Jinfa Huang

Jinfa Huang, Jinsheng Pan, Zhongwei Wan et al.

Recent advances show that two-stream approaches have achieved outstanding performance in hateful meme detection. However, hateful memes constantly evolve as new memes emerge by fusing progressive cultural ideas, making existing methods obsolete or ineffective. In this work, we explore the potential of Large Multimodal Models (LMMs) for hateful meme detection. To this end, we propose Evolver, which incorporates LMMs via Chain-of-Evolution (CoE) Prompting, by integrating the evolution attribute and in-context information of memes. Specifically, Evolver simulates the evolving and expressing process of memes and reasons through LMMs in a step-by-step manner. First, an evolutionary pair mining module retrieves the top-k most similar memes in the external curated meme set with the input meme. Second, an evolutionary information extractor is designed to summarize the semantic regularities between the paired memes for prompting. Finally, a contextual relevance amplifier enhances the in-context hatefulness information to boost the search for evolutionary processes. Extensive experiments on public FHM, MAMI, and HarM datasets show that CoE prompting can be incorporated into existing LMMs to improve their performance. More encouragingly, it can serve as an interpretive tool to promote the understanding of the evolution of social memes. [Homepage] (https://github.com/inFaaa/Evolver)

Peng Jin, Jinfa Huang, Fenglin Liu et al. · oxford

Most video-and-language representation learning approaches employ contrastive learning, e.g., CLIP, to project the video and text features into a common latent space according to the semantic similarities of text-video pairs. However, such learned shared latent spaces are not often optimal, and the modality gap between visual and textual representation can not be fully eliminated. In this paper, we propose Expectation-Maximization Contrastive Learning (EMCL) to learn compact video-and-language representations. Specifically, we use the Expectation-Maximization algorithm to find a compact set of bases for the latent space, where the features could be concisely represented as the linear combinations of these bases. Such feature decomposition of video-and-language representations reduces the rank of the latent space, resulting in increased representing power for the semantics. Extensive experiments on three benchmark text-video retrieval datasets prove that our EMCL can learn more discriminative video-and-language representations than previous methods, and significantly outperform previous state-of-the-art methods across all metrics. More encouragingly, the proposed method can be applied to boost the performance of existing approaches either as a jointly training layer or an out-of-the-box inference module with no extra training, making it easy to be incorporated into any existing methods.

Fengyang Xiao, Sujie Hu, Yuqi Shen et al. · tsinghua

Camouflaged Object Detection (COD) refers to the task of identifying and segmenting objects that blend seamlessly into their surroundings, posing a significant challenge for computer vision systems. In recent years, COD has garnered widespread attention due to its potential applications in surveillance, wildlife conservation, autonomous systems, and more. While several surveys on COD exist, they often have limitations in terms of the number and scope of papers covered, particularly regarding the rapid advancements made in the field since mid-2023. To address this void, we present the most comprehensive review of COD to date, encompassing both theoretical frameworks and practical contributions to the field. This paper explores various COD methods across four domains, including both image-level and video-level solutions, from the perspectives of traditional and deep learning approaches. We thoroughly investigate the correlations between COD and other camouflaged scenario methods, thereby laying the theoretical foundation for subsequent analyses. Beyond object-level detection, we also summarize extended methods for instance-level tasks, including camouflaged instance segmentation, counting, and ranking. Additionally, we provide an overview of commonly used benchmarks and evaluation metrics in COD tasks, conducting a comprehensive evaluation of deep learning-based techniques in both image and video domains, considering both qualitative and quantitative performance. Finally, we discuss the limitations of current COD models and propose 9 promising directions for future research, focusing on addressing inherent challenges and exploring novel, meaningful technologies. For those interested, a curated list of COD-related techniques, datasets, and additional resources can be found at https://github.com/ChunmingHe/awesome-concealed-object-segmentation

Hongjian Zhou, Fenglin Liu, Boyang Gu et al.

