmarking. To address these challenges, we present MedMASLab, a unified framework and benchmarking platform for multimodal medical multi-agent systems. MedMASLab introduces: (1) A standardized multimodal agent communication protocol that enables seamless integration of 11 heterogeneous MAS architectures across 24 medical modalities. (2) An automated clinical reasoning evaluator, a zero-shot semantic evaluation paradigm that overcomes the limitations of lexical string-matching by leveraging large vision-language models to verify diagnostic logic and visual grounding. (3) The most extensive benchmark to date, spanning 11 organ systems and 473 diseases, standardizing data from 11 clinical benchmarks. Our systematic evaluation reveals a critical domain-specific performance gap: while MAS improves reasoning depth, current architectures exhibit significant fragility when transitioning between specialized medical sub-domains. We provide a rigorous ablation of interaction mechanisms and cost-performance trade-offs, establishing a new technical baseline for future autonomous clinical systems. The source code and data is publicly available at: https://github.com/NUS-Project/MedMASLab/

Pengcheng Xu, Peng Tang, Donghao Luo et al. · tencent-ai

Unified Multimodal Models (UMMs) integrate multimodal understanding and generation, yet they are limited to maintaining visual consistency and disambiguating visual cues when referencing details across multiple input images. In this work, we propose a scalable multi-image editing framework for UMMs that explicitly distinguishes image identities and generalizes to variable input counts. Algorithmically, we introduce two innovations: 1) The learnable latent separators explicitly differentiate each reference image in the latent space, enabling accurate and disentangled conditioning. 2) The sinusoidal index encoding assigns visual tokens from the same image a continuous sinusoidal index embedding, which provides explicit image identity while allowing generalization and extrapolation on a variable number of inputs. To facilitate training and evaluation, we establish a high-fidelity benchmark using an inverse dataset construction methodology to guarantee artifact-free, achievable outputs. Experiments show clear improvements in semantic consistency, visual fidelity, and cross-image integration over prior baselines on diverse multi-image editing tasks, validating our advantages on consistency and generalization ability.

Xijie Huang, Chengming Xu, Donghao Luo et al. · tencent-ai

First-Frame Propagation (FFP) offers a promising paradigm for controllable video editing, but existing methods are hampered by a reliance on cumbersome run-time guidance. We identify the root cause of this limitation as the inadequacy of current training datasets, which are often too short, low-resolution, and lack the task diversity required to teach robust temporal priors. To address this foundational data gap, we first introduce FFP-300K, a new large-scale dataset comprising 300K high-fidelity video pairs at 720p resolution and 81 frames in length, constructed via a principled two-track pipeline for diverse local and global edits. Building on this dataset, we propose a novel framework designed for true guidance-free FFP that resolves the critical tension between maintaining first-frame appearance and preserving source video motion. Architecturally, we introduce Adaptive Spatio-Temporal RoPE (AST-RoPE), which dynamically remaps positional encodings to disentangle appearance and motion references. At the objective level, we employ a self-distillation strategy where an identity propagation task acts as a powerful regularizer, ensuring long-term temporal stability and preventing semantic drift. Comprehensive experiments on the EditVerseBench benchmark demonstrate that our method significantly outperforming existing academic and commercial models by receiving about 0.2 PickScore and 0.3 VLM score improvement against these competitors.

Zhihao Chen, Jiahui Wang, Yizhou Chen et al.

PET theranostics is transforming precision oncology, yet treatment response varies substantially; many patients receiving 177Lu-PSMA radioligand therapy (RLT) for metastatic castration-resistant prostate cancer (mCRPC) fail to respond, demanding reliable pre-therapy prediction. While LLM-based agents have shown remarkable potential in complex medical diagnosis, their application to PET theranostic outcome prediction remains unexplored, which faces three key challenges: (1) data and knowledge scarcity: RLT was only FDA-approved in 2022, yielding few training cases and insufficient domain knowledge in general LLMs; (2) heterogeneous information integration: robust prediction hinges on structured knowledge extraction from PET/CT, laboratory tests, and free-text clinical documentation; (3) evidence-grounded reasoning: clinical decisions must be anchored in trial evidence rather than LLM hallucinations. In this paper, we present TheraAgent, to our knowledge, the first agentic framework for PET theranostics, with three core innovations: (1) Multi-Expert Feature Extraction with Confidence-Weighted Consensus, where three specialized experts process heterogeneous inputs with uncertainty quantification; (2) Self-Evolving Agentic Memory (SEA-Mem), which learns prognostic patterns from accumulated cases, enabling case-based reasoning from limited data; (3) Evidence-Calibrated Reasoning, integrating a curated theranostics knowledge base to ground predictions in VISION/TheraP trial evidence. Evaluated on 35 real patients and 400 synthetic cases, TheraAgent achieves 75.7% overall accuracy on real patients and 87.0% on synthetic cases, outperforming MDAgents and MedAgent-Pro by over 20%. These results highlight a promising blueprint for trustworthy AI agents in PET theranostics, enabling trial-calibrated, multi-source decision support. Code will be released upon acceptance.

Jingxuan Wei, Siyuan Li, Yuhang Xu et al.

