Large Vision Language Models

Large Vision-Language Models (LVLMs) have achieved remarkable progress in visual understanding tasks such as image captioning and visual question answering. However, they remain susceptible to hallucinations, generating content that is inconsistent with the actual visual input. Existing methods primarily intervene at the decoding stage, while overlooking a critical source of hallucinations: irrelevant or noisy visual tokens that mislead the decoding process. To address this issue, we propose SeeMe, a training-free framework that introduces the concept of feature engineering from traditional machine learning into LVLMs. SeeMe restructures visual tokens through a three-stage token engineering process to suppress hallucination sources while preserving informative visual evidence. Experiments on MME, POPE, and AMBER benchmarks across four LVLMs demonstrate that SeeMe consistently reduces hallucinations and improves output consistency, providing a novel perspective for mitigating hallucinations in LVLMs.

Automated fine-grained perception of calligraphy styles--a task vital to cultural heritage preservation--remains a critical challenge for Large Vision-Language Models (LVLMs), largely constrained by existing datasets that suffer from modal mixture and flattened labels. To bridge this gap, we introduce HCSU, the first comprehensive dataset tailored for fine-grained Historical Calligraphy Style Understanding. HCSU comprises 39,307 meticulously curated character images from 49 historically prominent calligraphers across 10 dynasties, systematically decoupling authentic ink manuscripts (Tie) from stone rubbings (Bei) to resolve the long-standing modal mixture problem. Moving beyond conventional flattened labels, HCSU provides hierarchical expert-written aesthetic descriptions, enabling two rigorous evaluation protocols: fine-grained style discrimination and interpretable aesthetic reasoning. Extensive evaluations reveal a persistent gap between calligraphy-related knowledge and visually grounded style perception: state-of-the-art LVLMs show non-trivial performance but remain sensitive to script-level, textual, and source-specific cues, and often struggle to ground aesthetic judgments in fine-grained brushwork evidence. Ultimately, the HCSU benchmark exposes fundamental limitations in current multimodal architectures, aiming to inspire the evolution of expert-level visual reasoning for cultural heritage preservation. The dataset is available at https://huggingface.co/datasets/Tongji209/HCSU.

Vision-Language-Action (VLA) models acquire broad embodied capabilities through large-scale pretraining, yet their generalization remains far more fragile than that of LLMs and VLMs. The prevailing remedy, post-training via supervised fine-tuning or reinforcement learning, improves task-specific performance but narrows the generalist capability that makes pretraining valuable. We identify a key bottleneck: VLA failures stem not only from action generation but also from action evaluation. A diagnostic pass@k study confirms that frozen VLAs already contain competent behaviors in their output distribution, with overall success rates rising from 33% at pass@1 to 92% at pass@32. Inspired by this, we propose SVA (Search, Value, and Act), a simple framework that equips frozen VLA policies with long-term consequence awareness. SVA first uses Monte-Carlo tree search in simulation to fully explore the VLA's output distribution and collect diverse trajectories annotated with empirical returns; this knowledge is then distilled into a lightweight Q-value model that predicts the expected consequence of candidate actions; at deployment, the frozen VLA proposes multiple candidates and the evaluator selects the one with the highest uncertainty-regularized Q-value, requiring no simulator access. By decoupling action proposal from consequence evaluation, SVA preserves the generalization capacity of the VLA backbone while substantially improving task success rates. Experiments across embodied benchmarks show that SVA consistently improves generalization on unseen tasks and exhibits strong test-time scaling behavior. Strikingly, SVA enables a 9B VLA to outperform a 27B VLA by 7 points at 27% lower inference latency, suggesting that scaling test-time evaluation is more cost-effective than scaling model size.

Large vision language models (VLMs) report strong accuracy on medical question-answering, yet it remains unclear whether they reason from visual evidence or exploit textual shortcuts. We introduce a counterfactual evaluation framework that decouples visual and textual contributions by substituting input images with controlled surrogates blank, pixel-shuffled, image-absent, and CLIP-retrieved hard negatives and derive a suite of grounding metrics including the Visual Reliance Score (VRS) and Visual Hallucination Rate (VHR). We further introduce CORAL (COntrastive Retrieval-Augmented Learning), a 7B-parameter LoRA fine-tune of Qwen2.5-VL-7B trained with a Contrastive Grounding Objective (CGO) that penalises answer invariance under hard-negative image swaps. On a paired controlled evaluation across four closed-form medical VQA benchmarks (PathVQA, PMC-VQA, SLAKE, VQA-RAD; n=400 total), CORAL improves macro accuracy by +6.7 pp (P(Delta>0)=0.988) and reduces VHR by 8.0 pp (P<0.001) over the matched Qwen2.5-VL-7B base; neither MedVLThinker RL variant achieves a significant gain on either metric. Cross-domain diagnostics further reveal that image substitution costs only <=6.5 pp on medical benchmarks versus 48-61 pp on general-domain tasks, situating the grounding gap that CGO targets. We discuss evaluation limitations openly including train/eval benchmark overlap and underpowered secondary metrics and release our framework, training code, and model weights to support reproducible grounding audits of medical VLMs.

Affordance grounding aims to localize image regions that support a specific action, serving as a core capability for physical intelligence and embodied perception. Previous studies have primarily relied on weakly supervised learning with action labels from exocentric images. However, these methods often struggle with visually ambiguous exocentric images containing co-occurring actions; moreover, they fail to distinguish semantically similar actions because existing methods typically rely on brief action phrases that lack rich semantic details for action-specific localization. Although large vision-language models (LVLMs) encode rich action semantics and their action-conditioned textual outputs implicitly contain spatial cues, they do not directly provide action-specific spatial localization. To address these problems, we propose TokAG, a zero-shot affordance grounding framework that exploits the token-level semantic-spatial signals in LVLMs to localize action-relevant regions without external supervision. We observe that attention maps associated with different LVLM output tokens vary significantly, with many attending to irrelevant regions such as the background. Thus, we introduce a spatial-aware token-selection mechanism to systematically evaluate each output token and select the one whose attention maps exhibit dominant activation over the target object, instead of relying on arbitrary attention maps. By extracting these object-focused attention maps, we transform the LVLM's implicit semantic signals into zero-shot affordance heatmaps. Our zero-shot framework consistently outperforms prior weakly supervised approaches across multiple benchmarks, improving NSS by 10.7% on the unseen split of AGD20K and by 29.7% on HICO-IIF. The code and models will be made publicly available.

Multimodal retrieval-augmented generation (RAG) grounds a generator in evidence drawn from heterogeneous modalities -- text, tables, and images. The dominant deployment choice is binary and made before the model has tried to answer: either run a cheap text(+table) pipeline, or pay for an expensive vision-language model (VLM) over every image. Recent adaptive systems improve on this by selecting the modality or fidelity pre-retrieval, from a question-conditioned predictor of which modality will be needed. We show that this is the wrong decision point. Through an oracle headroom analysis on MultiModalQA, we find that the relevance of a modality to a question is a weak predictor of whether that modality is actually needed to answer correctly: a large fraction of questions whose gold support includes an image are nonetheless answerable from text and tables alone, and a pre-retrieval router that escalates on apparent visual relevance over-escalates substantially relative to an oracle. We propose \textbf{post-hoc selective modality escalation}: answer cheaply from text and tables, run a verifier on the (query, draft answer, evidence) tuple that localizes which modality is missing, and pay for VLM evidence only there. A calibrated value-of-escalation router then decides whether the expected accuracy gain justifies the visual cost. On MultiModalQA, our router recovers the accuracy of an always-on VLM pipeline while issuing far fewer visual calls, and closes most of the gap to the oracle escalation rate. The result extends a routing-signal hierarchy established for retrieval depth and reasoning hops to a third axis -- modality -- under a single cost-aware selective-escalation view.