Large language models (LLMs), such as ChatGPT, have received substantial attention due to their capabilities for understanding and generating human language. While there has been a burgeoning trend in research focusing on the employment of LLMs in supporting different medical tasks (e.g., enhancing clinical diagnostics and providing medical education), a review of these efforts, particularly their development, practical applications, and outcomes in medicine, remains scarce. Therefore, this review aims to provide a detailed overview of the development and deployment of LLMs in medicine, including the challenges and opportunities they face. In terms of development, we provide a detailed introduction to the principles of existing medical LLMs, including their basic model structures, number of parameters, and sources and scales of data used for model development. It serves as a guide for practitioners in developing medical LLMs tailored to their specific needs. In terms of deployment, we offer a comparison of the performance of different LLMs across various medical tasks, and further compare them with state-of-the-art lightweight models, aiming to provide an understanding of the advantages and limitations of LLMs in medicine. Overall, in this review, we address the following questions: 1) What are the practices for developing medical LLMs 2) How to measure the medical task performance of LLMs in a medical setting? 3) How have medical LLMs been employed in real-world practice? 4) What challenges arise from the use of medical LLMs? and 5) How to more effectively develop and deploy medical LLMs? By answering these questions, this review aims to provide insights into the opportunities for LLMs in medicine and serve as a practical resource. We also maintain a regularly updated list of practical guides on medical LLMs at https://github.com/AI-in-Health/MedLLMsPracticalGuide

Hao Li, Jinfa Huang, Peng Jin et al.

Text-based Visual Question Answering~(TextVQA) aims to produce correct answers for given questions about the images with multiple scene texts. In most cases, the texts naturally attach to the surface of the objects. Therefore, spatial reasoning between texts and objects is crucial in TextVQA. However, existing approaches are constrained within 2D spatial information learned from the input images and rely on transformer-based architectures to reason implicitly during the fusion process. Under this setting, these 2D spatial reasoning approaches cannot distinguish the fine-grain spatial relations between visual objects and scene texts on the same image plane, thereby impairing the interpretability and performance of TextVQA models. In this paper, we introduce 3D geometric information into a human-like spatial reasoning process to capture the contextual knowledge of key objects step-by-step. %we formulate a human-like spatial reasoning process by introducing 3D geometric information for capturing key objects' contextual knowledge. To enhance the model's understanding of 3D spatial relationships, Specifically, (i)~we propose a relation prediction module for accurately locating the region of interest of critical objects; (ii)~we design a depth-aware attention calibration module for calibrating the OCR tokens' attention according to critical objects. Extensive experiments show that our method achieves state-of-the-art performance on TextVQA and ST-VQA datasets. More encouragingly, our model surpasses others by clear margins of 5.7\% and 12.1\% on questions that involve spatial reasoning in TextVQA and ST-VQA valid split. Besides, we also verify the generalizability of our model on the text-based image captioning task.

Peng Jin, Jinfa Huang, Pengfei Xiong et al.

Contrastive learning-based video-language representation learning approaches, e.g., CLIP, have achieved outstanding performance, which pursue semantic interaction upon pre-defined video-text pairs. To clarify this coarse-grained global interaction and move a step further, we have to encounter challenging shell-breaking interactions for fine-grained cross-modal learning. In this paper, we creatively model video-text as game players with multivariate cooperative game theory to wisely handle the uncertainty during fine-grained semantic interaction with diverse granularity, flexible combination, and vague intensity. Concretely, we propose Hierarchical Banzhaf Interaction (HBI) to value possible correspondence between video frames and text words for sensitive and explainable cross-modal contrast. To efficiently realize the cooperative game of multiple video frames and multiple text words, the proposed method clusters the original video frames (text words) and computes the Banzhaf Interaction between the merged tokens. By stacking token merge modules, we achieve cooperative games at different semantic levels. Extensive experiments on commonly used text-video retrieval and video-question answering benchmarks with superior performances justify the efficacy of our HBI. More encouragingly, it can also serve as a visualization tool to promote the understanding of cross-modal interaction, which have a far-reaching impact on the community. Project page is available at https://jpthu17.github.io/HBI/.

Bin Lin, Zhenyu Tang, Yang Ye et al.