The construction of World Models capable of learning, simulating, and reasoning about objective physical laws constitutes a foundational challenge in the pursuit of Artificial General Intelligence. Recent advancements represented by video generation models like Sora have demonstrated the potential of data-driven scaling laws to approximate physical dynamics, while the emerging Unified Multimodal Model (UMM) offers a promising architectural paradigm for integrating perception, language, and reasoning. Despite these advances, the field still lacks a principled theoretical framework that defines the essential properties requisite for a General World Model. In this paper, we propose that a World Model must be grounded in the Trinity of Consistency: Modal Consistency as the semantic interface, Spatial Consistency as the geometric basis, and Temporal Consistency as the causal engine. Through this tripartite lens, we systematically review the evolution of multimodal learning, revealing a trajectory from loosely coupled specialized modules toward unified architectures that enable the synergistic emergence of internal world simulators. To complement this conceptual framework, we introduce CoW-Bench, a benchmark centered on multi-frame reasoning and generation scenarios. CoW-Bench evaluates both video generation models and UMMs under a unified evaluation protocol. Our work establishes a principled pathway toward general world models, clarifying both the limitations of current systems and the architectural requirements for future progress.

Zhangquan Chen, Manyuan Zhang, Xinlei Yu et al.

Dynamic spatial reasoning from monocular video is essential for bridging visual intelligence and the physical world, yet remains challenging for vision-language models (VLMs). Prior approaches either verbalize spatial-temporal reasoning entirely as text, which is inherently verbose and imprecise for complex dynamics, or rely on external geometric modules that increase inference complexity without fostering intrinsic model capability. In this paper, we present 4DThinker, the first framework that enables VLMs to "think with 4D" through dynamic latent mental imagery, i.e., internally simulating how scenes evolve within the continuous hidden space. Specifically, we first introduce a scalable, annotation-free data generation pipeline that synthesizes 4D reasoning data from raw videos. We then propose Dynamic-Imagery Fine-Tuning (DIFT), which jointly supervises textual tokens and 4D latents to ground the model in dynamic visual semantics. Building on this, 4D Reinforcement Learning (4DRL) further tackles complex reasoning tasks via outcome-based rewards, restricting policy gradients to text tokens to ensure stable optimization. Extensive experiments across multiple dynamic spatial reasoning benchmarks demonstrate that 4DThinker consistently outperforms strong baselines and offers a new perspective toward 4D reasoning in VLMs. Our code is available at https://github.com/zhangquanchen/4DThinker.

Qirui Yang, Yang Yang, Ying Zeng et al.

Text-guided diffusion models have greatly advanced image editing and generation. However, achieving physically consistent image retouching with precise parameter control (e.g., exposure, white balance, zoom) remains challenging. Existing methods either rely solely on ambiguous and entangled text prompts, which hinders precise camera control, or train separate heads/weights for parameter adjustment, which compromises scalability, multi-parameter composition, and sensitivity to subtle variations. To address these limitations, we propose CameraMaster, a unified camera-aware framework for image retouching. The key idea is to explicitly decouple the camera directive and then coherently integrate two critical information streams: a directive representation that captures the photographer's intent, and a parameter embedding that encodes precise camera settings. CameraMaster first uses the camera parameter embedding to modulate both the camera directive and the content semantics. The modulated directive is then injected into the content features via cross-attention, yielding a strongly camera-sensitive semantic context. In addition, the directive and camera embeddings are injected as conditioning and gating signals into the time embedding, enabling unified, layer-wise modulation throughout the denoising process and enforcing tight semantic-parameter alignment. To train and evaluate CameraMaster, we construct a large-scale dataset of 78K image-prompt pairs annotated with camera parameters. Extensive experiments show that CameraMaster produces monotonic and near-linear responses to parameter variations, supports seamless multi-parameter composition, and significantly outperforms existing methods.

Peng Tang, Zhiqiang Xu, Pengfei Wei et al.

We propose an efficient deep learning method for single image defocus deblurring (SIDD) by further exploring inverse kernel properties. Although the current inverse kernel method, i.e., kernel-sharing parallel atrous convolution (KPAC), can address spatially varying defocus blurs, it has difficulty in handling large blurs of this kind. To tackle this issue, we propose a Residual and Recursive Kernel-sharing Atrous Convolution (R$^2$KAC). R$^2$KAC builds on a significant observation of inverse kernels, that is, successive use of inverse-kernel-based deconvolutions with fixed size helps remove unexpected large blurs but produces ringing artifacts. Specifically, on top of kernel-sharing atrous convolutions used to simulate multi-scale inverse kernels, R$^2$KAC applies atrous convolutions recursively to simulate a large inverse kernel. Specifically, on top of kernel-sharing atrous convolutions, R$^2$KAC stacks atrous convolutions recursively to simulate a large inverse kernel. To further alleviate the contingent effect of recursive stacking, i.e., ringing artifacts, we add identity shortcuts between atrous convolutions to simulate residual deconvolutions. Lastly, a scale recurrent module is embedded in the R$^2$KAC network, leading to SR-R$^2$KAC, so that multi-scale information from coarse to fine is exploited to progressively remove the spatially varying defocus blurs. Extensive experimental results show that our method achieves the state-of-the-art performance.

