Alex Jinpeng Wang

Kevin Qinghong Lin, Alex Jinpeng Wang, Mattia Soldan et al. · microsoft-research, uw

Video-Language Pretraining (VLP), which aims to learn transferable representation to advance a wide range of video-text downstream tasks, has recently received increasing attention. Best performing works rely on large-scale, 3rd-person video-text datasets, such as HowTo100M. In this work, we exploit the recently released Ego4D dataset to pioneer Egocentric VLP along three directions. (i) We create EgoClip, a 1st-person video-text pretraining dataset comprising 3.8M clip-text pairs well-chosen from Ego4D, covering a large variety of human daily activities. (ii) We propose a novel pretraining objective, dubbed EgoNCE, which adapts video-text contrastive learning to the egocentric domain by mining egocentric-aware positive and negative samples. (iii) We introduce EgoMCQ, a development benchmark that is close to EgoClip and hence can support effective validation and fast exploration of our design decisions in EgoClip and EgoNCE. Furthermore, we demonstrate strong performance on five egocentric downstream tasks across three datasets: video-text retrieval on EPIC-KITCHENS-100; action recognition on Charades-Ego; natural language query, moment query, and object state change classification on Ego4D challenge benchmarks. The dataset and code are available at https://github.com/showlab/EgoVLP.

Kevin Qinghong Lin, Pengchuan Zhang, Joya Chen et al. · microsoft-research, uw

Video Temporal Grounding (VTG), which aims to ground target clips from videos (such as consecutive intervals or disjoint shots) according to custom language queries (e.g., sentences or words), is key for video browsing on social media. Most methods in this direction develop taskspecific models that are trained with type-specific labels, such as moment retrieval (time interval) and highlight detection (worthiness curve), which limits their abilities to generalize to various VTG tasks and labels. In this paper, we propose to Unify the diverse VTG labels and tasks, dubbed UniVTG, along three directions: Firstly, we revisit a wide range of VTG labels and tasks and define a unified formulation. Based on this, we develop data annotation schemes to create scalable pseudo supervision. Secondly, we develop an effective and flexible grounding model capable of addressing each task and making full use of each label. Lastly, thanks to the unified framework, we are able to unlock temporal grounding pretraining from large-scale diverse labels and develop stronger grounding abilities e.g., zero-shot grounding. Extensive experiments on three tasks (moment retrieval, highlight detection and video summarization) across seven datasets (QVHighlights, Charades-STA, TACoS, Ego4D, YouTube Highlights, TVSum, and QFVS) demonstrate the effectiveness and flexibility of our proposed framework. The codes are available at https://github.com/showlab/UniVTG.

Alex Jinpeng Wang, Yixiao Ge, Rui Yan et al. · tencent-ai

Mainstream Video-Language Pre-training models \cite{actbert,clipbert,violet} consist of three parts, a video encoder, a text encoder, and a video-text fusion Transformer. They pursue better performance via utilizing heavier unimodal encoders or multimodal fusion Transformers, resulting in increased parameters with lower efficiency in downstream tasks. In this work, we for the first time introduce an end-to-end video-language model, namely \textit{all-in-one Transformer}, that embeds raw video and textual signals into joint representations using a unified backbone architecture. We argue that the unique temporal information of video data turns out to be a key barrier hindering the design of a modality-agnostic Transformer. To overcome the challenge, we introduce a novel and effective token rolling operation to encode temporal representations from video clips in a non-parametric manner. The careful design enables the representation learning of both video-text multimodal inputs and unimodal inputs using a unified backbone model. Our pre-trained all-in-one Transformer is transferred to various downstream video-text tasks after fine-tuning, including text-video retrieval, video-question answering, multiple choice and visual commonsense reasoning. State-of-the-art performances with the minimal model FLOPs on nine datasets demonstrate the superiority of our method compared to the competitive counterparts. The code and pretrained model have been released in https://github.com/showlab/all-in-one.

Kevin Qinghong Lin, Alex Jinpeng Wang, Mattia Soldan et al. · microsoft-research, uw

In this report, we propose a video-language pretraining (VLP) based solution \cite{kevin2022egovlp} for four Ego4D challenge tasks, including Natural Language Query (NLQ), Moment Query (MQ), Object State Change Classification (OSCC), and PNR Localization (PNR). Especially, we exploit the recently released Ego4D dataset \cite{grauman2021ego4d} to pioneer Egocentric VLP from pretraining dataset, pretraining objective, and development set. Based on the above three designs, we develop a pretrained video-language model that is able to transfer its egocentric video-text representation or video-only representation to several video downstream tasks. Our Egocentric VLP achieves 10.46R@1&IoU @0.3 on NLQ, 10.33 mAP on MQ, 74% Acc on OSCC, 0.67 sec error on PNR. The code is available at https://github.com/showlab/EgoVLP.

Kevin Qinghong Lin, Alex Jinpeng Wang, Rui Yan et al. · microsoft-research, uw

In this report, we propose a video-language pretraining (VLP) based solution \cite{kevin2022egovlp} for the EPIC-KITCHENS-100 Multi-Instance Retrieval (MIR) challenge. Especially, we exploit the recently released Ego4D dataset \cite{grauman2021ego4d} to pioneer Egocentric VLP from pretraining dataset, pretraining objective, and development set. Based on the above three designs, we develop a pretrained video-language model that is able to transfer its egocentric video-text representation to MIR benchmark. Furthermore, we devise an adaptive multi-instance max-margin loss to effectively fine-tune the model and equip the dual-softmax technique for reliable inference. Our best single model obtains strong performance on the challenge test set with 47.39% mAP and 61.44% nDCG. The code is available at https://github.com/showlab/EgoVLP.