A frozen vision-language-action model (VLA) receives recent observations at every decision step, yet prior work has focused on adding memory rather than asking how existing history is represented and used. We study this temporal axis using layer-resolved linear probing and causal interchange interventions across three VLAs from two architecture families. We find a three-part dissociation. First, past-frame content remains linearly decodable throughout the network. Second, information unique to history beyond the current frame is nearly absent, indicating that stored history is largely a redundant copy of the present. Third, history is causally deployed only when the current frame is heavily degraded, while the action readout progressively loses dependence on history through the network. Although all models encode history similarly, their deployment strategies differ: under the same occlusion, one architecture increasingly relies on history as a fallback, whereas the other relies on it less. We further introduce a training-free temporal deployment audit that distinguishes these regimes. In the fallback regime, re-injecting history neither repairs occlusion nor disambiguates actions, confirming the redundancy of the stored representation. In the other regime, the same intervention reliably steers the predicted action toward the donor history. These results show that steerability depends on how history is deployed rather than whether it is encoded. VLAs do not forget the past; they largely fail to represent it as information distinct from the present. Our findings suggest that future memory augmentation should inject information unique to the past rather than simply more history.

Vision-language alignment powers open-vocabulary recognition, retrieval, and LVLM grounding, yet natural captions are often underspecified, making similarity brittle and overly confident under paraphrase and omitted details. We aim to learn representations whose matching is stable across caption views and whose confidence reflects how strongly text constrains an image. We propose Text as Partial Constraint (TPC), a core-residual alignment framework that treats multi-view captions as incomplete supervision. It distills a consensus semantic core as the alignment target, learns a single-view core predictor for standard inference with one query, and explicitly discourages vision-language similarity from depending on the orthogonal unsaid residual. An uncertainty-aware contrastive objective further softens alignment when caption views disagree, reducing overconfident updates under weak language constraints. Across zero-shot recognition and adversarial robustness, TPC achieves 81.42/64.05 Top-1 clean/robust accuracy on ImageNet and 76.19/52.03 on an Avg-14 transfer suite, while improving LVLM transfer with 85.16 POPE F1 and 59.57 OKVQA accuracy under an LLaVA-1.5-7B stack. These results suggest that modeling text as a partial constraint is a practical and principled route to more reliable vision-language representations under underspecified language supervision.

Large Vision-Language Models (LVLMs) exhibit remarkable vision-language capabilities and are increasingly deployed in real-world applications such as personal assistants, document analysis systems, and embodied agents. However, their dual-modal attack surfaces make them vulnerable to jailbreak attacks. Existing LVLM jailbreaks rely on simple designs, e.g., short text and out-of-distribution images. Nevertheless, recent advancements in both large language model backbones and multimodal mechanisms undermine these attacks, particularly their transferability among model architectures. To overcome this limitation, we propose a novel information overloading method that is equipped with both extensive text and multi-dimensional image attacks. These components are arranged in recursion-based image-typography layouts to exponentially increase multimodal information complexity. This overloading approach amplifies the cross-modal processing required, which undermines the safety alignment in LVLMs. Extensive experiments on both open-sourced and commercial LVLMs establish our method as a new state-of-the-art LVLM jailbreak attack. On open-source models, our method achieves an average ASR of 88.6%; on commercial LVLMs, it reaches an average ASR of 84.0%, exceeding the best baseline by 48.7%. Moreover, our prompts optimized on open-source surrogate models transfer effectively across model families. Beyond empirical results, we probe the safety-critical information flows within victim LVLMs. Our observations reveal that complex image-typography compositions induce intensified cross-modal processing and reduce the model's certainty in generating refusal responses. Together, these findings highlight information overloading as a practical and emerging safety risk for real-world LVLM deployments, underscoring the need for stronger defenses against complex multimodal jailbreak inputs.

We introduce VSeek, an agentic framework that transforms long-video question answering (LVQA) from a passive, single-pass perception task into a multi-turn retrieval process. VSeek utilizes a natural language-driven search to identify relevant context within long videos and is post-trained with reinforcement learning (RL) to jointly formulate targeted search queries and reason over retrieved clips for LVQA. While RL post-training has revolutionized reasoning in symbolic domains such as mathematics and code, its application to long-video understanding remains hindered by a lack of verified rewards. To ensure that the retrieved context is relevant, we propose a novel neuro-symbolic approach that bridges open-ended natural language with discrete visual verification. Specifically, complex user queries are compiled into formal temporal logic specifications for systematically decomposing natural language questions into a definitive checklist of required atomic visual primitives, such as key objects and activities, along with their temporal ordering. These systematically derived grounding events provide the critical feedback signal for RL post-training, enabling dense, verifiable rewards based on the successful retrieval of these specific visual elements rather than relying entirely on outcome-only answer accuracy. By explicitly optimizing for this verifiable evidence-seeking behavior, VSeek improves Pass@1 scores by up to 8% and Pass@4 scores by 15% on long-video understanding benchmarks compared to base models. We open-source our code at https://utaustin-swarmlab.github.io/VSeek.

Multilingual documents encapsulate rich regional cultures, scientific discoveries, and historical records. Parsing this content into structured, machine-readable formats is critical for unlocking global knowledge. However, existing benchmarks predominantly focus on high-resource languages like English and Chinese, creating an evaluation blind spot concerning model performance on other languages. While recent Vision-Language Models (VLMs) claim support for hundreds of languages, the lack of ground truth makes it impossible to empirically verify these capabilities. To bridge this gap, we introduce MORE, a large-scale benchmark designed for multilingual document parsing evaluation. MORE distinguishes itself through three key dimensions: (1) Unprecedented Scale: It covers 149 languages, making it the most linguistically diverse benchmark to date; (2) Structural Complexity: Unlike previous works, it extends evaluation beyond plain text to include structural elements such as code blocks, tables, and catalogs; and (3) Data Authenticity: All samples are curated from real-world documents via a model-assisted, human-refined annotation pipeline. We evaluate state-of-the-art models using MORE, establishing new performance baselines for long-tail languages and validating the benchmark's effectiveness in diagnosing model capabilities in realistic, diverse scenarios. The MORE dataset will be available at https://github.com/zimoqingfeng/MORE.

Vision Language Models (VLMs) are increasingly applied to software engineering artifacts, especially UML class diagrams whose meaning depends on visual notation. Yet, it is unclear whether VLMs actually read such diagrams or instead answer from pretrained priors about how classes typically relate. We introduce a controlled UML benchmark in which each prior-conforming diagram is paired with its prior-conflicting counterpart that (1) preserves the same class names and layout while (2) reverses only the relation arrow. We evaluate eight open-source VLMs from two model families, InternVL3.5 and Qwen3, alongside two closed-source frontier models GPT-5.4 and GPT-5.4 Mini. Across the eight open-source models, reversing the arrow reduces relation-direction accuracy by 33.48% on average, while GPT-5.4 Mini retains a 10% gap. In the harder three-class condition, accuracy drops sharply by 45.28% for open-source models, and even 18.62% for the GPT-5.4 family on average. Scaling provides only limited improvements and is family-dependent. Our benchmark presents a diagnostic prior-driven failure in diagram-grounded software understanding. Our artifact is available at https://anonymous.4open.science/r/UMLKnowledgeConflict-8461.

Cloud-edge Large Vision-Language Model (LVLM) inference enables efficient deployment by splitting computation between edge devices and cloud servers. In this process, intermediate vision tokens are transmitted from the edge to the cloud over a communication link, thereby exposing a new attack surface. We study vision token manipulation attack (VTM-Attack) under a black-box man-in-the-middle setting, where an adversary intercepts and manipulates a subset of transmitted vision tokens under a budget constraint. We propose four naïve attack strategies and an optimization-based token selection method. Experiments on 6 state-of-the-art LVLMs (3B-72B) across 4 benchmarks show that manipulating only 10% of vision tokens can reduce accuracy by up to 88.31%. These results reveal a critical vulnerability in cloud-edge LVLM inference.

Adapting frozen vision foundation models to fine-grained segmentation now largely depends on backbone selection. Whether the backbone applies global attention to a high-resolution token set predicts whether a low-rank adapter turns resolution into accuracy. Isotropic ViTs attend globally over the full grid and keep improving with resolution; hierarchical backbones confine early attention to local windows and pool the grid before their global stages, plateauing at lower resolutions. A controlled six-backbone study establishes the pattern, and editing the backbone points to the cause: pooling keeps the benefit, removing global attention does not. The effect is specific to low-rank adaptation. Under one fixed pipeline, SALT (Side-stem, Attention-gated U-Net, Low-rank Tuning), one RGB-only pass on a strong isotropic backbone wins the best S-measure on the four data-matched camouflaged sets, and leads every marine and salient set. It reaches a new state of the art on both marine-animal benchmarks (MAS3K mIoU 0.878).