Recent advances demonstrate that scaling Large Vision-Language Models (LVLMs) effectively improves downstream task performances. However, existing scaling methods enable all model parameters to be active for each token in the calculation, which brings massive training and inferring costs. In this work, we propose a simple yet effective training strategy MoE-Tuning for LVLMs. This strategy innovatively addresses the common issue of performance degradation in multi-modal sparsity learning, consequently constructing a sparse model with an outrageous number of parameters but a constant computational cost. Furthermore, we present the MoE-LLaVA, a MoE-based sparse LVLM architecture, which uniquely activates only the top-k experts through routers during deployment, keeping the remaining experts inactive. Extensive experiments show the significant performance of MoE-LLaVA in a variety of visual understanding and object hallucination benchmarks. Remarkably, with only approximately 3B sparsely activated parameters, MoE-LLaVA demonstrates performance comparable to the LLaVA-1.5-7B on various visual understanding datasets and even surpasses the LLaVA-1.5-13B in object hallucination benchmark. Through MoE-LLaVA, we aim to establish a baseline for sparse LVLMs and provide valuable insights for future research in developing more efficient and effective multi-modal learning systems. Code is released at https://github.com/PKU-YuanGroup/MoE-LLaVA.

Haoran Tang, Meng Cao, Jinfa Huang et al.

Text-Video Retrieval (TVR) aims to align and associate relevant video content with corresponding natural language queries. Most existing TVR methods are based on large-scale pre-trained vision-language models (e.g., CLIP). However, due to the inherent plain structure of CLIP, few TVR methods explore the multi-scale representations which offer richer contextual information for a more thorough understanding. To this end, we propose MUSE, a multi-scale mamba with linear computational complexity for efficient cross-resolution modeling. Specifically, the multi-scale representations are generated by applying a feature pyramid on the last single-scale feature map. Then, we employ the Mamba structure as an efficient multi-scale learner to jointly learn scale-wise representations. Furthermore, we conduct comprehensive studies to investigate different model structures and designs. Extensive results on three popular benchmarks have validated the superiority of MUSE.

Shenghai Yuan, Jinfa Huang, Xianyi He et al.

Identity-preserving text-to-video (IPT2V) generation aims to create high-fidelity videos with consistent human identity. It is an important task in video generation but remains an open problem for generative models. This paper pushes the technical frontier of IPT2V in two directions that have not been resolved in literature: (1) A tuning-free pipeline without tedious case-by-case finetuning, and (2) A frequency-aware heuristic identity-preserving DiT-based control scheme. We propose ConsisID, a tuning-free DiT-based controllable IPT2V model to keep human identity consistent in the generated video. Inspired by prior findings in frequency analysis of diffusion transformers, it employs identity-control signals in the frequency domain, where facial features can be decomposed into low-frequency global features and high-frequency intrinsic features. First, from a low-frequency perspective, we introduce a global facial extractor, which encodes reference images and facial key points into a latent space, generating features enriched with low-frequency information. These features are then integrated into shallow layers of the network to alleviate training challenges associated with DiT. Second, from a high-frequency perspective, we design a local facial extractor to capture high-frequency details and inject them into transformer blocks, enhancing the model's ability to preserve fine-grained features. We propose a hierarchical training strategy to leverage frequency information for identity preservation, transforming a vanilla pre-trained video generation model into an IPT2V model. Extensive experiments demonstrate that our frequency-aware heuristic scheme provides an optimal control solution for DiT-based models. Thanks to this scheme, our ConsisID generates high-quality, identity-preserving videos, making strides towards more effective IPT2V. Code: https://github.com/PKU-YuanGroup/ConsisID.

Jingyi Wang, Can Zhang, Jinfa Huang et al.

Recent advances in Scene Graph Generation (SGG) typically model the relationships among entities utilizing box-level features from pre-defined detectors. We argue that an overlooked problem in SGG is the coarse-grained interactions between boxes, which inadequately capture contextual semantics for relationship modeling, practically limiting the development of the field. In this paper, we take the initiative to explore and propose a generic paradigm termed Superpixel-based Interaction Learning (SIL) to remedy coarse-grained interactions at the box level. It allows us to model fine-grained interactions at the superpixel level in SGG. Specifically, (i) we treat a scene as a set of points and cluster them into superpixels representing sub-regions of the scene. (ii) We explore intra-entity and cross-entity interactions among the superpixels to enrich fine-grained interactions between entities at an earlier stage. Extensive experiments on two challenging benchmarks (Visual Genome and Open Image V6) prove that our SIL enables fine-grained interaction at the superpixel level above previous box-level methods, and significantly outperforms previous state-of-the-art methods across all metrics. More encouragingly, the proposed method can be applied to boost the performance of existing box-level approaches in a plug-and-play fashion. In particular, SIL brings an average improvement of 2.0% mR (even up to 3.4%) of baselines for the PredCls task on Visual Genome, which facilitates its integration into any existing box-level method.