Guanyu Cai, Yixiao Ge, Binjie Zhang et al. · tencent-ai

Recent dominant methods for video-language pre-training (VLP) learn transferable representations from the raw pixels in an end-to-end manner to achieve advanced performance on downstream video-language retrieval. Despite the impressive results, VLP research becomes extremely expensive with the need for massive data and a long training time, preventing further explorations. In this work, we revitalize region features of sparsely sampled video clips to significantly reduce both spatial and temporal visual redundancy towards democratizing VLP research at the same time achieving state-of-the-art results. Specifically, to fully explore the potential of region features, we introduce a novel bidirectional region-word alignment regularization that properly optimizes the fine-grained relations between regions and certain words in sentences, eliminating the domain/modality disconnections between pre-extracted region features and text. Extensive results of downstream video-language retrieval tasks on four datasets demonstrate the superiority of our method on both effectiveness and efficiency, \textit{e.g.}, our method achieves competing results with 80\% fewer data and 85\% less pre-training time compared to the most efficient VLP method so far \cite{lei2021less}. The code will be available at \url{https://github.com/showlab/DemoVLP}.

Alex Jinpeng Wang, Pan Zhou, Mike Zheng Shou et al.

Vision-Language Pre-Training (VLP) has shown promising capabilities to align image and text pairs, facilitating a broad variety of cross-modal learning tasks. However, we observe that VLP models often lack the visual grounding/localization capability which is critical for many downstream tasks such as visual reasoning. In this work, we propose a novel Position-guided Text Prompt (PTP) paradigm to enhance the visual grounding ability of cross-modal models trained with VLP. Specifically, in the VLP phase, PTP divides the image into $N\times N$ blocks, and identifies the objects in each block through the widely used object detector in VLP. It then reformulates the visual grounding task into a fill-in-the-blank problem given a PTP by encouraging the model to predict the objects in the given blocks or regress the blocks of a given object, e.g. filling `P" or ``O" in aPTP ``The block P has a O". This mechanism improves the visual grounding capability of VLP models and thus helps them better handle various downstream tasks. By introducing PTP into several state-of-the-art VLP frameworks, we observe consistently significant improvements across representative cross-modal learning model architectures and several benchmarks, e.g. zero-shot Flickr30K Retrieval (+4.8 in average recall@1) for ViLT \cite{vilt} baseline, and COCO Captioning (+5.3 in CIDEr) for SOTA BLIP \cite{blip} baseline. Moreover, PTP achieves comparable results with object-detector based methods, and much faster inference speed since PTP discards its object detector for inference while the later cannot. Our code and pre-trained weight will be released at \url{https://github.com/sail-sg/ptp}.

Yuying Ge, Yixiao Ge, Xihui Liu et al. · tencent-ai

Dominant pre-training work for video-text retrieval mainly adopt the "dual-encoder" architectures to enable efficient retrieval, where two separate encoders are used to contrast global video and text representations, but ignore detailed local semantics. The recent success of image BERT pre-training with masked visual modeling that promotes the learning of local visual context, motivates a possible solution to address the above limitation. In this work, we for the first time investigate masked visual modeling in video-text pre-training with the "dual-encoder" architecture. We perform Masked visual modeling with Injected LanguagE Semantics (MILES) by employing an extra snapshot video encoder as an evolving "tokenizer" to produce reconstruction targets for masked video patch prediction. Given the corrupted video, the video encoder is trained to recover text-aligned features of the masked patches via reasoning with the visible regions along the spatial and temporal dimensions, which enhances the discriminativeness of local visual features and the fine-grained cross-modality alignment. Our method outperforms state-of-the-art methods for text-to-video retrieval on four datasets with both zero-shot and fine-tune evaluation protocols. Our approach also surpasses the baseline models significantly on zero-shot action recognition, which can be cast as video-to-text retrieval.

Xiangxi Zheng, Kuang He, Jiayi Hu et al.

Chart-to-code generation demands strict visual precision and syntactic correctness from Vision-Language Models (VLMs). However, existing approaches are fundamentally constrained by data-centric limitations: despite the availability of growing chart-to-code datasets, simply scaling homogeneous chart-code pairs conflates visual perception with program logic, preventing models from fully leveraging the richness of multimodal supervision. We present CharTide, a novel data-centric framework that systematically redesigns both training and alignment data for chart-to-code generation. First, we construct a 2M-sample dataset via a Tri-Perspective Tuning strategy, explicitly decoupling training into visual perception, pure-text code logic, and modality fusion streams, enabling a 7B model to surpass specialized baselines using only supervised data. Second, we reformulate alignment as a data verification problem rather than a heuristic scoring task. To this end, we introduce an Inquiry-Driven RL framework grounded in the principle of information invariance: a downstream model should yield consistent answers to identical visual queries across both original and generated charts. Moving beyond rigid rule matching or VLM scoring, we employ a frozen Inspector to objectively verify generated charts through atomic QA tasks, providing verifiable reward signals based on answer accuracy. Experiments on ChartMimic, Plot2Code, and ChartX show that CharTide-7B/8B significantly outperforms open-source baselines, surpasses GPT-4o, and is competitive with GPT-5.