Feed-forward 3D models are commonly trained using either expensive geometric supervision or self-supervised photometric objectives, both of which provide incomplete learning signals. We introduce Vision-Language Reprojection Consistency (VLRC), a scalable auxiliary objective that exploits frozen vision-language representations as semantic multi-view supervision. Given a predicted 3D reconstruction, VLRC reprojects dense vision-language features across views and enforces feature consistency between corresponding image locations, requiring no additional 3D annotations. The objective integrates seamlessly with both self-supervised monocular reconstruction and supervised-pretrained feed-forward 3D models during unlabeled adaptation. By aligning geometry with language-grounded features, VLRC not only improves depth and camera estimation but also enables more coherent multi-view semantic fusion for open-vocabulary 3D scene understanding. Experiments on indoor and outdoor benchmarks demonstrate consistent gains in 3D reconstruction accuracy and zero-shot open-vocabulary 3D semantic segmentation.

Large vision-language models can reason over multimodal inputs by generating textual chains of thought (CoT). A key capability exhibited in CoT reasoning is self-reflection: revisiting earlier decisions and correcting previous errors. However, existing LVLMs often fail to properly attend to visual inputs during reflection, limiting their ability to translate feedback into grounded corrections, especially for out-of-distribution images. To address this issue, we propose a novel reinforcement learning training framework VRRL, with two components explicitly designed to elicit visually grounded self-reflection. First, we randomly mask trajectory prefixes during training to emphasize recovery from incorrect intermediate predictions rather than making early mistakes. Second, we introduce buffered roll-ins from an experience replay buffer to expose the model to diverse failure states that it must learn to correct. We evaluate our approach on visual grounding tasks involving tables and charts, as well as spatial navigation benchmarks. While off-the-shelf and conventionally fine-tuned models degrade substantially under distribution shift, our method substantially improves average out-of-distribution accuracy over standard RL and reflection-oriented fine-tuning baselines by using self-reflection effectively.

In the automotive domain, in-car scene understanding (ISU) enables the detection of safety-critical events, such as driver distraction, and supports drivers or passengers by analyzing the in-car scene and adapting the environment (e.g., ambient lighting). The industry is increasingly exploring vision-language models (VLMs) to interpret camera-recorded in-car scenes and extract information for downstream reasoning tasks. However, VLMs may generate incomplete, erroneous, or misleading scene descriptions, highlighting the need for systematic testing. Collecting real in-vehicle data is costly, difficult to scale, and often infeasible, particularly in early design stages. In this paper, we present ISU-Test, an automated testing approach that combines rendering-based scene generation with search-based testing to evaluate ISU systems. By framing testing as an optimization problem and systematically modifying scene parameters, our method generates diverse in-car scenarios and explores a wide range of configurations. We evaluate ISU-Test on both an industrial prototype and open-source VLMs across two case studies: question answering and captioning, comparing against randomized scenario generation. Results show that ISU-Test significantly outperforms the baseline, achieving up to 10 times higher failure rates and up to 3.6 times higher failure coverage.

Flow-matching vision-language-action policies generate robot action chunks through an iterative transport process, creating an opportunity for test-time guidance without retraining the base policy. We study this opportunity in Guided Action Flow, an inference-time framework that keeps a pretrained SmolVLA policy frozen and uses a learned action-chunk critic to guide its reverse-time flow sampler. The critic is trained from real success and failure rollouts, can condition on task-description features from the frozen SmolVLA language pathway, and is used only through action gradients during sampling. We evaluate the approach on LIBERO manipulation tasks. A single-task critic improves success from 68.0% to 82.0% on one seed window and from 82.0% to 86.0% on another. A multi-family task-description critic improves validation success from 46.0% to 56.0%, while the locked held-out test gain is positive but modest, from 65.0% to 67.5%. These results support the feasibility of Q-guided inference for frozen flow-matching VLA policies, while showing that critic generalization and uncertainty-aware guidance remain the central bottlenecks.

License Plate Recognition (LPR) systems are critical tools in traffic monitoring, security enforcement, and urban mobility management. Traditional LPR systems often rely on a multi-stage pipeline involving object detection using You Only Look Once (YOLO) and Optical Character Recognition (OCR), which suffer from limitations such as high resource demands, poor performance in unstructured environments, and the need for large annotated datasets. This study explores the potential of Vision-Language Models (VLMs) as a unified, zeroshot learning solution for Nigerian license plate recognition. Using a curated dataset of 88 challenging real-world images collected in Nigeria, we evaluate five selected VLMs: Gemini 2.0 Flash Exp (Google DeepMind), Qwen2.5-VL-7B-Instruct (Alibaba), GPT-4o (OpenAI), Claude 4 Sonnet (Anthropic), and Llama 3.2 Vision 90b (Meta). Results based on Character Error Rate (CER) reveal that Gemini and Qwen significantly outperform other models in both accuracy and robustness, on the challenging image scenarios. This work highlights the practical advantages of VLMs over YOLO+OCR, questions the claims by model providers, and compares the performances of the VLMs.

Using molecular large language models (LLMs) as a unified framework for understanding molecular structures and functions is emerging as a new trend in tasks such as molecular design and drug discovery. However, these models struggle to fully capture the visual representation of molecular structures, limiting their potential. While existing molecular vision-language models (VLMs) show promise, they still face challenges in structural alignment and lack the necessary topological modeling for accurate molecular understanding. To address this, we propose MolSight, a graph-aware vision-language model framework designed to enhance the understanding of molecular images by VLMs. MolSight integrates a Molecular Topology Module to inject chemical-bond adjacency information into vision tokens, and a Molecular Grounding Module to align visual features with chemical symbolic semantics. Our experiments demonstrate that MolSight significantly outperforms existing VLMs, molecular LLMs, and specialized tools across multiple chemical visual understanding tasks, achieving a new level of molecular image reasoning.

Large vision-language models (LVLMs) have achieved strong performance across many medical imaging tasks, yet their application to ultrasound remains limited due to its inherent complexity and variability. In this work, we revisit what is truly needed to enable real-world ultrasound understanding. Instead of introducing complex architectures or elaborate training strategies, we show that data scale and clinically faithful data alignment are the key factors. We construct a large-scale dataset of 1.5M real-world ultrasound examinations, containing 17.7M images, multi-organ coverage, and paired uncurated clinical reports. Crucially, we organize the data at the examination level, aligning multiple images with their corresponding reports to reflect real clinical workflows. We then fine-tune a standard LVLM using low-rank adaptation (LoRA) on this dataset without task-specific modifications. Surprisingly, this simple recipe already leads to strong performance across diverse ultrasound understanding tasks, outperforming prior methods designed with more complex pipelines. Beyond these results, we present model and data scaling analyses that provide insights into the role of scale in ultrasound LVLMs.

Large Vision-Language Models (LVLMs) have achieved remarkable progress in multimodal understanding, yet their enormous parameter scale and cross-modal computation incur substantial memory and latency overhead, severely limiting real-world deployment on resource-constrained devices. Binarization offers an attractive solution by drastically reducing storage and computational costs. However, existing binarization methods neglect the varying importance of weights across different layers and modalities. This causes parameters irrelevant to downstream tasks to be unnecessarily retained, whereas modality-critical weights may not be adequately optimized, resulting in significant performance degradation. To address these challenges, we develop a novel \underline{S}ignificance-\underline{A}ware \underline{B}inarization for \underline{L}arge \underline{V}ision-\underline{L}anguage \underline{M}odels (SAB-LVLM). Specifically, after constructing Hessian matrices for textual and visual inputs, we propose a spatial significance map to distinguish full-precision weights activated under a single modality from those activated across modalities. We then devise a modality-guided integration strategy to obtain the significance-aware binarization map, which measures weight significance across layers and modalities. Subsequently, this binarization map is incorporated into the binarization objective as an error reweighting term, and binarization fitting is performed through an alternating significance-weighted update scheme. Extensive experiments illustrate the superiority of our SAB-LVLM over existing binary PTQ methods under an approximately 1-bit compression constraint. Our code is accessible at https://github.com/LyuQi127/SAB_LVLM.