Yongdong Luo, Xiawu Zheng, Xiao Yang et al.

Existing large video-language models (LVLMs) struggle to comprehend long videos correctly due to limited context. To address this problem, fine-tuning long-context LVLMs and employing GPT-based agents have emerged as promising solutions. However, fine-tuning LVLMs would require extensive high-quality data and substantial GPU resources, while GPT-based agents would rely on proprietary models (e.g., GPT-4o). In this paper, we propose Video Retrieval-Augmented Generation (Video-RAG), a training-free and cost-effective pipeline that employs visually-aligned auxiliary texts to help facilitate cross-modality alignment while providing additional information beyond the visual content. Specifically, we leverage open-source external tools to extract visually-aligned information from pure video data (e.g., audio, optical character, and object detection), and incorporate the extracted information into an existing LVLM as auxiliary texts, alongside video frames and queries, in a plug-and-play manner. Our Video-RAG offers several key advantages: (i) lightweight with low computing overhead due to single-turn retrieval; (ii) easy implementation and compatibility with any LVLM; and (iii) significant, consistent performance gains across long video understanding benchmarks, including Video-MME, MLVU, and LongVideoBench. Notably, our model demonstrates superior performance over proprietary models like Gemini-1.5-Pro and GPT-4o when utilized with a 72B model.

Shenghai Yuan, Jinfa Huang, Yujun Shi et al.

Recent advances in Text-to-Video generation (T2V) have achieved remarkable success in synthesizing high-quality general videos from textual descriptions. A largely overlooked problem in T2V is that existing models have not adequately encoded physical knowledge of the real world, thus generated videos tend to have limited motion and poor variations. In this paper, we propose \textbf{MagicTime}, a metamorphic time-lapse video generation model, which learns real-world physics knowledge from time-lapse videos and implements metamorphic generation. First, we design a MagicAdapter scheme to decouple spatial and temporal training, encode more physical knowledge from metamorphic videos, and transform pre-trained T2V models to generate metamorphic videos. Second, we introduce a Dynamic Frames Extraction strategy to adapt to metamorphic time-lapse videos, which have a wider variation range and cover dramatic object metamorphic processes, thus embodying more physical knowledge than general videos. Finally, we introduce a Magic Text-Encoder to improve the understanding of metamorphic video prompts. Furthermore, we create a time-lapse video-text dataset called \textbf{ChronoMagic}, specifically curated to unlock the metamorphic video generation ability. Extensive experiments demonstrate the superiority and effectiveness of MagicTime for generating high-quality and dynamic metamorphic videos, suggesting time-lapse video generation is a promising path toward building metamorphic simulators of the physical world. Code: https://github.com/PKU-YuanGroup/MagicTime

Tianyu Xie, Jinfa Huang, Yuexiao Ma et al.

Omni-modal large language models (OLMs) redefine human-machine interaction by natively integrating audio, vision, and text. However, existing OLM benchmarks remain anchored to static, accuracy-centric tasks, leaving a critical gap in assessing social interactivity, the fundamental capacity to navigate dynamic cues in natural dialogues. To this end, we propose SocialOmni, a comprehensive benchmark that operationalizes the evaluation of this conversational interactivity across three core dimensions: (i) speaker separation and identification (who is speaking), (ii) interruption timing control (when to interject), and (iii) natural interruption generation (how to phrase the interruption). SocialOmni features 2,000 perception samples and a quality-controlled diagnostic set of 209 interaction-generation instances with strict temporal and contextual constraints, complemented by controlled audio-visual inconsistency scenarios to test model robustness. We benchmarked 12 leading OLMs, which uncovers significant variance in their social-interaction capabilities across models. Furthermore, our analysis reveals a pronounced decoupling between a model's perceptual accuracy and its ability to generate contextually appropriate interruptions, indicating that understanding-centric metrics alone are insufficient to characterize conversational social competence. More encouragingly, these diagnostics from SocialOmni yield actionable signals for bridging the perception-interaction divide in future OLMs.

Jing Xiong, Gongye Liu, Lun Huang et al.