Kevin Qinghong Lin, Yuhao Zheng, Hangyu Ran et al.

Code has emerged as a precise and executable medium for reasoning and action in the agent era. Yet, progress has largely focused on language-centric tasks such as program synthesis and debugging, leaving visual-centric coding underexplored. Inspired by how humans reason over sketches, we advocate SVG code as a compact, interpretable, and executable visual representation. We introduce VCode, a benchmark that reframes multimodal understanding as code generation: given an image, a model must produce SVG that preserves symbolic meaning for downstream reasoning. VCode covers three domains - general commonsense (MM-Vet), professional disciplines (MMMU), and visual-centric perception (CV-Bench). To assess symbolic fidelity, we propose CodeVQA, a novel evaluation protocol in which a policy model answers questions over rendered SVGs; correct answers indicate faithful symbolic preservation. Empirically, frontier VLMs struggle to generate faithful SVGs, revealing a persistent gap between language-centric and visual-centric coding. To close this gap, we introduce VCoder, an agentic framework that augments VLMs along two axes: (i) Thinking with Revision, which iteratively analyzes discrepancies and refines SVG code; and (ii) Acting with Visual Tools, where detectors and parsers supply structured cues such as objects, shapes, and text beyond the model's intrinsic capacity. Across benchmarks, frontier VLMs with strong reasoning capabilities score well overall yet remain limited in professional knowledge and 3D reasoning. VCoder delivers a 12.3-point overall gain over the top-performing Claude-4-Opus. Human studies show that both humans and VLMs perform worse on rendered SVGs, their consistency reveals the promise of symbolic visual representation. The benchmark and code are available at https://github.com/CSU-JPG/VCode.

Zhuobai Dong, Rui Zhao, Songjie Wu et al.

Diffusion models have achieved remarkable success in image generation, yet their deployment remains constrained by the heavy computational cost and the need for numerous inference steps. Previous efforts on fewer-step distillation attempt to skip redundant steps by training compact student models, yet they often suffer from heavy retraining costs and degraded generalization. In this work, we take a different perspective: we accelerate smartly, not evenly, applying smaller speedups to early semantic stages and larger ones to later redundant phases. We instantiate this phase-aware strategy with two experts that specialize in slow and fast denoising phases. Surprisingly, instead of investing massive effort in retraining student models, we find that simply equipping the base model with lightweight LoRA adapters achieves both efficient acceleration and strong generalization. We refer to these two adapters as Slow-LoRA and Fast-LoRA. Through extensive experiments, our method achieves up to 5 acceleration over the base model while maintaining comparable visual quality across diverse benchmarks. Remarkably, the LoRA experts are trained with only 1 samples on a single V100 within one hour, yet the resulting models generalize strongly on unseen prompts.

Junchao Yi, Rui Zhao, Jiahao Tang et al.

Multimodal generation has long been dominated by text-driven pipelines where language dictates vision but cannot reason or create within it. We challenge this paradigm by asking whether all modalities, including textual descriptions, spatial layouts, and editing instructions, can be unified into a single visual representation. We present FlowInOne, a framework that reformulates multimodal generation as a purely visual flow, converting all inputs into visual prompts and enabling a clean image-in, image-out pipeline governed by a single flow matching model. This vision-centric formulation naturally eliminates cross-modal alignment bottlenecks, noise scheduling, and task-specific architectural branches, unifying text-to-image generation, layout-guided editing, and visual instruction following under one coherent paradigm. To support this, we introduce VisPrompt-5M, a large-scale dataset of 5 million visual prompt pairs spanning diverse tasks including physics-aware force dynamics and trajectory prediction, alongside VP-Bench, a rigorously curated benchmark assessing instruction faithfulness, spatial precision, visual realism, and content consistency. Extensive experiments demonstrate that FlowInOne achieves state-of-the-art performance across all unified generation tasks, surpassing both open-source models and competitive commercial systems, establishing a new foundation for fully vision-centric generative modeling where perception and creation coexist within a single continuous visual space.

Dongxing Mao, Yilin Wang, Linjie Li et al.

Despite recent advances in text-to-image generation, models still struggle to accurately render prompt-specified text with correct spatial layout -- especially in multi-span, structured settings. This challenge is driven not only by the lack of datasets that align prompts with the exact text and layout expected in the image, but also by the absence of effective metrics for evaluating layout quality. To address these issues, we introduce TextGround4M, a large-scale dataset of over 4 million prompt-image pairs, each annotated with span-level text grounded in the prompt and corresponding bounding boxes. This enables fine-grained supervision for layout-aware, prompt-grounded text rendering. Building on this, we propose a lightweight training strategy for autoregressive T2I models that appends layout-aware span tokens during training, without altering model architecture or inference behavior. We further construct a benchmark with stratified layout complexity to evaluate both open-source and proprietary models in a zero-shot setting. In addition, we introduce two layout-aware metrics to address the long-standing lack of spatial evaluation in text rendering. Our results show that models trained on TextGround4M outperform strong baselines in text fidelity, spatial accuracy, and prompt consistency, highlighting the importance of fine-grained layout supervision for grounded T2I generation.