Visual-spatial understanding, defined as the ability to infer object relationships and scene layouts from visual inputs, is fundamental to downstream tasks such as robotic navigation and embodied interaction. However, pre-trained vision-language models (VLMs) remain constrained by spatial uncertainty stemming from inherently 2D observations and by the scarcity of data for 3D spatial understanding. To address these limitations, we proposed a novel framework, SpaceEra, in the NeurIPS 2025 Spotlight paper. Although it achieved significant performance gains, we further observed that its effectiveness is hindered by insufficient input from scanning videos and weak reasoning constraints. To tackle these newly emerged challenges, we extend the original framework into a comprehensive system, termed SpaceEra++, which spans data construction, model design, training optimization, and prompting inference. Specifically, to alleviate input insufficiency, we introduce ScenePick, a frame sampling strategy that balances spatial coverage with object semantics to produce compact yet comprehensive scene representations. In addition, to enhance spatial reasoning, we develop SpaceAlign, which enforces pairwise object constraints by jointly exploiting absolute coordinates and relative spatial relations, thereby aligning optimization with spatial accuracy. Extensive experiments across multiple benchmarks demonstrate consistent improvements over strong baselines, while ablation studies validate both the individual and joint contributions of each component, and further analyses provide guidance for future research.

Open-vocabulary object detection aims to localize and classify objects beyond the fixed set of categories seen dur ing training. Recent open-vocabulary object detection methods improve localization and classification for unseen categories by leveraging a frozen VLM as a detector backbone. However, VLM classification score lacks recognizing position and scale of the object in an image. We observe that pretrained VLMs en able to classify foreground and background regions. According to this observation, we propose a simple inference-time Pro posal Calibration (ProCal) that improves localization quality of the classification score. ProCal computes a proposal prior by combining two scores: localization-aware foreground score and background-aware suppression score. Localization-aware foreground score captures whether a proposal contains an object area. Background-aware suppression score measures the extent to which the proposal resembles background. We analyze that ProCal suppresses false novel activation on background proposals and consistently ranks true novel proposals above background and partial novel proposals. Applied to CLIPSelf ViT-L/14, ProCal improves APr +2.5 on OV-LVIS. The analyses show that proposal-level localization-aware reranking effects to mitigate ranking miscalibration for novel objects.

Large vision-language models (LVLMs) exhibit strong reasoning ability but suffer from visual forgetting during long-horizon decoding, where attention progressively drifts away from visual evidence. Existing methods largely treat this issue as a late-stage attention decay problem or attempt to mitigate it through heuristic reminders or post-hoc attention lifting. Through systematic empirical analysis, we find that performance degradation under visual forgetting is largely driven by two overlooked factors: early-stage attention decay disrupts evidence acquisition, and attention concentration on a subset of task-irrelevant visual sink tokens. Motivated by these insights, we propose LASER, a post-training framework that regulates both the visual attention trajectory and intra-visual token attention distribution during reasoning. Technically, LASER introduces two complementary rewards: a Visual Grounding Reward, which encourages the model to maintain attention on semantically salient visual tokens throughout decoding, and a Sink Suppression Reward, which penalizes excessive attention concentration on visual sink tokens. Together, these rewards preserve early-stage grounding while preventing attention collapse onto uninformative regions. Extensive experiments on eight benchmark datasets demonstrate that LASER consistently outperforms strong baselines, validating attention-aware training as an effective remedy for visual forgetting.

Visual recognition models often fail when deployed in new environments. Domain Generalization (DG) addresses this by learning representations that remain invariant to environment-specific variations. Recent approaches increasingly rely on large vision-language models, assuming that preserving their expressive visual representations improves robustness. However, we show that such visual expressiveness can instead propagate spurious cues that tie representations to the training environments, hindering invariant learning. We therefore discard visual guidance and instead treat the language embedding space as the primary source of domain invariance, naturally acting as an information bottleneck that preserves core semantics while suppressing domain-specific variations. Extensive experiments across diverse backbones exhibit state-of-the-art performance and further analyze what makes guidance effective for robust generalization. These findings shift the focus of DG from improving representations to designing supervision that enforces invariance.

Large Vision-Language Models (LVLMs) have been increasingly integrated into robotic systems. However, these models may exhibit overthinking behaviors, where they generate excessively long reasoning traces, incurring an excessive inference time. This overthinking behavior poses a serious risk to robotic systems, as the adversary can deliberately trigger overthinking to slow down the decision making of a victim robotic system, causing a variety of safety issues (i.e., an overthinking-induced slowdown attack). To initiate this attack, an adversary can embed carefully crafted, human-readable scene text into the visual scene observed by a victim robotic agent, causing significant inference delays even under a strict black-box setting. Therefore, the embedded scene text serves as a significant "trigger" for the attack. This work systematically identifies and validates transferable triggers of overthinking in robotic systems by introducing a three-stage framework. First, we construct a diverse corpus of reasoning-intensive scene text and extract overthinking-correlated lexical features from short response prefixes. Second, we perform an efficient black-box search guided by a prefix-based proxy score while selectively confirming a small set of top candidates with full latency measurements. Third, we evaluate black-box transfer using a fixed pool of triggers on unseen images and multiple LVLMs, reporting latency amplification and attack success rates under standard thresholds. Across three representative LVLMs, all triggers yield slowdown ratios greater than 1.0x, with the strongest single-trigger case reaching 6.96x. The physical printing of the text trigger still causes up to 4.74x latency amplification. These results demonstrate that our discovered triggers are transferred between multiple LVLM models and consistently cause significant slowdowns in robotic systems.

Given one or more images of a railway crossing, can we leverage visual cues that allow us to robustly estimate how safe it is? Can we improve our ability to do so by introducing structured data (such as official accident reports) about the accident history of that crossing into our models? In this work, we explore how to best answer those questions towards building an AI system that can ingest multi-modal data for railway crossings and provide safety assessment and scores that align with expert opinion and with safety scoring used by the Federal Railroad Administration (FRA). To that end, we propose a proof-of-concept pipeline that delivers on that goal, while at the same time exploring and tackling a number of critical research challenges that pertain to different parts of the pipeline, from data preparation to different learning paradigms that can allow us to realize such a system. Indicatively, our proposed system identifies HIGH-RISK and LOW-RISK crossings with a macro F1 score of 0.757 and estimates FRA-based safety scores with an RMSE of 0.078 and correlation of 0.492 using a routed fine-tuned compact VLM pipeline, while producing qualitative results that align with domain-expert assessment.

Fine-grained visual reasoning remains challenging for vision-language models, especially when small but critical visual cues are buried in high-resolution images. Existing approaches rely on repeated cropping or test-time visual search to introduce local evidence, but they typically do not explicitly distinguish perception from reasoning. In this paper, we propose Perceive-to-Reason (P2R), a unified framework that formulates fine-grained visual reasoning as a two-stage process: the model first localizes question-relevant evidence as a Perceiver, and then answers the question as a Reasoner based on the annotated image and cropped regions. To better align training with this decoupled formulation, we further introduce Perception-Reasoning Alternating GRPO (PRA-GRPO), a role-aware reinforcement learning strategy that alternates between perception-focused and reasoning-focused updates using only final-answer supervision. Built on top of Qwen3-VL-Instruct-2B/4B/8B, P2R consistently improves performance across model scales. In particular, P2R-4B achieves 93.2% on V-Star, 81.9% on HR-Bench-4K, and 80.5% on HR-Bench-8K, substantially outperforming its corresponding backbone. Further experiments show that the benefits of P2R extend beyond high-resolution benchmarks to broader multimodal reasoning tasks. These results suggest that explicitly decoupling perception from reasoning provides an effective framework for fine-grained visual reasoning.

