Vision-Language Models

Also known as VLM

Ask about this topic

Overview

Vision-language models study how systems can connect images or videos with language to recognize, describe, retrieve, and reason about visual content. Common approaches include contrastive image-text encoders such as CLIP, generative models that connect vision encoders to language models, and instruction tuning, sometimes combined with retrieval or tools. Researchers evaluate them on image-text retrieval, visual question answering, OCR and document understanding, visual grounding, and video benchmarks. A central challenge is that fluent answers do not guarantee visual grounding: hallucinations, reliance on textual shortcuts, weak compositional and temporal reasoning, and the costs of efficient deployment remain important concerns.

Latest papers

CardsList

738 papers77 papers in the last four weeks, up 20% on the four weeks before. 0.4% of all new papers.

Mar 25, 2026

VFIG: Vectorizing Complex Figures in SVG with Vision-Language Models

Scalable Vector Graphics (SVG) are essential for technical illustration and digital design, offering resolution independence and semantic editability. In practice, original vector files are frequently lost, leaving only rasterized versions (e.g., PNG, JPEG) that resist modification, while manual reconstruction is prohibitively expensive. Progress on automating raster-to-SVG conversion has been bottlenecked by two gaps: existing SVG datasets are dominated by icons and decorative graphics that lack the complexity of professional diagrams, and existing benchmarks rely on pixel- or embedding-level similarity that fails to capture structural correctness (e.g., broken connectivity, misplaced arrows). We close both gaps with paired contributions targeting diagram-centric figures (e.g., model architectures, flowcharts, schematics). For training, we introduce VFIG-Data, the largest figure-to-SVG dataset of its kind at 66K pairs, combining real paper figures converted via a describe-and-generate pipeline with programmatic diagrams that supply noise-free supervision over arrow styles, fonts, and geometry. For evaluation, we introduce VFIG-Bench, a structure-aware evaluation suite, paired with VFIG-Bench-OOD, an out-of-distribution set of figures manually curated from highly cited arXiv papers. Beyond pixel and embedding similarity, our protocol reports rubric-based VLM-Judge scores and Elo ratings from pairwise human preference evaluation. Built on these contributions, VFIG is a VLM family trained with a simple-to-complex SFT curriculum followed by RL with rendering-aware rewards. VFIG achieves state-of-the-art open-source performance, outperforming the best open-source VLM baseline by over 30%, and matches Claude Sonnet 4.6 on VFIG-BENCH: Gemini-Judge 78.2% vs. 76.7% and GPT-Judge 87.5% vs. 87.4%. It remains slightly behind the strongest proprietary models GPT-5.2 and Gemini-3.

Qijia He, Xunmei Liu, Hammaad Memon +6

University of Washington · Allen Institute for Artificial Intelligence · UNC-Chapel Hill

Vector Graphics GenerationVision-Language ModelsImage-to-Image TranslationVLM Evaluation
AskPDF
Mar 25, 2026

Mitigating Object Hallucinations in LVLMs via Attention Imbalance Rectification

Object hallucination in Large Vision-Language Models (LVLMs) severely compromises their reliability in real-world applications, posing a critical barrier to their deployment in high-stakes scenarios such as autonomous driving and medical image analysis. Through systematic empirical investigation, we identify that the imbalanced attention allocation, both across modalities (i.e., vision and language) and within modalities (among individual tokens), exhibits a strong causal correlation with the occurrence of object hallucination. Leveraging this insight, we introduce a novel concept termed attention imbalance, which not only quantifies the degree of attention disparity but also visually delineates the underlying patterns (e.g., over-attentiveness to irrelevant language tokens or under-attentiveness to discriminative visual features) that drive object hallucination. To mitigate object hallucination, we further propose Attention Imbalance Rectification (AIR), a lightweight decoding-time intervention method that reallocates attention weights and adjusts attention distributions to rectify modality-wise and token-wise imbalances. Extensive evaluations on four mainstream LVLMs and three benchmarks (CHAIR, POPE, and MM-Vet) with seven baselines demonstrate that AIR consistently reduces object hallucination rates, achieving up to a 35.1% reduction compared to the baselines, while improving up to 15.9% of LVLMs' general capability across diverse vision-language tasks.