Alex Jinpeng Wang, Dongxing Mao, Jiawei Zhang et al. · microsoft-research

Text-conditioned image generation has gained significant attention in recent years and are processing increasingly longer and comprehensive text prompt. In everyday life, dense and intricate text appears in contexts like advertisements, infographics, and signage, where the integration of both text and visuals is essential for conveying complex information. However, despite these advances, the generation of images containing long-form text remains a persistent challenge, largely due to the limitations of existing datasets, which often focus on shorter and simpler text. To address this gap, we introduce TextAtlas5M, a novel dataset specifically designed to evaluate long-text rendering in text-conditioned image generation. Our dataset consists of 5 million long-text generated and collected images across diverse data types, enabling comprehensive evaluation of large-scale generative models on long-text image generation. We further curate 3000 human-improved test set TextAtlasEval across 3 data domains, establishing one of the most extensive benchmarks for text-conditioned generation. Evaluations suggest that the TextAtlasEval benchmarks present significant challenges even for the most advanced proprietary models (e.g. GPT4o with DallE-3), while their open-source counterparts show an even larger performance gap. These evidences position TextAtlas5M as a valuable dataset for training and evaluating future-generation text-conditioned image generation models.

Xiangxi Zheng, Linjie Li, Zhengyuan Yang et al. · microsoft-research

Recent advancements in Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities in visual-text processing. However, existing static image-text benchmarks are insufficient for evaluating their dynamic perception and interactive reasoning abilities. We introduce Vision-centric Multiple Abilities Game Evaluation(V-MAGE), a novel game-based evaluation framework designed to systematically assess MLLMs' visual reasoning in interactive, continuous-space environments. V-MAGE features five distinct video games comprising over 30 carefully constructed evaluation scenarios. These scenarios are set in free-form, visually complex environments that require models to interpret dynamic game states and make decisions based solely on visual input, thereby closely reflecting the conditions encountered by human players. To ensure robust and interpretable comparisons across models, V-MAGE employs a dynamic Elo-based ranking system that accounts for varying difficulty levels and task diversity. Benchmarking state-of-the-art MLLMs against human baselines reveals that while leading models approach human-level performance in simple tasks, their performance drops significantly in complex scenarios requiring advanced reasoning and task orchestration. This persistent performance gap highlights fundamental limitations in current MLLMs' ability to perform real-time, vision-grounded interactions. Through extensive analyses, we demonstrate the utility of V-MAGE in uncovering these limitations and providing actionable insights for improving the visual and reasoning capabilities of MLLMs in dynamic, interactive settings. Code is publicly available at https://github.com/CSU-JPG/V-MAGE.

Haochen Han, Alex Jinpeng Wang, Fangming Liu et al.

In this paper, we study a practical but less-touched problem in Vision-Language Models (VLMs), \ie, negation understanding. Specifically, many real-world applications require models to explicitly identify what is false or non-existent, \eg, radiologists may search for images that exclude specific conditions. Despite the impressive transferability of VLMs through large-scale training, they suffer from a critical limitation that fails to handle negation. To address this challenge, existing methods attribute its root cause to the scarcity of negation training data and propose to fine-tune VLMs on massive data containing explicit negation. Undoubtedly, such data-centric solutions demand substantial data and computational resources, limiting their sustainable widespread adoption. To tackle negation in a low-carbon manner, we empirically observe that the key obstacle lies in the dual-concept shifts between the affirmation and negation distributions. Therefore, we propose a Negation-Aware Test-Time Adaptation (NEAT) method to efficiently adjust distribution-related parameters during inference. In brief, NEAT can reduce distribution shift in consistent semantics while eliminating false distributional consistency in unrelated semantics. Extensive experiments on the various negation understanding tasks verify the effectiveness of the proposed method. Remarkably, with less than 0.01\% of trainable parameters, NEAT achieves comparable or superior performance to state-of-the-art post-training approaches. Our code is available at https://github.com/hhc1997/NEAT.

Yin Tang, Jiawei Ma, Jinrui Zhang et al.

Continuous navigation in complex environments is critical for Unmanned Aerial Vehicle (UAV). However, the existing Vision-Language Navigation (VLN) models follow the dead-reckoning, which iteratively updates its position for the next waypoint prediction, and subsequently construct the complete trajectory. Then, such stepwise manner will inevitably lead to accumulated errors of position over time, resulting in misalignment between internal belief and objective coordinates, which is known as "state drift" and ultimately compromises the full trajectory prediction. Drawing inspiration from classical control theory, we propose to correct for errors by formulating such sequential prediction as a recursive Bayesian state estimation problem. In this paper, we design NeuroKalman, a novel framework that decouples navigation into two complementary processes: a Prior Prediction, based on motion dynamics and a Likelihood Correction, from historical observation. We first mathematically associate Kernel Density Estimation of the measurement likelihood with the attention-based retrieval mechanism, which then allows the system to rectify the latent representation using retrieved historical anchors without gradient updates. Comprehensive experiments on TravelUAV benchmark demonstrate that, with only 10% of the training data fine-tuning, our method clearly outperforms strong baselines and regulates drift accumulation.