Autoregressive modeling has been a huge success in the field of natural language processing (NLP). Recently, autoregressive models have emerged as a significant area of focus in computer vision, where they excel in producing high-quality visual content. Autoregressive models in NLP typically operate on subword tokens. However, the representation strategy in computer vision can vary in different levels, i.e., pixel-level, token-level, or scale-level, reflecting the diverse and hierarchical nature of visual data compared to the sequential structure of language. This survey comprehensively examines the literature on autoregressive models applied to vision. To improve readability for researchers from diverse research backgrounds, we start with preliminary sequence representation and modeling in vision. Next, we divide the fundamental frameworks of visual autoregressive models into three general sub-categories, including pixel-based, token-based, and scale-based models based on the representation strategy. We then explore the interconnections between autoregressive models and other generative models. Furthermore, we present a multifaceted categorization of autoregressive models in computer vision, including image generation, video generation, 3D generation, and multimodal generation. We also elaborate on their applications in diverse domains, including emerging domains such as embodied AI and 3D medical AI, with about 250 related references. Finally, we highlight the current challenges to autoregressive models in vision with suggestions about potential research directions. We have also set up a Github repository to organize the papers included in this survey at: https://github.com/ChaofanTao/Autoregressive-Models-in-Vision-Survey.

Haoyu Huang, Jinfa Huang, Zhongwei Wan et al.

Agentic multimodal large language models (MLLMs) (e.g., OpenAI o3 and Gemini Agentic Vision) achieve remarkable reasoning capabilities through iterative visual tool invocation. However, the cascaded perception, reasoning, and tool-calling loops introduce significant sequential overhead. This overhead, termed agentic depth, incurs prohibitive latency and seriously limits system-level concurrency. To this end, we propose SpecEyes, an agentic-level speculative acceleration framework that breaks this sequential bottleneck. Our key insight is that a lightweight, tool-free MLLM can serve as a speculative planner to predict the execution trajectory, enabling early termination of expensive tool chains without sacrificing accuracy. To regulate this speculative planning, we introduce a cognitive gating mechanism based on answer separability, which quantifies the model's confidence for self-verification without requiring oracle labels. Furthermore, we design a heterogeneous parallel funnel that exploits the stateless concurrency of the small model to mask the stateful serial execution of the large model, maximizing system throughput. Extensive experiments on V* Bench, HR-Bench, and POPE demonstrate that SpecEyes achieves 1.1-3.35x speedup over the agentic baseline while preserving or even improving accuracy (up to +6.7%), thereby boosting serving throughput under concurrent workloads.

Hanjia Lyu, Jinfa Huang, Daoan Zhang et al.

Recent research has offered insights into the extraordinary capabilities of Large Multimodal Models (LMMs) in various general vision and language tasks. There is growing interest in how LMMs perform in more specialized domains. Social media content, inherently multimodal, blends text, images, videos, and sometimes audio. Understanding social multimedia content remains a challenging problem for contemporary machine learning frameworks. In this paper, we explore GPT-4V(ision)'s capabilities for social multimedia analysis. We select five representative tasks, including sentiment analysis, hate speech detection, fake news identification, demographic inference, and political ideology detection, to evaluate GPT-4V. Our investigation begins with a preliminary quantitative analysis for each task using existing benchmark datasets, followed by a careful review of the results and a selection of qualitative samples that illustrate GPT-4V's potential in understanding multimodal social media content. GPT-4V demonstrates remarkable efficacy in these tasks, showcasing strengths such as joint understanding of image-text pairs, contextual and cultural awareness, and extensive commonsense knowledge. Despite the overall impressive capacity of GPT-4V in the social media domain, there remain notable challenges. GPT-4V struggles with tasks involving multilingual social multimedia comprehension and has difficulties in generalizing to the latest trends in social media. Additionally, it exhibits a tendency to generate erroneous information in the context of evolving celebrity and politician knowledge, reflecting the known hallucination problem. The insights gleaned from our findings underscore a promising future for LMMs in enhancing our comprehension of social media content and its users through the analysis of multimodal information.

Rui-Jie Zhu, Tianhao Peng, Tianhao Cheng et al.