Weixiang Shen, Bailiang Jian, Jun Li et al.

Tool-augmented large language model (LLM) agents can orchestrate specialist classifiers, segmentation models, and visual question-answering modules to interpret chest X-rays. However, these agents still solve each case in isolation: they fail to accumulate experience across cases, correct recurrent reasoning mistakes, or adapt their tool-use behavior without expensive reinforcement learning. While a radiologist naturally improves with every case, current agents remain static. In this work, we propose Evo-MedAgent, a self-evolving memory module that equips a medical agent with the capacity for inter-case learning at test time. Our memory comprises three complementary stores: (1)~\emph{Retrospective Clinical Episodes} that retrieve problem-solving experiences from similar past cases, (2)~an \emph{Adaptive Procedural Heuristics} bank curating priority-tagged diagnostic rules that evolves via reflection, much like a physician refining their internal criteria, and (3)~a \emph{Tool Reliability Controller} that tracks per-tool trustworthiness. On ChestAgentBench, Evo-MedAgent raises multiple-choice question (MCQ) accuracy from 0.68 to 0.79 on GPT-5-mini, and from 0.76 to 0.87 on Gemini-3 Flash. With a strong base model, evolving memory improves performance more effectively than orchestrating external tools on qualitative diagnostic tasks. Because Evo-MedAgent requires no training, its per-case overhead is bounded by one additional retrieval pass and a single reflection call, making it deployable on top of any frozen model.

Hangyu Lin, Chao Wen, Chengming Xu et al.

Flow matching based video generative models have been increasingly relying on prepended Vision-Language Models (VLMs) to handle complex, instruction-based video editing. The prevailing assumption underlying this paradigm is that a connector module can seamlessly align the VLM's rich multi-modal reasoning with the original text embedding space of DiTs. However, we hypothesize that this alignment acts as a severe semantic bottleneck, degrading fine-grained structural variables. Verifying this is challenging, as end-to-end evaluations conflate alignment failures with generation errors, and natural datasets lack disentangled annotations. To rigorously investigate this, we propose a controlled data processing pipeline based on video composition that results in TRACE-Edit, a diagnostic dataset focusing on relation-based editing. Leveraging this dataset, we propose a comprehensive diagnostic protocol to analyze two important designs of meta-query and connector in the existing video editing models. Systematic evaluation of four representative model cases reveals that fine-grained structural semantics can be severely degraded during alignment. Our findings overturn the assumption of lossless semantic transfer, identifying the VLM-to-DiT alignment as a major bottleneck and providing a new diagnostic foundation for future multi-modal alignment architectures.

Jiangning Zhang, Junwei Zhu, Teng Hu et al.

Native 4K (2160$\times$3840) video generation remains a critical challenge due to the quadratic computational explosion of full-attention as spatiotemporal resolution increases, making it difficult for models to strike a balance between efficiency and quality. This paper proposes a novel Transformer retrofit strategy termed $\textbf{T3}$ ($\textbf{T}$ransform $\textbf{T}$rained $\textbf{T}$ransformer) that, without altering the core architecture of full-attention pretrained models, significantly reduces compute requirements by optimizing their forward logic. Specifically, $\textbf{T3-Video}$ introduces a multi-scale weight-sharing window attention mechanism and, via hierarchical blocking together with an axis-preserving full-attention design, can effect an "attention pattern" transformation of a pretrained model using only modest compute and data. Results on 4K-VBench show that $\textbf{T3-Video}$ substantially outperforms existing approaches: while delivering performance improvements (+4.29$\uparrow$ VQA and +0.08$\uparrow$ VTC), it accelerates native 4K video generation by more than 10$\times$. Project page at https://zhangzjn.github.io/projects/T3-Video

Yongqi Yang, Huayang Huang, Xu Peng et al.

Recent hybrid video generation models combine autoregressive temporal dynamics with diffusion-based spatial denoising, but their sequential, iterative nature leads to error accumulation and long inference times. In this work, we propose a distillation-based framework for efficient causal video generation that enables high-quality synthesis with extremely limited denoising steps. Our approach builds upon the Distribution Matching Distillation (DMD) framework and proposes a novel Adversarial Self-Distillation (ASD) strategy, which aligns the outputs of the student model's n-step denoising process with its (n+1)-step version at the distribution level. This design provides smoother supervision by bridging small intra-student gaps and more informative guidance by combining teacher knowledge with locally consistent student behavior, substantially improving training stability and generation quality in extremely few-step scenarios (e.g., 1-2 steps). In addition, we present a First-Frame Enhancement (FFE) strategy, which allocates more denoising steps to the initial frames to mitigate error propagation while applying larger skipping steps to later frames. Extensive experiments on VBench demonstrate that our method surpasses state-of-the-art approaches in both one-step and two-step video generation. Notably, our framework produces a single distilled model that flexibly supports multiple inference-step settings, eliminating the need for repeated re-distillation and enabling efficient, high-quality video synthesis.

Caoshuo Li, Zengmao Ding, Xiaobin Hu et al.