Yiqi Lin, Alex Jinpeng Wang, Linjie Li et al.

Contrastive vision-language models such as CLIP have demonstrated strong performance across a wide range of multimodal tasks by learning from aligned image-text pairs. However, their ability to handle complex, real-world web documents remains limited, particularly in scenarios where text and images are interleaved, loosely aligned, or embedded in visual form. To address these challenges, we propose Vision-Centric Contrastive Learning (VC2L), a unified framework that models text, images, and their combinations using a single vision transformer. VC2L operates entirely in pixel space by rendering all inputs, whether textual, visual, or combined, as images, thus eliminating the need for OCR, text tokenization, or modality fusion strategy. To capture complex cross-modal relationships in multimodal web documents, VC2L employs a snippet-level contrastive learning objective that aligns consecutive multimodal segments, leveraging the inherent coherence of documents without requiring explicitly paired image-text data. To assess the effectiveness of this approach, we introduce three retrieval benchmarks, AnyCIR, SeqCIR, and CSR, designed to evaluate cross-modal retrieval, fine-grained sequential understanding, and generalization to unseen data, respectively. Empirical results show that VC2L achieves competitive or superior performance compared to CLIP-style models on both the proposed benchmarks and established datasets such as M-BEIR and MTEB. These findings underscore the potential of multimodal web data as a valuable training resource for contrastive learning and illustrate the scalability of a unified, vision-centric approach for multimodal representation learning. Code and models are available at: https://github.com/showlab/VC2L.

Jiahao Tang, Henry Hengyuan Zhao, Lijian Wu et al.

We introduce Chart2Code, a new benchmark for evaluating the chart understanding and code generation capabilities of large multimodal models (LMMs). Chart2Code is explicitly designed from a user-driven perspective, capturing diverse real-world scenarios and progressively increasing task difficulty. It consists of three levels: Level 1 (Chart Reproduction) reproduces charts from a reference figure and user query; Level 2 (Chart Editing) involves complex modifications such as changing chart types or adding elements; and Level 3 (Long-Table to Chart Generation) requires models to transform long, information-dense tables into faithful charts following user instructions. To our knowledge, this is the first hierarchical benchmark that reflects practical chart2code usage while systematically scaling task complexity. In total, Chart2Code contains 2,023 tasks across 22 chart types, paired with multi-level evaluation metrics that assess both code correctness and the visual fidelity of rendered charts. We benchmark 25 state-of-the-art (SoTA) LMMs, including both proprietary and the latest open-source models such as GPT-5, Qwen2.5-VL, InternVL3/3.5, MiMo-VL, and Seed-1.6-VL. Experimental results demonstrate that even the SoTA model GPT-5 averages only 0.57 on code-based evaluation and 0.22 on chart-quality assessment across the editing tasks, underscoring the difficulty of Chart2Code. We anticipate this benchmark will drive advances in multimodal reasoning and foster the development of more robust and general-purpose LMMs. Our code and data are available on Chart2Code.

Alex Jinpeng Wang, Kevin Qinghong Lin, David Junhao Zhang et al.

This paper examines the problems of severe image-text misalignment and high redundancy in the widely-used large-scale Vision-Language Pre-Training (VLP) datasets. To address these issues, we propose an efficient and straightforward Vision-Language learning algorithm called TL;DR, which aims to compress the existing large VLP data into a small, high-quality set. Our approach consists of two major steps. First, a codebook-based encoder-decoder captioner is developed to select representative samples. Second, a new caption is generated to complement the original captions for selected samples, mitigating the text-image misalignment problem while maintaining uniqueness. As the result, TL;DR enables us to reduce the large dataset into a small set of high-quality data, which can serve as an alternative pre-training dataset. This algorithm significantly speeds up the time-consuming pretraining process. Specifically, TL;DR can compress the mainstream VLP datasets at a high ratio, e.g., reduce well-cleaned CC3M dataset from 2.82M to 0.67M ($\sim$24\%) and noisy YFCC15M from 15M to 2.5M ($\sim$16.7\%). Extensive experiments with three popular VLP models over seven downstream tasks show that VLP model trained on the compressed dataset provided by TL;DR can perform similar or even better results compared with training on the full-scale dataset. The code will be made available at \url{https://github.com/showlab/datacentric.vlp}.

Rui Yan, Mike Zheng Shou, Yixiao Ge et al.

Video-Text pre-training aims at learning transferable representations from large-scale video-text pairs via aligning the semantics between visual and textual information. State-of-the-art approaches extract visual features from raw pixels in an end-to-end fashion. However, these methods operate at frame-level directly and thus overlook the spatio-temporal structure of objects in video, which yet has a strong synergy with nouns in textual descriptions. In this work, we propose a simple yet effective module for video-text representation learning, namely RegionLearner, which can take into account the structure of objects during pre-training on large-scale video-text pairs. Given a video, our module (1) first quantizes visual features into semantic clusters, then (2) generates learnable masks and uses them to aggregate the features belonging to the same semantic region, and finally (3) models the interactions between different aggregated regions. In contrast to using off-the-shelf object detectors, our proposed module does not require explicit supervision and is much more computationally efficient. We pre-train the proposed approach on the public WebVid2M and CC3M datasets. Extensive evaluations on four downstream video-text retrieval benchmarks clearly demonstrate the effectiveness of our RegionLearner. The code will be available at https://github.com/ruiyan1995/Region_Learner.