The evaluation of long-term video quality understanding remains an open challenge for large vision-language models (LVLMs). Existing video quality benchmarks predominantly focus on short clips and isolated distortions, overlooking the temporal continuity, cumulative degradation, and reasoning complexity inherent in long-duration content. To address these limitations, we present LongVQUBench, a comprehensive benchmark for long-term video quality understanding. LongVQUBench contains over 1200 diverse videos spanning movies, documentaries, surveillance footage, egocentric recordings, and animated content, accompanied by 1500 multiple-choice and open-ended questions for validation and testing. To assess perceptual reasoning across different temporal scopes, we introduce three progressively complex evaluation levels: (i) local event quality understanding (LQU) for analyzing localized distortions; (ii) cross-event quality reasoning (CQR) for integrating multiple degraded events; and (iii) global quality understanding (GQU) for holistic perceptual evaluation over extended durations. Furthermore, a needle distortion question-answering (NDQA) paradigm is embedded across all three levels, where spatial or temporal artifacts are sparsely inserted to probe fine-grained detection and reasoning capabilities. Extensive experiments on 14 state-of-the-art LVLMs reveal significant performance degradation with increasing video length and reasoning depth, highlighting their limited capacity for long-range temporal integration and perceptual attribution. We envision LongVQUBench as a foundational step toward the systematic, hierarchical, and explainable evaluation of LVLMs' long-term video quality understanding.

Video world models are emerging as a scalable alternative for evaluating generalist robot policies, bypassing the physical constraints and engineering burdens of real-world deployment. However, evaluating policies with video world models remains challenging, as world-model errors can make generated rollouts unreliable and slow inference limits large-scale throughput. We introduce RoboWorld, an automated evaluation pipeline that pairs a fast autoregressive video world model with a task-progress-aware vision-language model scoring. To enable reliable long-horizon autoregressive world-model rollouts, we propose Step Forcing, which combines anchored and one-step self-forwarded contexts to reduce train-test mismatch while preserving action-observation dynamics. Together, these components enable RoboWorld to align strongly with real-world robot evaluation across tasks and environments, achieving Pearson's r = 0.989 and Spearman's $ρ$ = 0.970.

Industrial inspection requires more than binary anomaly detection: a practical system should determine whether an anomaly exists, localize the defective region, identify the defect type, and provide interpretable visual evidence. Existing CLIP-based methods detect and localize anomalies well but offer limited language-level defect understanding, while instruction-tuned vision-language models can describe defects but do not natively produce pixel-level masks. We introduce GenAU, a Generalist vision-language framework for industrial Anomaly Understanding that unifies image-level detection, pixel-level segmentation, multi-type anomaly detection, and defect analysis in a single instruction-following model. GenAU augments a vision-language model with two segmentation tokens, [SEG_defect] and [SEG_normal], whose hidden states act as language-grounded queries over multi-scale visual features for pixel-level localization; the image-level score fuses this map with the decoder's textual normal/defect decision, while the language decoder produces structured defect-aware responses. Trained with a joint language-modeling and segmentation objective, GenAU covers all four tasks within one architecture and recipe, adding zero-shot multi-type detection and language-grounded defect analysis at a quantified cost to detection and segmentation. Across cross-dataset benchmarks, GenAU attains the strongest image-level detection among CLIP-based zero-shot methods on VisA and Real-IAD, with segmentation approaching but not surpassing specialized CLIP baselines.

Memory-based discrete vision-language navigation (VLN) agents must act under partial observability, yet even strong frozen backbones remain vulnerable at test time. Two common failure modes are stale historical evidence at memory readout and inefficient local backtracking during action selection. We present DART-VLN, a training-free test-time control framework for discrete VLN. DART-VLN combines Test-Time Memory Decay, a read-side memory reweighting rule that suppresses stale and redundant evidence without rewriting stored content, with Anti-Loop Regularization, a lightweight next-hop penalty that discourages immediate reversals during action selection. The framework introduces no new learnable parameters and leaves the learned backbone unchanged. Experiments on R2R and REVERIE show a consistent pattern: decay-only provides stable read-side gains, while decay+anti-loop achieves the best overall quality-efficiency trade-off, yielding shorter trajectories, lower runtime, and improved navigation performance in key settings. Behavioral analysis further confirms that anti-loop regularization reduces local backtracking and improves path efficiency under frozen backbones. Overall, the results show that modest test-time control can make memory-based discrete VLN more reliable and efficient without retraining.

Real-world e-commerce image editing often requires multiple, localized, and auditable operations rather than global restyling. This compositional nature poses a dual challenge: models must precisely apply all requested edits to the correct regions while preserving unmodified content, even under ambiguous instructions. Existing one-shot editors conflate intent resolution, spatial grounding, and synthesis into a single step, frequently resulting in partial execution failures, which is unacceptable for commercial scenarios. To address this, we introduce GMO-E$^2$DIT, an agentic editing framework that couples a Vision-Language Model (VLM) with a mask-conditioned image editor to tackle structured multi-turn task completion. Given an underspecified instruction, the VLM agent constructs a region-grounded edit agenda, effectively decoupling cognitive reasoning from generative rendering. The framework then executes sub-programs via operation-aware masks and references, utilizing a reflection-driven loop to inspect intermediate results and determine the subsequent state. This iterative mechanism reliably preserves safe partial progress, retries unfinished operations, and recovers from errors. Furthermore, we develop a unified data pipeline providing aligned supervision for planning, execution, and reflection, alongside EComEditBench, a comprehensive benchmark for instruction-driven evaluation. Extensive experiments demonstrate that GMO-E$^2$DIT achieves competitive performance compared to strong closed-source models, yielding superior instruction accuracy and edit fidelity over existing baselines.

Vision-language pretraining remains dominated by contrastive objectives, whereas vision-only self-supervised learning has largely adopted non-contrastive methods. At the same time, the role of vision-language encoders has shifted: they are increasingly deployed not as zero-shot classifiers but as the frozen visual backbone of vision-language models and dense prediction systems, which consume the full grid of patch tokens rather than a single pooled embedding. We introduce LeVLJEPA, the first fully non-contrastive end-to-end vision-language pretraining method. LeVLJEPA learns through cross-modal prediction with stop-gradient targets and per-modality distributional regularization, without negatives, temperature, momentum encoder, or teacher-student schedule, and trains stably at large scale. We find that the resulting encoder provides markedly stronger dense semantic features for downstream use: as a frozen vision-language-model backbone, LeVLJEPA is the strongest of the evaluated encoders across GQA, VQAv2, and POPE under two distinct language models, and outperforms contrastive baselines on semantic segmentation, while remaining on par on global readouts such as linear probing. These results establish non-contrastive pretraining as an effective means of producing dense semantic vision features.

Benchmarks for vision-language models (VLMs) mostly test observational spatial reasoning: models describe relations already visible in the input. Existing what-if tasks typically vary the observer while keeping the scene fixed. Can VLMs instead predict the consequences of hypothetically moving or rotating an object? We introduce MindEdit-Bench, a benchmark of six spatial reasoning tasks built from three-photo smartphone triplets of newly captured indoor scenes via an automatic in-the-wild 3D scene-graph extraction pipeline. Four tasks probe perception and perspective transformation over observed structure; two new tasks, L4 (spatial editing) and L5 (cross-view visibility editing), probe object-level counterfactual reasoning, where correct answers are absent from all input images. Each question provides 8-24 structured answer choices, enabling answer-letter-level diagnosis of spatial and fallback errors. The benchmark covers 120 private indoor scenes not drawn from public datasets, reducing public-data pretraining-overlap risk. Across 15 VLMs on 1,003 human-verified questions, task-wise mean VLM accuracy is only 8%-31%, versus 81%-97% human majority-vote accuracy. The pooled human--best-VLM gap is 53 pp, with at least 39 pp on every task. The structured answer space further reveals non-uniform failures, including weaker camera-depth-axis inference and fallback behavior on difficult visibility-editing cases.

Large Vision-Language Models (LVLMs) rely extensively on Visual Instruction Tuning (VIT) to elicit their multimodal reasoning capabilities. However, we find a discrepancy: VIT often packs multiple language tasks about the same image for conversational, multi-turn training, whereas existing benchmarks evaluate LVLMs in isolated, single-turn scenarios. The models can suffer from visual attention decay and contextual overfitting during multi-turn training, making it hard for them to realize their full potential in the mismatched test phase. To close the gap, we propose learning with Stochastic Turn Depth (StochasT), which stochastically groups language tasks for the same image into clusters of varying sizes (turn depth) while preserving their organic order. Hence, while StochasT draws on Dropout and stochastic depth for ResNets, it does not actually drop anything to maximize the utility of the training data. Furthermore, we introduce a challenging, benchmark-agnostic evaluation mechanism based on the Balanced Latin Square to measure LVLMs' robustness under varying contextual dependencies. Extensive experiments demonstrate that StochasT effectively grants LVLMs strong, harmonized capabilities for both single-turn and multi-turn use cases.