As one of the earliest writing systems, Oracle Bone Script (OBS) preserves the cultural and intellectual heritage of ancient civilizations. However, current OBS research faces two major challenges: (1) the interpretation of OBS involves a complex workflow comprising multiple serial and parallel sub-tasks, and (2) the efficiency of OBS information organization and retrieval remains a critical bottleneck, as scholars often spend substantial effort searching for, compiling, and managing relevant resources. To address these challenges, we present OracleAgent, the first agent system designed for the structured management and retrieval of OBS-related information. OracleAgent seamlessly integrates multiple OBS analysis tools, empowered by large language models (LLMs), and can flexibly orchestrate these components. Additionally, we construct a comprehensive domain-specific multimodal knowledge base for OBS, which is built through a rigorous multi-year process of data collection, cleaning, and expert annotation. The knowledge base comprises over 1.4M single-character rubbing images and 80K interpretation texts. OracleAgent leverages this resource through its multimodal tools to assist experts in retrieval tasks of character, document, interpretation text, and rubbing image. Extensive experiments demonstrate that OracleAgent achieves superior performance across a range of multimodal reasoning and generation tasks, surpassing leading mainstream multimodal large language models (MLLMs) (e.g., GPT-4o). Furthermore, our case study illustrates that OracleAgent can effectively assist domain experts, significantly reducing the time cost of OBS research. These results highlight OracleAgent as a significant step toward the practical deployment of OBS-assisted research and automated interpretation systems.

Zhifei Xie, Kaiyu Pang, Haobin Zhang et al.

Despite rapid advances in automatic speech recognition (ASR) and large audio-language models, robust recognition in real-world environments remains limited by an "acoustic robustness bottleneck": models often lose acoustic grounding and produce omissions or hallucinations under severe, compositional distortions. We propose Mega-ASR, a unified ASR-in-the-wild framework that combines scalable compound-data construction with progressive acoustic-to-semantic optimization. We introduce Voices-in-the-Wild-2M, covering 7 classic acoustic phenomena and 54 physically plausible compound scenarios, and train Mega-ASR with Acoustic-to-Semantic Progressive Supervised Fine-Tuning and Dual-Granularity WER-Gated Policy Optimization. Extensive experiments demonstrate that Mega-ASR achieves significant advantages over prior state-of-the-art systems on adverse-condition ASR benchmarks (45.69% vs. 54.01% on VOiCES R4-B-F, and 21.49% vs. 29.34% on NOIZEUS Sta-0). On complex compositional acoustic scenarios, Mega-ASR further delivers over 30% relative WER reduction against strong open- and closed-source baselines, establishing a scalable paradigm for robust ASR in-the-wild.

Juntao Jiang, Jiangning Zhang, Yali Bi et al.

Chain-of-Thought (CoT) reasoning has proven effective in enhancing large language models by encouraging step-by-step intermediate reasoning, and recent advances have extended this paradigm to Multimodal Large Language Models (MLLMs). In the medical domain, where diagnostic decisions depend on nuanced visual cues and sequential reasoning, CoT aligns naturally with clinical thinking processes. However, Current benchmarks for medical image understanding generally focus on the final answer while ignoring the reasoning path. An opaque process lacks reliable bases for judgment, making it difficult to assist doctors in diagnosis. To address this gap, we introduce a new M3CoTBench benchmark specifically designed to evaluate the correctness, efficiency, impact, and consistency of CoT reasoning in medical image understanding. M3CoTBench features 1) a diverse, multi-level difficulty dataset covering 24 examination types, 2) 13 varying-difficulty tasks, 3) a suite of CoT-specific evaluation metrics (correctness, efficiency, impact, and consistency) tailored to clinical reasoning, and 4) a performance analysis of multiple MLLMs. M3CoTBench systematically evaluates CoT reasoning across diverse medical imaging tasks, revealing current limitations of MLLMs in generating reliable and clinically interpretable reasoning, and aims to foster the development of transparent, trustworthy, and diagnostically accurate AI systems for healthcare. Project page at https://juntaojianggavin.github.io/projects/M3CoTBench/.

Wencan Jiang, Jiangning Zhang, Jianbiao Mei et al.

Long-horizon multimodal agents in open-world games must stay goal-directed across many low-level interactions under tight token and latency budgets. Existing approaches often trade off costly per-step reasoning against reactive execution that can drift, repeat failures, and recover poorly. Our key idea is to reuse strategic reasoning across locally stable segments and reinvoke it at event boundaries. We present SPIKE, an adaptive dual controller framework for cost-efficient long-horizon game control. Its Strategic Controller performs low-frequency global planning, failure analysis, and recovery, while its Reactive Controller handles fast local execution under a strict token budget. An Event Trigger monitors visual change, task progress, repeated actions, and failure signals to decide when control should stay reactive or escalate to strategic reasoning. Hierarchical Memory separates short-term experience reuse in the State-Action Memory Bank (SA-MB) from structured evidence in the State Action Knowledge Graph (SA-KG), allowing each controller to retrieve the context it needs. This design reuses strategic proposals over multiple reactive steps, supports local override when plans become stale, and reserves expensive reasoning for moments where extra deliberation is useful. On the Lite-100 split of StarDojo, SPIKE improves Lite-100 success rate (SR) by 5.0 percentage points (38.5% relative) over the strongest Lite-100 baseline and Budgeted SR by 9.3 points (75.6% relative) over the strongest budgeted baseline. It also reduces token consumption by 54.9% and latency by 40.8%. Ablations show that event triggering, reactive override, and heterogeneous memory each contribute to success and recovery, supporting selective reasoning rather than reasoning at every step.