Alex Jinpeng Wang, Yixiao Ge, Guanyu Cai et al.

Recently, by introducing large-scale dataset and strong transformer network, video-language pre-training has shown great success especially for retrieval. Yet, existing video-language transformer models do not explicitly fine-grained semantic align. In this work, we present Object-aware Transformers, an object-centric approach that extends video-language transformer to incorporate object representations. The key idea is to leverage the bounding boxes and object tags to guide the training process. We evaluate our model on three standard sub-tasks of video-text matching on four widely used benchmarks. We also provide deep analysis and detailed ablation about the proposed method. We show clear improvement in performance across all tasks and datasets considered, demonstrating the value of a model that incorporates object representations into a video-language architecture. The code will be released at \url{https://github.com/FingerRec/OA-Transformer}.

Rui Gui, Yang Wan, Haochen Han et al.

Text rendering has recently emerged as one of the most challenging frontiers in visual generation, drawing significant attention from large-scale diffusion and multimodal models. However, text editing within images remains largely unexplored, as it requires generating legible characters while preserving semantic, geometric, and contextual coherence. To fill this gap, we introduce TextEditBench, a comprehensive evaluation benchmark that explicitly focuses on text-centric regions in images. Beyond basic pixel manipulations, our benchmark emphasizes reasoning-intensive editing scenarios that require models to understand physical plausibility, linguistic meaning, and cross-modal dependencies. We further propose a novel evaluation dimension, Semantic Expectation (SE), which measures reasoning ability of model to maintain semantic consistency, contextual coherence, and cross-modal alignment during text editing. Extensive experiments on state-of-the-art editing systems reveal that while current models can follow simple textual instructions, they still struggle with context-dependent reasoning, physical consistency, and layout-aware integration. By focusing evaluation on this long-overlooked yet fundamental capability, TextEditBench establishes a new testing ground for advancing text-guided image editing and reasoning in multimodal generation.

Alex Jinpeng Wang, Linjie Li, Kevin Qinghong Lin et al. · microsoft-research

In the evolution of Vision-Language Pre-training, shifting from short-text comprehension to encompassing extended textual contexts is pivotal. Recent autoregressive vision-language models like \cite{flamingo, palme}, leveraging the long-context capability of Large Language Models, have excelled in few-shot text generation tasks but face challenges in alignment tasks. Addressing this gap, we introduce the contrastive loss into text generation models, presenting the COntrastive-Streamlined MultimOdal framework (\ModelName), strategically partitioning the language model into dedicated unimodal text processing and adept multimodal data handling components. \ModelName, our unified framework, merges unimodal and multimodal elements, enhancing model performance for tasks involving textual and visual data while notably reducing learnable parameters. However, these models demand extensive long-text datasets, yet the availability of high-quality long-text video datasets remains limited. To bridge this gap, this work introduces \VideoDatasetName, an inaugural interleaved video-text dataset featuring comprehensive captions, marking a significant step forward. Demonstrating its impact, we illustrate how \VideoDatasetName{} enhances model performance in image-text tasks. With 34% learnable parameters and utilizing 72\% of the available data, our model demonstrates significant superiority over OpenFlamingo~\cite{openflamingo}. For instance, in the 4-shot flickr captioning task, performance notably improves from 57.2% to 65.\%. The contributions of \ModelName{} and \VideoDatasetName{} are underscored by notable performance gains across 14 diverse downstream datasets encompassing both image-text and video-text tasks.

Yiqi Lin, Conghui He, Alex Jinpeng Wang et al.

Despite CLIP being the foundation model in numerous vision-language applications, the CLIP suffers from a severe text spotting bias. Such bias causes CLIP models to `Parrot' the visual text embedded within images while disregarding the authentic visual semantics. We uncover that in the most popular image-text dataset LAION-2B, the captions also densely parrot (spell) the text embedded in images. Our analysis shows that around 50% of images are embedded with visual text content, and around 30% of captions words are in these embedded visual content. Based on such observation, we thoroughly inspect the different released versions of CLIP models and verify that the visual text is the dominant factor in measuring the LAION-style image-text similarity for these models. To examine whether these parrot captions shape the text spotting bias, we train a series of CLIP models with LAION subsets curated by different parrot-caption-oriented criteria. We show that training with parrot captions easily shapes such bias but harms the expected visual-language representation learning in CLIP models. This suggests that it is urgent to revisit either the design of CLIP-like models or the existing image-text dataset curation pipeline built on CLIP score filtering.

Zihui Cheng, Qiguang Chen, Xiao Xu et al.