Large Language Models (LLMs) have demonstrated impressive reasoning capabilities, especially when guided by explicit chain-of-thought (CoT) reasoning that verbalizes intermediate steps. While CoT improves both interpretability and accuracy, its dependence on natural language reasoning limits the model's expressive bandwidth. Latent reasoning tackles this bottleneck by performing multi-step inference entirely in the model's continuous hidden state, eliminating token-level supervision. To advance latent reasoning research, this survey provides a comprehensive overview of the emerging field of latent reasoning. We begin by examining the foundational role of neural network layers as the computational substrate for reasoning, highlighting how hierarchical representations support complex transformations. Next, we explore diverse latent reasoning methodologies, including activation-based recurrence, hidden state propagation, and fine-tuning strategies that compress or internalize explicit reasoning traces. Finally, we discuss advanced paradigms such as infinite-depth latent reasoning via masked diffusion models, which enable globally consistent and reversible reasoning processes. By unifying these perspectives, we aim to clarify the conceptual landscape of latent reasoning and chart future directions for research at the frontier of LLM cognition. An associated GitHub repository collecting the latest papers and repos is available at: https://github.com/multimodal-art-projection/LatentCoT-Horizon/.

Yongdong Luo, Wang Chen, Xiawu Zheng et al.

Recent advances in long video understanding typically mitigate visual redundancy through visual token pruning based on attention distribution. However, while existing methods employ post-hoc low-response token pruning in decoder layers, they overlook the input-level semantic correlation between visual tokens and instructions (query). In this paper, we propose QuoTA, an ante-hoc training-free modular that extends existing large video-language models (LVLMs) for visual token assignment based on query-oriented frame-level importance assessment. The query-oriented token selection is crucial as it aligns visual processing with task-specific requirements, optimizing token budget utilization while preserving semantically relevant content. Specifically, (i) QuoTA strategically allocates frame-level importance scores based on query relevance, enabling one-time visual token assignment before cross-modal interactions in decoder layers, (ii) we decouple the query through Chain-of-Thoughts reasoning to facilitate more precise LVLM-based frame importance scoring, and (iii) QuoTA offers a plug-and-play functionality that extends to existing LVLMs. Extensive experimental results demonstrate that implementing QuoTA with LLaVA-Video-7B yields an average performance improvement of 3.2% across six benchmarks (including Video-MME and MLVU) while operating within an identical visual token budget as the baseline. Codes are open-sourced at https://github.com/MAC-AutoML/QuoTA.

Yanshu Li, Jianjiang Yang, Tian Yun et al.

Multimodal in-context learning (ICL) has emerged as a key mechanism for harnessing the capabilities of large vision-language models (LVLMs). However, its effectiveness remains highly sensitive to the quality of input ICL sequences, particularly for tasks involving complex reasoning or open-ended generation. A major limitation is our limited understanding of how LVLMs actually exploit these sequences during inference. To bridge this gap, we systematically interpret multimodal ICL through the lens of task mapping, which reveals how local and global relationships within and among demonstrations guide model reasoning. Building on this insight, we present TACO, a lightweight transformer-based model equipped with task-aware attention that dynamically configures ICL sequences. By injecting task-mapping signals into the autoregressive decoding process, TACO creates a bidirectional synergy between sequence construction and task reasoning. Experiments on five LVLMs and nine datasets demonstrate that TACO consistently surpasses baselines across diverse ICL tasks. These results position task mapping as a novel and valuable perspective for interpreting and improving multimodal ICL.

Shaofeng Zhang, Jinfa Huang, Qiang Zhou et al.

Image outpainting aims to generate the content of an input sub-image beyond its original boundaries. It is an important task in content generation yet remains an open problem for generative models. This paper pushes the technical frontier of image outpainting in two directions that have not been resolved in literature: 1) outpainting with arbitrary and continuous multiples (without restriction), and 2) outpainting in a single step (even for large expansion multiples). Moreover, we develop a method that does not depend on a pre-trained backbone network, which is in contrast commonly required by the previous SOTA outpainting methods. The arbitrary multiple outpainting is achieved by utilizing randomly cropped views from the same image during training to capture arbitrary relative positional information. Specifically, by feeding one view and positional embeddings as queries, we can reconstruct another view. At inference, we generate images with arbitrary expansion multiples by inputting an anchor image and its corresponding positional embeddings. The one-step outpainting ability here is particularly noteworthy in contrast to previous methods that need to be performed for $N$ times to obtain a final multiple which is $N$ times of its basic and fixed multiple. We evaluate the proposed approach (called PQDiff as we adopt a diffusion-based generator as our embodiment, under our proposed \textbf{P}ositional \textbf{Q}uery scheme) on public benchmarks, demonstrating its superior performance over state-of-the-art approaches. Specifically, PQDiff achieves state-of-the-art FID scores on the Scenery (\textbf{21.512}), Building Facades (\textbf{25.310}), and WikiArts (\textbf{36.212}) datasets. Furthermore, under the 2.25x, 5x and 11.7x outpainting settings, PQDiff only takes \textbf{40.6\%}, \textbf{20.3\%} and \textbf{10.2\%} of the time of the benchmark state-of-the-art (SOTA) method.