Xinlei Yu, Chengming Xu, Guibin Zhang et al.

Multi-Agent System (MAS) powered by Visual Language Models (VLMs) enables challenging tasks but suffers from a novel failure term, multi-agent visual hallucination snowballing, where hallucinations are seeded in a single agent and amplified by following ones due to the over-reliance on textual flow to relay visual information. Through turn-, layer-, and token-wise attention analyses, we provide detailed insights into the essence of hallucination snowballing regarding the reduction of visual attention allocation. It leads us to identify a subset of vision tokens with a unimodal attention peak in middle layers that best preserve visual evidence but gradually diminish in deeper agent turns, resulting in the visual hallucination snowballing in MAS. Thus, we propose ViF, a lightweight, plug-and-play mitigation paradigm that relays inter-agent messages with Visual Flow powered by the selected visual relay tokens and applies attention reallocation to amplify this pattern. The experiment results demonstrate that our method markedly reduces hallucination snowballing, consistently improving the performance across eight benchmarks based on four common MAS structures and ten base models. The source code is publicly available at: https://github.com/YU-deep/ViF.git.

Yuji Wang, Moran Li, Xiaobin Hu et al.

Identity-preserving text-to-video (IPT2V) generation, which aims to create high-fidelity videos with consistent human identity, has become crucial for downstream applications. However, current end-to-end frameworks suffer a critical spatial-temporal trade-off: optimizing for spatially coherent layouts of key elements (e.g., character identity preservation) often compromises instruction-compliant temporal smoothness, while prioritizing dynamic realism risks disrupting the spatial coherence of visual structures. To tackle this issue, we propose a simple yet effective spatial-temporal decoupled framework that decomposes representations into spatial features for layouts and temporal features for motion dynamics. Specifically, our paper proposes a semantic prompt optimization mechanism and stage-wise decoupled generation paradigm. The former module decouples the prompt into spatial and temporal components. Aligned with the subsequent stage-wise decoupled approach, the spatial prompts guide the text-to-image (T2I) stage to generate coherent spatial features, while the temporal prompts direct the sequential image-to-video (I2V) stage to ensure motion consistency. Experimental results validate that our approach achieves excellent spatiotemporal consistency, demonstrating outstanding performance in identity preservation, text relevance, and video quality. By leveraging this simple yet robust mechanism, our algorithm secures the runner-up position in 2025 ACM MultiMedia Challenge. Our code is available at https://github.com/rain152/IPVG.

Xingyu Zhou, Qifan Li, Xiaobin Hu et al.

The diffusion model presents a powerful ability to capture the entire (conditional) data distribution. However, due to the lack of sufficient training and data to learn to cover low-probability areas, the model will be penalized for failing to generate high-quality images corresponding to these areas. To achieve better generation quality, guidance strategies such as classifier free guidance (CFG) can guide the samples to the high-probability areas during the sampling stage. However, the standard CFG often leads to over-simplified or distorted samples. On the other hand, the alternative line of guiding diffusion model with its bad version is limited by carefully designed degradation strategies, extra training and additional sampling steps. In this paper, we proposed a simple yet effective strategy Internal Guidance (IG), which introduces an auxiliary supervision on the intermediate layer during training process and extrapolates the intermediate and deep layer's outputs to obtain generative results during sampling process. This simple strategy yields significant improvements in both training efficiency and generation quality on various baselines. On ImageNet 256x256, SiT-XL/2+IG achieves FID=5.31 and FID=1.75 at 80 and 800 epochs. More impressively, LightningDiT-XL/1+IG achieves FID=1.34 which achieves a large margin between all of these methods. Combined with CFG, LightningDiT-XL/1+IG achieves the current state-of-the-art FID of 1.19.

Yuyang Hu, Shichun Liu, Yanwei Yue et al.

Memory has emerged, and will continue to remain, a core capability of foundation model-based agents. As research on agent memory rapidly expands and attracts unprecedented attention, the field has also become increasingly fragmented. Existing works that fall under the umbrella of agent memory often differ substantially in their motivations, implementations, and evaluation protocols, while the proliferation of loosely defined memory terminologies has further obscured conceptual clarity. Traditional taxonomies such as long/short-term memory have proven insufficient to capture the diversity of contemporary agent memory systems. This work aims to provide an up-to-date landscape of current agent memory research. We begin by clearly delineating the scope of agent memory and distinguishing it from related concepts such as LLM memory, retrieval augmented generation (RAG), and context engineering. We then examine agent memory through the unified lenses of forms, functions, and dynamics. From the perspective of forms, we identify three dominant realizations of agent memory, namely token-level, parametric, and latent memory. From the perspective of functions, we propose a finer-grained taxonomy that distinguishes factual, experiential, and working memory. From the perspective of dynamics, we analyze how memory is formed, evolved, and retrieved over time. To support practical development, we compile a comprehensive summary of memory benchmarks and open-source frameworks. Beyond consolidation, we articulate a forward-looking perspective on emerging research frontiers, including memory automation, reinforcement learning integration, multimodal memory, multi-agent memory, and trustworthiness issues. We hope this survey serves not only as a reference for existing work, but also as a conceptual foundation for rethinking memory as a first-class primitive in the design of future agentic intelligence.