Large Vision-Language Models (LVLMs) have achieved significant success in multimodal tasks, with multimodal chain-of-thought (MCoT) further enhancing performance and interpretability. Recent MCoT methods fall into two categories: (i) Textual-MCoT (T-MCoT), which takes multimodal input and produces textual output; and (ii) Interleaved-MCoT (I-MCoT), which generates interleaved image-text outputs. Despite advances in both approaches, the mechanisms driving these improvements are not fully understood. To fill this gap, we first reveal that MCoT boosts LVLMs by incorporating visual thoughts, which convey image information to the reasoning process regardless of the MCoT format, depending only on clarity and conciseness of expression. Furthermore, to explore visual thoughts systematically, we define four distinct forms of visual thought expressions and analyze them comprehensively. Our findings demonstrate that these forms differ in clarity and conciseness, yielding varying levels of MCoT improvement. Additionally, we explore the internal nature of visual thoughts, finding that visual thoughts serve as intermediaries between the input image and reasoning to deeper transformer layers, enabling more advanced visual information transmission. We hope that the visual thoughts can inspire further breakthroughs for future MCoT research.

Ling Xing, Alex Jinpeng Wang, Rui Yan et al.

Real-world applications are stretching context windows to hundreds of thousand of tokens while Large Language Models (LLMs) swell from billions to trillions of parameters. This dual expansion send compute and memory costs skyrocketing, making token compression indispensable. We introduce Vision Centric Token Compression (Vist), a slow-fast compression framework that mirrors human reading: the fast path renders distant tokens into images, letting a frozen, lightweight vision encoder skim the low-salience context; the slow path feeds the proximal window into the LLM for fine-grained reasoning. A Probability-Informed Visual Enhancement (PVE) objective masks high-frequency tokens during training, steering the Resampler to concentrate on semantically rich regions-just as skilled reader gloss over function words. On eleven in-context learning benchmarks, Vist achieves the same accuracy with 2.3 times fewer tokens, cutting FLOPs by 16% and memory by 50%. This method delivers remarkable results, outperforming the strongest text encoder-based compression method CEPE by 7.6% on average over benchmarks like TriviaQA, NQ, PopQA, NLUI, and CLIN, setting a new standard for token efficiency in LLMs. The source code will be released.

Alex Jinpeng Wang, Linjie Li, Zhengyuan Yang et al. · microsoft-research

Recent advancements in autoregressive and diffusion models have led to strong performance in image generation with short scene text words. However, generating coherent, long-form text in images, such as paragraphs in slides or documents, remains a major challenge for current generative models. We present the first work specifically focused on long text image generation, addressing a critical gap in existing text-to-image systems that typically handle only brief phrases or single sentences. Through comprehensive analysis of state-of-the-art autoregressive generation models, we identify the image tokenizer as a critical bottleneck in text generating quality. To address this, we introduce a novel text-focused, binary tokenizer optimized for capturing detailed scene text features. Leveraging our tokenizer, we develop \ModelName, a multimodal autoregressive model that excels in generating high-quality long-text images with unprecedented fidelity. Our model offers robust controllability, enabling customization of text properties such as font style, size, color, and alignment. Extensive experiments demonstrate that \ModelName~significantly outperforms SD3.5 Large~\cite{sd3} and GPT4o~\cite{gpt4o} with DALL-E 3~\cite{dalle3} in generating long text accurately, consistently, and flexibly. Beyond its technical achievements, \ModelName~opens up exciting opportunities for innovative applications like interleaved document and PowerPoint generation, establishing a new frontier in long-text image generating.

Zhuobai Dong, Junchao Yi, Ziyuan Zheng et al.

Understanding the physical world - governed by laws of motion, spatial relations, and causality - poses a fundamental challenge for multimodal large language models (MLLMs). While recent advances such as OpenAI o3 and GPT-4o demonstrate impressive perceptual and reasoning capabilities, our investigation reveals these models struggle profoundly with visual physical reasoning, failing to grasp basic physical laws, spatial interactions, and causal effects in complex scenes. More importantly, they often fail to follow coherent reasoning chains grounded in visual evidence, especially when multiple steps are needed to arrive at the correct answer. To rigorously evaluate this capability, we introduce MVPBench, a curated benchmark designed to rigorously evaluate visual physical reasoning through the lens of visual chain-of-thought (CoT). Each example features interleaved multi-image inputs and demands not only the correct final answer but also a coherent, step-by-step reasoning path grounded in evolving visual cues. This setup mirrors how humans reason through real-world physical processes over time. To ensure fine-grained evaluation, we introduce a graph-based CoT consistency metric that verifies whether the reasoning path of model adheres to valid physical logic. Additionally, we minimize shortcut exploitation from text priors, encouraging models to rely on visual understanding. Experimental results reveal a concerning trend: even cutting-edge MLLMs exhibit poor visual reasoning accuracy and weak image-text alignment in physical domains. Surprisingly, RL-based post-training alignment - commonly believed to improve visual reasoning performance - often harms spatial reasoning, suggesting a need to rethink current fine-tuning practices.

Jiacheng Hou, Yining Sun, Ruochong Jin et al.

Recent advances in large image editing models have shifted the paradigm from text-driven instructions to vision-prompt editing, where user intent is inferred directly from visual inputs such as marks, arrows, and visual-text prompts. While this paradigm greatly expands usability, it also introduces a critical and underexplored safety risk: the attack surface itself becomes visual. In this work, we propose Vision-Centric Jailbreak Attack (VJA), the first visual-to-visual jailbreak attack that conveys malicious instructions purely through visual inputs. To systematically study this emerging threat, we introduce IESBench, a safety-oriented benchmark for image editing models. Extensive experiments on IESBench demonstrate that VJA effectively compromises state-of-the-art commercial models, achieving attack success rates of up to 80.9% on Nano Banana Pro and 70.1% on GPT-Image-1.5. To mitigate this vulnerability, we propose a training-free defense based on introspective multimodal reasoning, which substantially improves the safety of poorly aligned models to a level comparable with commercial systems, without auxiliary guard models and with negligible computational overhead. Our findings expose new vulnerabilities, provide both a benchmark and practical defense to advance safe and trustworthy modern image editing systems. Warning: This paper contains offensive images created by large image editing models.