Bin Zhu, Munan Ning, Peng Jin et al.

In the multi-modal domain, the dependence of various models on specific input formats leads to user confusion and hinders progress. To address this challenge, we introduce \textbf{LLMBind}, a novel framework designed to unify a diverse array of multi-modal tasks. By harnessing a Mixture-of-Experts (MoE) Large Language Model (LLM), LLMBind processes multi-modal inputs and generates task-specific tokens, enabling the invocation of corresponding models to accomplish tasks. This unique approach empowers LLMBind to interpret inputs and generate outputs across various modalities, including image, text, video, and audio. Furthermore, we have constructed an interaction dataset comprising 400k instructions, which unlocks the ability of LLMBind for interactive visual generation and editing tasks. Extensive experimentation demonstrates that LLMBind achieves very superior performance across diverse tasks and outperforms existing models in user evaluations conducted in real-world scenarios. Moreover, the adaptability of LLMBind allows for seamless integration with the latest models and extension to new modality tasks, highlighting its potential to serve as a unified AI agent for modeling universal modalities.

Xiaodong Wang, Jinfa Huang, Li Yuan et al.

Most Video Large Language Models (Video-LLMs) adopt preference alignment techniques, e.g., DPO~\citep{rafailov2024dpo}, to optimize the reward margin between a winning response ($y_w$) and a losing response ($y_l$). However, the likelihood displacement observed in DPO indicates that both $\log π_θ(y_w\mid x)$ and $\log π_θ(y_l\mid x) $ often decrease during training, inadvertently boosting the probabilities of non-target responses. In this paper, we systematically revisit this phenomenon from LLMs to Video-LLMs, showing that it intensifies when dealing with the redundant complexity of video content. To alleviate the impact of this phenomenon, we propose \emph{Lean Preference Optimization} (LeanPO), a reference-free approach that reformulates the implicit reward as the average likelihood of the response with respect to the policy model. A key component of LeanPO is the reward-trustworthiness correlated self-generated preference data pipeline, which carefully infuses relevant prior knowledge into the model while continuously refining the preference data via self-reflection. This allows the policy model to obtain high-quality paired data and accurately estimate the newly defined reward, thus mitigating the unintended drop. In addition, we introduce a dynamic label smoothing strategy that mitigates the impact of noise in responses from diverse video content, preventing the model from overfitting to spurious details. Extensive experiments demonstrate that LeanPO significantly enhances the performance of state-of-the-art Video-LLMs, consistently boosting baselines of varying capacities with minimal additional training overhead. Moreover, LeanPO offers a simple yet effective solution for aligning Video-LLM preferences with human trustworthiness, paving the way toward the reliable and efficient Video-LLMs.

Shenghai Yuan, Jinfa Huang, Yongqi Xu et al.