Vision-language models (VLMs) have become a paradigm for multimodal learning, yet remain unstable due to object hallucination, weak visual grounding, and catastrophic forgetting after full-parameter instruction tuning. We claim these failures result from a lack of explicit control over visual representation learning during the standard next-token prediction objective. As a result, visual embeddings thus become passively optimized and prone to injecting redundant or spurious signals. To counter this, we introduce Information-Regularized Attention (IRA), a stochastic attention mechanism that explicitly regulates the amount of visual information injected into the hidden states of intermediate transformer layers. This local reparameterization translates uncertainty about visual representations into local noise that is independent across data points. Beyond evaluating model performance, we also quantify embedding properties, where IRA produces smoother curvature trajectories and suppresses attention-sink across all layers, indicating a more stable transformation of the visual signal. Our results suggest that stochastic attention is not merely a regularizer but a key contributor to representation learning in a generative architecture, offering a new direction for building more reliable VLMs.

We present DroneIQA-VLE, our solution to the ICME 2026 Drone-IQA Grand Challenge on Target-aware Image Quality Assessment for Low-altitude UAV Images. The framework jointly predicts global, target, and background quality scores by ensembling two complementary pipelines: (1) SigLIP2 vision encoders with multi-task regression heads, and (2) a LoRA-adapted Qwen3.5-9B multimodal large language model for quality score regression. The final global quality prediction is obtained by arithmetically averaging the outputs of both pipelines. Our method achieves 2nd place in the challenge, demonstrating its effectiveness. The code is available at https://github.com/sunwei925/DroneIQA-VLE.

Autonomous vehicles must safely navigate complex environments where planning-critical agents may be hidden from view. Current approaches often treat all occlusions with uniform conservatism, yielding needlessly defensive driving, or they infer hidden spaces without estimating the impact on the planner. This work bridges the critical gap between perception and planning by enabling Vision-Language Models (VLMs) to identify and reason about the specific hidden agents that are most critical to the ego-vehicle's trajectory. We introduce a novel framework that uses Planning KL-divergence (PKL), an information-theoretic metric, to systematically identify and rank occluded agents based on their impact on the ego vehicle's plan. Using this planning-aware ranking, we employ an expert VLM (GPT-5) to generate rich, structured annotations that capture the visual evidence and reasoning required for this task. We apply this framework to the nuScenes dataset to create a new benchmark focused on high-impact scenarios. We conduct comprehensive experiments on a wide range of general-purpose and domain-adapted VLMs, demonstrating that fine-tuning on our PKL-guided data yields dramatic performance improvements across all models. Notably, our results show that smaller, fine-tuned models significantly outperform their much larger zero-shot counterparts, and that our PKL-guided data selection strategy improves performance by approximately 30% over random sampling. Our work presents the first systematic approach for training VLMs to focus on planning-critical occlusions, enabling more semantically grounded and efficient risk assessment in autonomous driving.

Camera pose estimation from image streams is a critical component of spatial world models that integrate perception into planning and decision-making. Nearly all Visual Odometry (VO) and Simultaneous Localization and Mapping (V-SLAM) systems have focused on datasets containing raw, uncompressed videos. Many working systems instead use ubiquitous hardware units to efficiently compress and decode video streams, saving orders of magnitude in storage and bandwidth. However, this lossy compression introduces visual artifacts that hinder the performance of traditional tracking systems. We present VOCA, a causal stereo visual-odometry method that exploits codec information to improve tracking performance. We achieve state-of-the-art performance on causal VO for relative trajectory error, efficiency, and absolute trajectory error on compressed streams. This work highlights the potential of leveraging widely available video codec information for vision tasks.

We present a black-box model-stealing attack that recovers private vision-tokenizer configurations of deployed vision-language models (VLMs), including the visual patch size and input preprocessing pipeline. The key idea is a task-level side channel induced by ViT-style patchification: when a synthetic grid image is aligned with the hidden patch grid, boundary cues are erased at tokenization, causing periodic accuracy drop. By sweeping the grid cell size and measuring these collapses, we infer the patch size; by introducing padding and a consistency-check test, we further identify whether preprocessing is dynamic- or fixed-resolution and recover the target resize resolution. Across open-source Qwen-VL variants and proprietary models including GPT and Claude, we reliably recover tokenizer-related parameters. Finally, we show that such leakage enables preprocessing-aware transfer attacks and model-targeted adversarial manipulation.

In collaborative dialogue, shared perception does not guarantee shared interpretation. Mutual understanding must be established through interaction. We investigate whether vision-language models (VLMs) can distinguish what could be shared from what has been shared between dialogue participants through grounding. We formulate this as an interpretation-matching task on 13,077 annotated reference expressions from HCRC MapTask dialogues, and evaluate VLMs under systematically controlled manipulations of dialogue context and map-information access. Our results show that providing authentic map images improves overall performance but shifts models toward over-predicting alignment. Textual descriptions of the same map content reproduce this bias, while non-informative images suppress alignment predictions entirely, indicating that the bias is driven by task-relevant map content, not the visual channel. This improvement comes at the cost of degraded accuracy on non-aligned cases. Calibration analysis and reference-chain tracking further suggest that models rely on static referential cues on the maps rather than tracking how grounding unfolds through dialogue history. We observe these patterns most clearly in Qwen3-VL-8B-Instruct and, to varying degrees, in four additional models from two architecture families. In models that exhibit the bias, map content, whether presented visually or textually, is treated as evidence of mutual understanding, conflating potential with established common ground.

Vision-Language-Action (VLA) models have made significant strides in embodied intelligence by integrating the powerful representations of pre-trained Vision-Language Models (VLMs). However, the massive parameter scale of VLAs imposes a heavy computational burden, and these models exhibit extreme sensitivity to parameter pruning. Current paradigms often treat the resulting performance degradation as inevitable, relying on fine-tuning or low-rank corrections to recover efficacy. We challenge this convention by questioning whether the removed parameters are truly redundant if VLA pruning necessitates performance recovery to be effective, or if this paradigm masks the indiscriminate pruning of critical parameters. We revisit parameter redundancy through the lens of VLM-to-VLA adaptation, first quantifying the spatial distribution of parameter divergence during adaptation to reveal structured patterns across different modules. Subsequently, we introduce controlled pruning as a diagnostic probe: by comparing the direct impact of removing different parameter subsets on VLA performance without any fine-tuning, we establish a causal link between adaptation-induced divergence signals and functional contributions. Based on the discovered modular heterogeneities, we design a multi-module joint pruning scheme. Evaluations on the LIBERO benchmark demonstrate that our approach reduces the parameters of OpenVLA and $π_{0.5}$ by 12%--30% while maintaining approximately 90% of the original performance without any post-pruning recovery. In contrast, existing parameter pruning criteria result in total performance collapse when evaluated under the same recovery-free constraints. Our study reveals the parameter evolution mechanism in VLA adaptation and provides a new path for deploying efficient, robust robotic policies in resource-constrained environments.

Generalizable robotic grasping in cluttered environments is essential for deploying manipulators in unstructured human spaces, yet existing VLM-based methods rely on visual similarity for object matching, neglecting physical affordances such as handle graspability and material fragility, and operate open-loop without spatial reasoning or failure recovery, limiting their effectiveness when objects are densely packed or physically diverse. We present Agentic RAG-VLM, a unified framework that bridges VLM-based semantic understanding and physically grounded grasp execution by integrating retrieval-augmented generation (RAG) with vision-language models (VLMs) and agentic self-reflective planning. Agentic RAG-VLM introduces three tightly coupled components: (1) a Hierarchical Affordance-Aware RAG (HAA-RAG) that encodes four-dimensional affordance descriptors, including type, material, fragility, and graspable region, and retrieves strategies by functional affordance compatibility rather than visual appearance; (2) a Scene Graph Constraint Reasoner that constructs spatial relationship graphs from VLM perception and translates proximity, occlusion, and support constraints into concrete grasp parameter adjustments; and (3) an Agentic Self-Reflective Pipeline with a 14-type failure taxonomy and three-level adaptive retry for closed-loop grasp refinement. Evaluated on a 12-task benchmark spanning single-grasp, interactive, and long-horizon scenarios with 360 trials per configuration, Agentic RAG-VLM achieves 78.3 percent overall success, a 53.3 percentage-point absolute gain over VLM-only baselines, demonstrating that affordance-aware retrieval, scene graph reasoning, and agentic recovery are jointly essential for robust manipulation.