Rifen Lin, Alex Jinpeng Wang, Jiawei Mo et al.

Multimodal pretraining has revolutionized visual understanding, but its impact on video-based person re-identification (ReID) remains underexplored. Existing approaches often rely on video-text pairs, yet suffer from two fundamental limitations: (1) lack of genuine multimodal pretraining, and (2) text poorly captures fine-grained temporal motion-an essential cue for distinguishing identities in video. In this work, we take a bold departure from text-based paradigms by introducing the first skeleton-driven pretraining framework for ReID. To achieve this, we propose Contrastive Skeleton-Image Pretraining for ReID (CSIP-ReID), a novel two-stage method that leverages skeleton sequences as a spatiotemporally informative modality aligned with video frames. In the first stage, we employ contrastive learning to align skeleton and visual features at sequence level. In the second stage, we introduce a dynamic Prototype Fusion Updater (PFU) to refine multimodal identity prototypes, fusing motion and appearance cues. Moreover, we propose a Skeleton Guided Temporal Modeling (SGTM) module that distills temporal cues from skeleton data and integrates them into visual features. Extensive experiments demonstrate that CSIP-ReID achieves new state-of-the-art results on standard video ReID benchmarks (MARS, LS-VID, iLIDS-VID). Moreover, it exhibits strong generalization to skeleton-only ReID tasks (BIWI, IAS), significantly outperforming previous methods. CSIP-ReID pioneers an annotation-free and motion-aware pretraining paradigm for ReID, opening a new frontier in multimodal representation learning.

Ling Xing, Alex Jinpeng Wang, Rui Yan et al.

People see text. Humans read by recognizing words as visual objects, including their shapes, layouts, and patterns, before connecting them to meaning, which enables us to handle typos, distorted fonts, and various scripts effectively. Modern large language models (LLMs), however, rely on subword tokenization, fragmenting text into pieces from a fixed vocabulary. While effective for high-resource languages, this approach over-segments low-resource languages, yielding long, linguistically meaningless sequences and inflating computation. In this work, we challenge this entrenched paradigm and move toward a vision-centric alternative. Our method, SeeTok, renders text as images (visual-text) and leverages pretrained multimodal LLMs to interpret them, reusing strong OCR and text-vision alignment abilities learned from large-scale multimodal training. Across three different language tasks, SeeTok matches or surpasses subword tokenizers while requiring 4.43 times fewer tokens and reducing FLOPs by 70.5%, with additional gains in cross-lingual generalization, robustness to typographic noise, and linguistic hierarchy. SeeTok signals a shift from symbolic tokenization to human-like visual reading, and takes a step toward more natural and cognitively inspired language models.

Haochen Han, Alex Jinpeng Wang, Peijun Ye et al.

The data appetite for Vision-Language Models (VLMs) has continuously scaled up from the early millions to billions today, which faces an untenable trade-off with data quality and inevitably introduces Noisy Correspondence (NC) samples. Undoubtedly, such semantically unrelated data significantly impairs the performance of VLMs. Previous efforts mainly address this challenge by estimating refined alignment for more precise guidance. However, such resource-intensive pipelines that train VLMs from scratch struggle to meet realistic data demands. In this paper, we present a brand new perspective that seeks to directly eliminate the harmful effects of NC in pre-trained VLMs. Specifically, we propose NCU, a Noisy Correspondence Unlearning fine-tuning framework that efficiently enhances VLMs' robustness by forgetting learned noisy knowledge. The key to NCU is learning the hardest negative information, which can provide explicit unlearning direction for both false positives and false negatives. Such twin goals unlearning process can be formalized into one unified optimal transport objective for fast fine-tuning. We validate our approach with the prevailing CLIP model over various downstream tasks. Remarkably, NCU surpasses the robust pre-trained method on zero-shot transfer while with lower computational overhead. The code will be released upon acceptance.

Alex Jinpeng Wang, Linjie Li, Yiqi Lin et al.

Training models with longer in-context lengths is a significant challenge for multimodal model due to substantial GPU memory and computational costs. This exploratory study does not present state-of-the-art models; rather, it introduces an innovative method designed to increase in-context text length in multi-modality large language models (MLLMs) efficiently. We present Visualized In-Context Text Processing (VisInContext), which processes long in-context text using visual tokens. This technique significantly reduces GPU memory usage and floating point operations (FLOPs) for both training and inferenceing stage. For instance, our method expands the pre-training in-context text length from 256 to 2048 tokens with nearly same FLOPs for a 56 billion parameter MOE model. Experimental results demonstrate that model trained with VisInContext delivers superior performance on common downstream benchmarks for in-context few-shot evaluation. Additionally, VisInContext is complementary to existing methods for increasing in-context text length and enhances document understanding capabilities, showing great potential in document QA tasks and sequential document retrieval.