We propose a novel text-to-video (T2V) generation benchmark, ChronoMagic-Bench, to evaluate the temporal and metamorphic capabilities of the T2V models (e.g. Sora and Lumiere) in time-lapse video generation. In contrast to existing benchmarks that focus on visual quality and textual relevance of generated videos, ChronoMagic-Bench focuses on the model's ability to generate time-lapse videos with significant metamorphic amplitude and temporal coherence. The benchmark probes T2V models for their physics, biology, and chemistry capabilities, in a free-form text query. For these purposes, ChronoMagic-Bench introduces 1,649 prompts and real-world videos as references, categorized into four major types of time-lapse videos: biological, human-created, meteorological, and physical phenomena, which are further divided into 75 subcategories. This categorization comprehensively evaluates the model's capacity to handle diverse and complex transformations. To accurately align human preference with the benchmark, we introduce two new automatic metrics, MTScore and CHScore, to evaluate the videos' metamorphic attributes and temporal coherence. MTScore measures the metamorphic amplitude, reflecting the degree of change over time, while CHScore assesses the temporal coherence, ensuring the generated videos maintain logical progression and continuity. Based on ChronoMagic-Bench, we conduct comprehensive manual evaluations of ten representative T2V models, revealing their strengths and weaknesses across different categories of prompts, and providing a thorough evaluation framework that addresses current gaps in video generation research. Moreover, we create a large-scale ChronoMagic-Pro dataset, containing 460k high-quality pairs of 720p time-lapse videos and detailed captions ensuring high physical pertinence and large metamorphic amplitude. [Homepage](https://pku-yuangroup.github.io/ChronoMagic-Bench/).

Zhongwei Wan, Ziang Wu, Che Liu et al.

Long-context Multimodal Large Language Models (MLLMs) demand substantial computational resources for inference as the growth of their multimodal Key-Value (KV) cache, in response to increasing input lengths, challenges memory and time efficiency. Unlike single-modality LLMs that manage only textual contexts, the KV cache of long-context MLLMs includes representations from multiple images with temporal and spatial relationships and related textual contexts. The predominance of image tokens means traditional optimizations for LLMs' KV caches are unsuitable for multimodal long-context settings, and no prior works have addressed this challenge. In this work, we introduce LOOK-M, a pioneering, fine-tuning-free approach that efficiently reduces the multimodal KV cache size while maintaining performance comparable to a full cache. We observe that during prompt prefill, the model prioritizes more textual attention over image features, and based on the multimodal interaction observation, a new proposed text-prior method is explored to compress the KV cache. Furthermore, to mitigate the degradation of image contextual information, we propose several compensatory strategies using KV pairs merging. LOOK-M demonstrates that with a significant reduction in KV Cache memory usage, such as reducing it by 80% in some cases, it not only achieves up to 1.5x faster decoding but also maintains or even enhances performance across a variety of long context multimodal tasks.

Peng Jin, Hao Li, Zesen Cheng et al.

Text-video retrieval is a challenging cross-modal task, which aims to align visual entities with natural language descriptions. Current methods either fail to leverage the local details or are computationally expensive. What's worse, they fail to leverage the heterogeneous concepts in data. In this paper, we propose the Disentangled Conceptualization and Set-to-set Alignment (DiCoSA) to simulate the conceptualizing and reasoning process of human beings. For disentangled conceptualization, we divide the coarse feature into multiple latent factors related to semantic concepts. For set-to-set alignment, where a set of visual concepts correspond to a set of textual concepts, we propose an adaptive pooling method to aggregate semantic concepts to address the partial matching. In particular, since we encode concepts independently in only a few dimensions, DiCoSA is superior at efficiency and granularity, ensuring fine-grained interactions using a similar computational complexity as coarse-grained alignment. Extensive experiments on five datasets, including MSR-VTT, LSMDC, MSVD, ActivityNet, and DiDeMo, demonstrate that our method outperforms the existing state-of-the-art methods.

Jingyi Wang, Jinfa Huang, Can Zhang et al.

Dynamic scene graphs generated from video clips could help enhance the semantic visual understanding in a wide range of challenging tasks such as environmental perception, autonomous navigation, and task planning of self-driving vehicles and mobile robots. In the process of temporal and spatial modeling during dynamic scene graph generation, it is particularly intractable to learn time-variant relations in dynamic scene graphs among frames. In this paper, we propose a Time-variant Relation-aware TRansformer (TR$^2$), which aims to model the temporal change of relations in dynamic scene graphs. Explicitly, we leverage the difference of text embeddings of prompted sentences about relation labels as the supervision signal for relations. In this way, cross-modality feature guidance is realized for the learning of time-variant relations. Implicitly, we design a relation feature fusion module with a transformer and an additional message token that describes the difference between adjacent frames. Extensive experiments on the Action Genome dataset prove that our TR$^2$ can effectively model the time-variant relations. TR$^2$ significantly outperforms previous state-of-the-art methods under two different settings by 2.1% and 2.6% respectively.