Xiaobin Hu, Enpu Zuo, Lanping Hu et al.

Privacy protection has become a critical requirement in the era of ubiquitous visual data sharing, imposing higher demands on efficient and robust privacy detection algorithms. However, current robust detection models are severely hindered by the lack of comprehensive datasets. Existing privacy-oriented datasets often suffer from limited scale, coarse-grained annotations, and narrow domain coverage, failing to capture the intricate details of sensitive information in realworld environments. To bridge this gap, we present a large-scale, fine-grained Visual Privacy Dataset (VPD-100K), designed to facilitate generalized privacy detection. We establish a holistic taxonomy comprising four primary domains: Human Presence, On-Screen Personally Identifiable Information (PII), Physical Identifiers, and Location Indicators, containing 100,000 images annotated with 33 fine-grained classes and over 190,000 object instances. Statistical analysis reveals that our dataset features long-tailed distributions, small object scales, and high visual complexity. These characteristics make the dataset particularly valuable for demanding, unconstrained applications such as live streaming, where actors frequently face unintentional, realtime information leakage. Furthermore, we design an effective frequency-enhanced lightweight module consisting of frequency-domain attention fusion and adaptive spectral gating mechanism that breaks the limitations of spatial pixel intensity to better capture the subtle details of sensitive information. Extensive experiments conducted on both diverse image and streaming videos benchmarks consistently demonstrate the effectiveness of our VPD-100K dataset and the wellcurated frequency mechanism. The code and dataset are available at https://vpd-100k.github.io/.

Haiyue Sun, Qingdong He, Jinlong Peng et al.

Autoregressive Model (AR) has shown remarkable success in conditional image generation. However, these approaches for multiple reference generation struggle with decoupling different reference identities. In this work, we propose the TokenAR framework, specifically focused on a simple but effective token-level enhancement mechanism to address reference identity confusion problem. Such token-level enhancement consists of three parts, 1). Token Index Embedding clusters the tokens index for better representing the same reference images; 2). Instruct Token Injection plays as a role of extra visual feature container to inject detailed and complementary priors for reference tokens; 3). The identity-token disentanglement strategy (ITD) explicitly guides the token representations toward independently representing the features of each identity.This token-enhancement framework significantly augments the capabilities of existing AR based methods in conditional image generation, enabling good identity consistency while preserving high quality background reconstruction. Driven by the goal of high-quality and high-diversity in multi-subject generation, we introduce the InstructAR Dataset, the first open-source, large-scale, multi-reference input, open domain image generation dataset that includes 28K training pairs, each example has two reference subjects, a relative prompt and a background with mask annotation, curated for multiple reference image generation training and evaluating. Comprehensive experiments validate that our approach surpasses current state-of-the-art models in multiple reference image generation task. The implementation code and datasets will be made publicly. Codes are available, see https://github.com/lyrig/TokenAR

Yuansen Liu, Haiming Tang, Jinlong Peng et al. · tencent-ai

Multimodal Large Language Models (MLLMs) have demonstrated significant advances in visual understanding tasks. However, their capacity to comprehend human-centric scenes has rarely been explored, primarily due to the absence of comprehensive evaluation benchmarks that take into account both the human-oriented granular level and higher-dimensional causal reasoning ability. Such high-quality evaluation benchmarks face tough obstacles, given the physical complexity of the human body and the difficulty of annotating granular structures. In this paper, we propose Human-MME, a curated benchmark designed to provide a more holistic evaluation of MLLMs in human-centric scene understanding. Compared with other existing benchmarks, our work provides three key features: 1. Diversity in human scene, spanning 4 primary visual domains with 15 secondary domains and 43 sub-fields to ensure broad scenario coverage. 2. Progressive and diverse evaluation dimensions, evaluating the human-based activities progressively from the human-oriented granular perception to the higher-dimensional reasoning, consisting of eight dimensions with 19,945 real-world image question pairs and an evaluation suite. 3. High-quality annotations with rich data paradigms, constructing the automated annotation pipeline and human-annotation platform, supporting rigorous manual labeling to facilitate precise and reliable model assessment. Our benchmark extends the single-target understanding to the multi-person and multi-image mutual understanding by constructing the choice, short-answer, grounding, ranking and judgment question components, and complex questions of their combination. The extensive experiments on 17 state-of-the-art MLLMs effectively expose the limitations and guide future MLLMs research toward better human-centric image understanding. All data and code are available at https://github.com/Yuan-Hou/Human-MME.