3D Visual Grounding (3DVG) aims to localize target objects in 3D scenes given natural language descriptions. Existing approaches typically perform reasoning over the entire scene, leading to ambiguous predictions and high computational cost, especially in cluttered environments. We observe that many referential expressions rely on local spatial context and often correspond to restricted spatial regions rather than the full scene. Motivated by this insight, we propose PruneGround, an effective plug-and-play framework for 3DVG built upon three key components. First, we introduce Language-Guided Spatial Pruning (LGSP), which leverages a frozen Vision Language Model (VLM) to identify language-relevant regions, thereby reducing spatial computation and grounding candidates in the narrower search space. Second, we propose MultiView-Conditioned Description Reformulation (MCDR), which decomposes complex expressions into simplified target-anchor relations and augments missing spatial cues through multi-view reasoning. Finally, we propose LLM-Grounder, which repurposes a detection-pretrained spatial LLM into a language-conditioned grounding model by aligning point cloud and linguistic representations within the pruned region. Extensive experiments on the three most popular point cloud benchmarks demonstrate that our method achieves state-of-the-art results on all three ScanRefer settings and on 9 out of 10 Nr3D/Sr3D settings. Code and models are publicly available: https://github.com/leduckhai/PruneGround

Vision-and-language navigation (VLN) in continuous environments requires an agent to ground instructions in egocentric observations while maintaining spatial understanding across long action sequences. Recent navigation foundation models have shown strong progress by scaling vision-language models, but they often learn navigation primarily as direct action generation, without explicitly modeling world states or predicting their future evolution. We introduce FutureNav, a VLM-based unified world-action modeling framework for vision-and-language navigation. Specifically, FutureNav jointly encodes text, visual, and spatial features and feeds them into the LLM, and optimizes four objectives for simultaneous world and action modeling: an action policy objective for navigation action prediction, inverse and forward dynamics objectives for modeling state transitions, and a future generation objective for predicting future spatial states. This unified architecture strengthens action prediction while explicitly modeling the world, without sacrificing inference speed. Extensive experiments show that, with only a 4B-scale backbone, FutureNav achieves state-of-the-art performance on multiple VLN benchmarks and substantially outperforms prior VLN methods, paving the way toward future world-action models for VLN. We will release the code and models to support future research.

Large Vision Language Models (LVLMs) have achieved remarkable success on vision-language tasks, yet fine-grained perception over high-resolution images and long-context videos remains challenging. As the number of visual tokens increases, the visual attention sink phenomenon becomes increasingly severe, causing irrelevant tokens to absorb a disproportionate amount of attention mass. Recent approaches attempt to mitigate this issue by explicitly predicting bounding boxes or temporal spans and re-encoding the cropped visual regions. Such methods depend on unreliable numeric localization in the discrete token space and incur significant computational overhead due to additional forward passes. In this work, we propose VisReflect, a simple yet effective framework that improves fine-grained perception in long visual contexts through latent visual reflection. Instead of decoding intermediate predictions into discrete tokens, the model generates continuous visual reflection that represents question-relevant visual features in the latent space. These reflections selectively emphasize salient regions or frames, guiding attention towards relevant visual tokens within a single forward pass. We conduct comprehensive evaluations on challenging high-resolution image benchmarks, including BLINK, V*, and HRBench-4K/8K, as well as video understanding benchmarks such as MVBench, VideoMME, and MLVU. Our method consistently improves over strong baselines, achieving gains of 4.1% on image benchmarks and 1.8% on video benchmarks. Compared with zooming-based methods, our model achieves comparable performance while reducing inference time by roughly 44% on video understanding.

High benchmark accuracy does not guarantee genuine use of visual evidence. We study this problem in traffic accident Video Question Answering (VideoQA), where correct answers should depend on scene-specific visual evidence but may instead be inferred from textual shortcuts. Through an audit of four public benchmarks, we find that several recent open-weight Vision-Language Models (VLMs) perform competitively, and sometimes better, without video input. On the MM-AU benchmark, removing video consistently improves accuracy, and adding more frames further degrades performance. To quantify visual dependence, we introduce two dataset-level diagnostics: Blind Gap, measuring above-chance text-only performance, and Visual Gain, measuring the marginal benefit of adding video. We further propose an instance-level Shortcut Score that combines text-only confidence with visual necessity signals, enabling continuous, training-free filtering of shortcut-prone questions. The resulting subsets reduce shortcut bias and improve visual grounding. Our findings reveal large differences in grounding quality across benchmarks and show that visually grounded evaluation, not just high accuracy, is essential in safety-critical VideoQA.

Existing domain-incremental learning (DIL) strategies call for massive amounts of data to adapt to new domains and suffer from the overfitting problem in the case of data scarcity. This paper puts forward a relatively uncharted problem, namely, few-shot domain incremental learning (FSDIL), taking into account the problem of extreme data shortages in the realm of DIL. A novel algorithm, namely Continual Vision-Language Consolidation (CVLC), is proposed to address the FSDIL problem, where the key idea lies in the concept of latent space reservation in the base domain coupled with dual coalescent projection (DCP) as a parameter-efficient fine-tuning method. First, the vision prototype is calibrated while multiple templates and synonyms are generated via LLMs to induce the language prototype. The vision and language prototypes are fused. Adaptation to never-ending arrivals of new domains is done by the DCP technique, fine-tuned in such a way to prepare the model to unseen domains via latent-space reservations committed in the base domain. CVLC is structured under shared and domain-specific components to combine general knowledge and domain-specific details. The advantage of our approach is demonstrated through a range of benchmark problems and comparisons with prior arts, in which CVLC outperforms them by up to a 16% gap. Our codes are shared publicly in https://github.com/Naeem-Paeedeh/CVLC .

Improving vision-language models (VLMs) on visual reasoning typically requires retraining or hand-designed prompts and tools. We present Dynamo, a training-free framework that adapts a frozen VLM without any weight updates. On a small labeled training subset, the agent inspects its own correct and incorrect attempts and evolves two complementary capabilities: reusable reasoning skills for cognitive bottlenecks, and executable visual tools for perceptual ones. Each generated tool is paired with a skill that specifies when to invoke it, and both capability types accumulate in a persistent library. Across four visual reasoning benchmarks and five VLM backbones, Dynamo improves direct inference on all 20 model--benchmark settings (avg. +5.6 acc). When the tool set is given in advance, the framework learns when to call each tool, and per-step tool choice improves on every tested backbone. Against task-specific RL (VTool-R1, DeepEyes), Dynamo closes 65--99% of the RL gap at a fraction of the compute, and combines additively with RL when available.

Reinforcement Learning (RL) is an important paradigm for improving the reasoning capabilities of Vision-Language Models (VLMs). However, directly applying RL to rollout multimodal reasoning can lead to instability, due to the exploitation of language priors, the neglect of visual evidence, and the generation of reasoning traces that are fluent yet not visually grounded. The question arises: Can initially steer the policy toward visually faithful reasoning regime before applying reinforcement learning? To this end, we propose a Faithful Warm-Start (FWS) strategy that first curates samples with explicit vision-language causal relationships from six general VQA benchmarks to construct the FaithfulQA dataset, where each of the image-question pairs gains a certain degree of visual observations, question requirements, commonsense knowledge, domain knowledge, and the final answer. Subsequently, a VLM-based judge is employed to further purify the dataset, ensuring strong causal consistency and visual faithfulness. This warm-start stage equips the model with the capability to understand causally grounded vision-language patterns before subsequent RL optimization under sparse answer-level rewards. Experimental results show that such faithful supervision improves answer accuracy, stabilizes RL training, and reduces visually unsupported reasoning.