Kun Wang, Guibin Zhang, ManKit Ye et al.

The past two years have witnessed the meteoric rise of Large Language Model (LLM)-powered multi-agent systems (MAS), which harness collective intelligence and exhibit a remarkable trajectory toward self-evolution. This paradigm has rapidly progressed from manually engineered systems that require bespoke configuration of prompts, tools, roles, and communication protocols toward frameworks capable of automated orchestration. Yet, dominant automatic multi-agent systems, whether generated by external modules or a single LLM agent, largely adhere to a rigid ``\textit{generate-once-and-deploy}'' paradigm, rendering the resulting systems brittle and ill-prepared for the dynamism and uncertainty of real-world environments. To transcend this limitation, we introduce MAS$^2$, a paradigm predicated on the principle of recursive self-generation: a multi-agent system that autonomously architects bespoke multi-agent systems for diverse problems. Technically, we devise a ``\textit{generator-implementer-rectifier}'' tri-agent team capable of dynamically composing and adaptively rectifying a target agent system in response to real-time task demands. Collaborative Tree Optimization is proposed to train and specialize these meta-agents. Extensive evaluation across seven benchmarks reveals that MAS$^2$ achieves performance gains of up to $19.6\%$ over state-of-the-art MAS in complex scenarios such as deep research and code generation. Moreover, MAS$^2$ exhibits superior cross-backbone generalization, effectively leveraging previously unseen LLMs to yield improvements of up to $15.1\%$. Crucially, these gains are attained without incurring excessive token costs, as MAS$^2$ consistently resides on the Pareto frontier of cost-performance trade-offs. The source codes are available at https://github.com/yeyeyeah2/MAS2.

Yuji Wang, Moran Li, Xiaobin Hu et al.

Current video generation models struggle with identity preservation under large facial angles, primarily facing two challenges: the difficulty in exploring an effective mechanism to integrate identity features into DiT structure, and the lack of targeted coverage of large facial angles in existing open-source video datasets. To address these, we present two key innovations. First, we introduce a Mixture of Facial Experts (MoFE) that dynamically combines complementary cues from three specialized experts, each designed to capture distinct but mutually reinforcing aspects of facial attributes. The identity expert captures cross-pose identity-sensitive features, the semantic expert extracts high-level visual semantxics, and the detail expert preserves pixel-level features (e.g., skin texture, color gradients). Furthermore, to mitigate dataset limitations, we have tailored a data processing pipeline centered on two key aspects: Face Constraints and Identity Consistency. Face Constraints ensure facial angle diversity and a high proportion of facial regions, while Identity Consistency preserves coherent person-specific features across temporal sequences, collectively addressing the scarcity of large facial angles and identity-stable training data in existing datasets. Leveraging this pipeline, we have curated and refined a Large Face Angles (LFA) Dataset from existing open-source human video datasets, comprising 460K video clips with annotated facial angles. Experimental results on the LFA benchmark demonstrate that our method, empowered by the LFA dataset, significantly outperforms prior SOTA methods in face similarity, face FID, and CLIP semantic alignment. The code and dataset will be made publicly available at https://github.com/rain152/LFA-Video-Generation.

Na Zhang, Moran Li, Chengming Xu et al.

Realistic hair strand generation is crucial for applications like computer graphics and virtual reality. While diffusion models can generate hairstyles from text or images, these inputs lack precision and user-friendliness. Instead, we propose the first sketch-based strand generation model, which offers finer control while remaining user-friendly. Our framework tackles key challenges, such as modeling complex strand interactions and diverse sketch patterns, through two main innovations: a learnable strand upsampling strategy that encodes 3D strands into multi-scale latent spaces, and a multi-scale adaptive conditioning mechanism using a transformer with diffusion heads to ensure consistency across granularity levels. Experiments on several benchmark datasets show our method outperforms existing approaches in realism and precision. Qualitative results further confirm its effectiveness. Code will be released at [GitHub](https://github.com/fighting-Zhang/StrandDesigner).

Lingjie Kong, Kai Wu, Xiaobin Hu et al.

Recent advances in diffusion-based text-to-image models have simplified creating high-fidelity images, but preserving the identity (ID) of specific elements, like a personal dog, is still challenging. Object customization, using reference images and textual descriptions, is key to addressing this issue. Current object customization methods are either object-specific, requiring extensive fine-tuning, or object-agnostic, offering zero-shot customization but limited to specialized domains. The primary issue of promoting zero-shot object customization from specific domains to the general domain is to establish a large-scale general ID dataset for model pre-training, which is time-consuming and labor-intensive. In this paper, we propose a novel pipeline to construct a large dataset of general objects and build the Multi-Category ID-Consistent (MC-IDC) dataset, featuring 315k text-image samples across 10k categories. With the help of MC-IDC, we introduce Customizing Anything (CustAny), a zero-shot framework that maintains ID fidelity and supports flexible text editing for general objects. CustAny features three key components: a general ID extraction module, a dual-level ID injection module, and an ID-aware decoupling module, allowing it to customize any object from a single reference image and text prompt. Experiments demonstrate that CustAny outperforms existing methods in both general object customization and specialized domains like human customization and virtual try-on. Our contributions include a large-scale dataset, the CustAny framework and novel ID processing to advance this field. Code and dataset will be released soon in https://github.com/LingjieKong-fdu/CustAny.