Quantum computing provides a powerful paradigm for representing and transforming high-dimensional information through superposition, entanglement, and measurement-induced nonlinear features. While current quantum hardware is not yet practical for direct large-scale vision-language model (VLM) inference, simulated quantum computation can be used during model construction to generate structured parameters for compact classical AI systems. We build RiverONE, a lightweight vision-language model for quantum calibration plot understanding, using simulated quantum computation. It employs a specialized visual encoder and an InternVL-based language backbone. To compensate for compression-induced information loss, we introduce quantum-generated parameters, which are materialized as classical tensors after training. This allows RiverONE to run entirely on classical GPUs at inference time, with no quantum hardware or runtime quantum simulation. With approximately 1.9 billion parameters, RiverONE achieves at least 95% of the performance of NVIDIA Ising Calibration 1 on quantum calibration plot understanding tasks while using less than 10% of its parameter count. These results suggest that simulated quantum computation can serve as a practical construction-stage mechanism for building lightweight, knowledge-intensive scientific VLMs. Our code is available at https://github.com/THeWakeSystems/RiverOne.

Vision-Language Models (VLMs) often achieve high performance on benchmarks while remaining "black boxes", yet they remain prone to hallucination or rely on superficial shortcuts. In this work, we propose a framework designed to enhance both performance and interpretability through De-compositional Evidence Grounding. Unlike monolithic inference approaches, our approach forces the model to decompose a global query into a sequence of atomic sub-questions, each requiring an explicit sub-answer and critically a localized evidence bounding box. By grounding intermediate logical steps (e.g. identifying a container, analyzing liquid properties, and assessing environmental context) in specific visual regions, we construct a structured reasoning path that mirrors human-like deduction. This allows the final answer to emerge as a logical consequence of verified visual facts rather than a statistical guess.

Table parsing aims to convert table images into structured, machine-readable representations, a task requiring the joint perception of complex spatial layouts and textual content. While recent vision-language models (VLMs) enable end-to-end parsing, they typically rely on direct supervision of the final output, thereby bypassing the explicit intermediate reasoning that is crucial for understanding complex table structures. Furthermore, attempts to optimize these models using reinforcement learning (RL) are often hindered by unstable or ambiguous reward designs, limiting potential performance gains. To address these limitations, we propose StrucTab, a table parsing model learned through intermediate structural supervision and reward decomposition. At the modeling level, by decomposing the parsing process into human-inspired subtasks, such as row-column counting and merged-cell analysis, StrucTab progressively unifies them through a sequential reasoning strategy. At the optimization level, we introduce Uni-TabRL, a unified RL framework that leverages decomposed rewards (validity, structure, and content) to provide stable and informative optimization signals. Finally, at the evaluation level, we present TableVerse-5K, a large-scale, challenging benchmark encompassing diverse, real-world table scenarios. Extensive experiments demonstrate the state-of-the-art performance of StrucTab across all evaluated public benchmarks and significant improvements on TableVerse-5K, validating the effectiveness of explicit structural modeling and decomposed reward optimization. Code and benchmark are publicly available at https://github.com/VirtualLUOUCAS/StrucTab.

Robotic-assisted endovascular interventions demand accurate, stable, and context-aware guidewire navigation in complex and patient-specific vascular anatomies. Despite recent advances in robotic precision and learning-based control, existing autonomous navigation methods remain limited by their reliance on static reward functions and the lack of explicit procedural reasoning regarding anatomical context and task progression. To address these challenges, this paper proposes a vision-language procedural reasoning (VL-PR) framework for autonomous guidewire navigation. The framework integrates a multimodal large language model (MLLM) as a procedural reasoning module that interprets real-time visual observations to infer high-level navigation contexts. Instead of generating low-level control commands, the inferred procedural insights enable context-aware reward adaptation by dynamically adjusting the importance of reward components across different navigation phases. This approach allows a single policy to resolve competing objectives and handle complex transitions while preserving a consistent global task goal. Experiments on a physical robotic platform across diverse vascular scenarios demonstrate enhanced task reliability and streamlined navigational efficiency, highlighting the advantages over static-reward methods and offering a scalable solution for complex and multi-task robotic endovascular procedures.

Vision-Language Models (VLMs) provide powerful semantic understanding and commonsense reasoning for End-to-End Autonomous Driving (E2E-AD) planning. However, trajectories directly generated by VLMs often encode only coarse driving intentions and remain insufficient for geometrically accurate, future-aware, and multi-view-grounded planning. To address these limitations, we develop the Layer-Wise World-Model-Guided Driving framework (LWDrive). LWDrive is a VLM planning framework that refines coarse trajectories through layer-wise world-model guidance. Instead of treating the VLM output as the final trajectory, LWDrive uses it as an intent-aware coarse plan, expands a diverse candidate space around it, and progressively refines the candidates through a Foresight Cascade Planner (FCP). Specifically, we introduce future-frame generation supervision to encourage the VLM to learn forward-looking scene representations, thereby injecting planning-relevant predictive dynamics into its internal hidden states. Built upon these world-model-supervised representations, FCP exploits VLM features across multiple layers and integrates historical temporal states, Action-Query representations, and current-frame multi-view Bird's-Eye-View (BEV) features to refine candidate trajectories in a coarse-to-fine manner. This design enables progressive correction of spatial positions and motion trends while grounding trajectory refinement with multi-view scene cues and preserving the high-level driving intention produced by the large model. Finally, a score head evaluates the refined candidates and selects the best trajectory as the final planning output. Experiments show that LWDrive achieves a score of 92.0 on the NAVSIM benchmark and 89.6 on NAVSIM-v2. Code and models will be made publicly available.

Large Vision-Language Models (LVLMs) excel at multimodal tasks but remain prone to object hallucinations. Prior training-free remedies often uniformly strengthen visual signals, which may also amplify irrelevant regions and introduce spurious evidence, harming fluency. We propose Context-aware Attention Intervention (CAI), a training-free inference-time mechanism that enforces a see only when needed principle via two-axis selectivity: where to look and when to intervene. At each decoding step, CAI derives token-specific visual relevance from early-layer representations to localize semantically aligned regions, and applies a conservative, entropy- and depth-gated attention tilt only for uncertainty-spiking tokens in deeper layers where visual grounding degrades, leaving confident tokens and irrelevant regions largely unchanged. This targeted intervention strengthens visual grounding while preserving linguistic fluency, and it yields consistent improvements even without contrastive decoding, which remains optional as an auxiliary bias-suppression module. Extensive experiments across multiple LVLM backbones and benchmarks show that CAI achieves state-of-the-art hallucination mitigation, and our analysis characterizes CAI as a KL-minimal attention reweighting with bounded interference under inactive gates or small tilts. Code is available at https://github.com/Iris1946/CAI.

Large Vision-Language Models (VLMs) are increasingly deployed as content moderation tools, yet they remain vulnerable to jailbreak attacks in which harmful text is visually encoded as ASCII art. This can allow inappropriate or harmful content to bypass moderation systems. To address this vulnerability, this paper investigates how image resolution affects VLM detection of harmful ASCII art across eight character construction modes (L1-L8), ranging from dense block characters to word-embedded designs. We evaluate eight state-of-the-art VLMs on English and Chinese corpora using a pipeline that generates ASCII art images at ten resolution scales, probing whether a consistent detection-failure threshold exists across models, modes, and languages. Results indicate that detection rates decline sharply above certain resolution thresholds, and that word-based modes are the most resistant to detection across the full resolution range. These findings reveal a systematic vulnerability in VLM-based content moderation systems and motivate resolution-aware evaluation standards.

Vision-language foundation models have shown strong potential in medical image analysis. Although foundation models for ultrasound imaging have recently emerged, the domain remains particularly challenging due to severe speckle noise, acquisition variability, and subtle anatomical boundaries, leading to high inter-observer variability. Existing CLIP-based models rely primarily on global image-text alignment, limiting their sensitivity to clinically decisive local structures. We propose SonoCLIP, the first million-scale region-controllable fetal ultrasound vision-language foundation model that integrates segmentation masks as mask-channel visual prompts within the vision encoder, enabling joint global-local contrastive representation learning. To support scalable region-text alignment, we introduce a sigmoid-based pairwise contrastive loss that improves stability under large-scale supervision. We further curate a 1.44M-image multimodal fetal ultrasound dataset spanning 24 standard planes for large-scale pretraining. Extensive cross-center evaluations demonstrate that SonoCLIP achieves superior zero-shot transfer performance under both global and mask-guided inference, establishing a controllable and clinically oriented foundation model for fetal ultrasound analysis. Our code and data are available at https://github.com/Harrison-one/SonoCLIP.