Yinan Chen, Jiangning Zhang, Yali Bi et al.

Image inversion is a fundamental task in generative models, aiming to map images back to their latent representations to enable downstream applications such as editing, restoration, and style transfer. This paper provides a comprehensive review of the latest advancements in image inversion techniques, focusing on two main paradigms: Generative Adversarial Network (GAN) inversion and diffusion model inversion. We categorize these techniques based on their optimization methods. For GAN inversion, we systematically classify existing methods into encoder-based approaches, latent optimization approaches, and hybrid approaches, analyzing their theoretical foundations, technical innovations, and practical trade-offs. For diffusion model inversion, we explore training-free strategies, fine-tuning methods, and the design of additional trainable modules, highlighting their unique advantages and limitations. Additionally, we discuss several popular downstream applications and emerging applications beyond image tasks, identifying current challenges and future research directions. By synthesizing the latest developments, this paper aims to provide researchers and practitioners with a valuable reference resource, promoting further advancements in the field of image inversion. We keep track of the latest works at https://github.com/RyanChenYN/ImageInversion

Zhiyao Wang, Xu Chen, Chengming Xu et al.

Face Restoration (FR) is a crucial area within image and video processing, focusing on reconstructing high-quality portraits from degraded inputs. Despite advancements in image FR, video FR remains relatively under-explored, primarily due to challenges related to temporal consistency, motion artifacts, and the limited availability of high-quality video data. Moreover, traditional face restoration typically prioritizes enhancing resolution and may not give as much consideration to related tasks such as facial colorization and inpainting. In this paper, we propose a novel approach for the Generalized Video Face Restoration (GVFR) task, which integrates video BFR, inpainting, and colorization tasks that we empirically show to benefit each other. We present a unified framework, termed as stable video face restoration (SVFR), which leverages the generative and motion priors of Stable Video Diffusion (SVD) and incorporates task-specific information through a unified face restoration framework. A learnable task embedding is introduced to enhance task identification. Meanwhile, a novel Unified Latent Regularization (ULR) is employed to encourage the shared feature representation learning among different subtasks. To further enhance the restoration quality and temporal stability, we introduce the facial prior learning and the self-referred refinement as auxiliary strategies used for both training and inference. The proposed framework effectively combines the complementary strengths of these tasks, enhancing temporal coherence and achieving superior restoration quality. This work advances the state-of-the-art in video FR and establishes a new paradigm for generalized video face restoration. Code and video demo are available at https://github.com/wangzhiyaoo/SVFR.git.

Xiaomin Yu, Yijiang Li, Yuhui Zhang et al.

Training multimodal large language models has long been limited by the scarcity of high-quality paired multimodal data. Recent studies show that the shared representation space of pretrained multimodal contrastive models can serve as a bridge, enabling models to perform multimodal training with unimodal data. However, the key premise of this paradigm remains insufficiently understood: can representations from different modalities be reliably interchanged? The core obstacle lies in the persistent Modality Gap in the shared space. In this work, we revisit the geometric nature of the modality gap. We find that modality representations already share compatible dominant semantic geometry. What truly hinders modality interchangeability is not a simple global shift, but an anisotropic residual structure concentrated along a small number of dominant directions. Based on this finding, we further propose the principle of anisotropic modality gap alignment: effective modality alignment should align with the target-modality distribution while preserving the semantic structure of the source modality. Guided by this principle, we propose an anisotropic geometric correction framework, AnisoAlign, for unpaired modality alignment. This framework leverages the internal geometric prior of the target modality and performs bounded correction on source-modality representations, thereby constructing substitute representations in the target modality. Experiments confirm its benefits in both geometric diagnostics and text-only MLLM training. Overall, this work recasts the modality gap from an empirical observation into a correctable, structured geometric phenomenon and provides a new representation alignment perspective for training multimodal models with unimodal data.

Qingdong He, Chaoyi Wang, Peng Tang et al.

Video subtitle removal aims to distinguish text overlays from background content while preserving temporal coherence. Existing diffusion-based methods necessitate explicit mask sequences during both training and inference phases, which restricts their practical deployment. In this paper, we present CLEAR (Context-aware Learning for End-to-end Adaptive Video Subtitle Removal), a mask-free framework that achieves truly end-to-end inference through context-aware adaptive learning. Our two-stage design decouples prior extraction from generative refinement: Stage I learns disentangled subtitle representations via self-supervised orthogonality constraints on dual encoders, while Stage II employs LoRA-based adaptation with generation feedback for dynamic context adjustment. Notably, our method only requires 0.77% of the parameters of the base diffusion model for training. On Chinese subtitle benchmarks, CLEAR outperforms mask-dependent baselines by + 6.77dB PSNR and -74.7% VFID, while demonstrating superior zero-shot generalization across six languages (English, Korean, French, Japanese, Russian, German), a performance enabled by our generation-driven feedback mechanism that ensures robust subtitle removal without ground-truth masks during inference.