Han Sun, Qin Li, Peixin Wang +1

Shanghai Key Laboratory of Trustworthy Computing, East China Normal University

Object Hallucination in VLMsVLM Hallucination MitigationVision-Language ModelsLarge Vision-Language ModelsAttention Mechanisms
AskPDF
Mar 24, 2026

VISion On Request: Enhanced VLLM efficiency with sparse, dynamically selected, vision-language interactions

Existing approaches for improving the efficiency of Large Vision-Language Models (LVLMs) are largely based on the concept of visual token reduction. This approach, however, creates an information bottleneck that impairs performance, especially on challenging tasks that require fine-grained understanding and reasoning. In this work, we challenge this paradigm by introducing VISion On Request (VISOR), a method that reduces inference cost without discarding visual information. Instead of compressing the image, VISOR improves efficiency by sparsifying the interaction between image and text tokens. Specifically, the language model attends to the full set of high-resolution visual tokens through a small, strategically placed set of attention layers: general visual context is provided by efficient cross-attention between text-image, while a few well-placed and dynamically selected self-attention layers refine the visual representations themselves, enabling complex, high-resolution reasoning when needed. Based on this principle, we first train a single universal network on a range of computational budgets by varying the number of self-attention layers, and then introduce a lightweight policy mechanism that dynamically allocates visual computation based on per-sample complexity. Extensive experiments show that VISOR drastically reduces computational cost while matching or exceeding state-of-the-art results across a diverse suite of benchmarks, and excels in challenging tasks that require detailed visual understanding.

Adrian Bulat, Alberto Baldrati, Ioannis Maniadis Metaxas +2

Samsung AI Cambridge · Technical University of Iasi · Queen Mary University of London

Efficient VLM InferenceVision-Language ModelsLarge Vision-Language Models
AskPDF
Mar 24, 2026

ARGENT: Adaptive Hierarchical Image-Text Representations

Large-scale Vision-Language Models (VLMs) such as CLIP learn powerful semantic representations but operate in Euclidean space, which fails to capture the inherent hierarchical structure of visual and linguistic concepts. Hyperbolic geometry, with its exponential volume growth, offers a principled alternative for embedding such hierarchies with low distortion. However, existing hyperbolic VLMs use entailment losses that are unstable: as parent embeddings contract toward the origin, their entailment cones widen toward a half-space, causing catastrophic cone collapse that destroys the intended hierarchy. Additionally, hierarchical evaluation of these models remains unreliable, being largely retrieval-based and correlation-based metrics and prone to taxonomy dependence and ambiguous negatives. To address these limitations, we propose an adaptive entailment loss paired with a norm regularizer that prevents cone collapse without heuristic aperture clipping. We further introduce an angle-based probabilistic entailment protocol (PEP) for evaluating hierarchical understanding, scored with AUC-ROC and Average Precision. This paper introduces a stronger hyperbolic VLM baseline ARGENT, Adaptive hieRarchical imaGe-tExt represeNTation. ARGENT improves the SOTA hyperbolic VLM by 0.7, 1.1, and 0.8 absolute points on image classification, text-to-image retrieval, and proposed hierarchical metrics, respectively.

Chuong Huynh, Hossein Souri, Abhinav Kumar +3

University of Maryland, College Park · Samsung Research America, AI Center – Mountain View

Vision-Language ModelsHierarchical Representation LearningHyperbolic Representation LearningVLM Evaluation
AskPDF
Mar 24, 2026

Gaze-Regularized VLMs for Ego-Centric Behavior Understanding

Eye gaze, encompassing fixations and saccades, provides critical insights into human intentions and future actions. This study introduces a gaze-regularized framework that enhances Vision Language Models (VLMs) for egocentric behavior understanding. Unlike existing methods that rely solely on visual data and overlook gaze information, our approach directly incorporates gaze information into the VLM architecture during training. By generating gaze-based queries, the model dynamically focuses on gaze-highlighted regions, while a gaze-regularization mechanism ensures the alignment of model attention with human attention patterns. To better understand how gaze can be effectively integrated into VLMs, we conducted extensive experiments exploring various strategies for incorporating gaze data. These innovations enable the prediction of future events with detailed action descriptions. Experimental results demonstrate a nearly 13 % improvement in semantic scores compared to baseline models not leveraging gaze data, highlighting the effectiveness of our approach. This work establishes a foundation for leveraging the human gaze in VLMs, significantly boosting their predictive capabilities in applications requiring accurate and robust future event prediction.

Anupam Pani, Yanchao Yang

HKU Musketeers Foundation Institute of Data Science, The University of Hong Kong

Vision-Language ModelsEgocentric Video UnderstandingNext-Action Prediction
AskPDF
Mar 23, 2026

A Vision Language Model for Generating Procedural Plant Architecture Representations from Simulated Images

Three-dimensional (3D) procedural plant architecture models have emerged as an important tool for simulation-based studies of plant structure and function, extracting plant architectural parameters from field measurements, and for generating realistic plants in computer graphics. However, measuring the architectural parameters and nested structures for these models at the field scales remains prohibitively labor-intensive. We present a novel algorithm that generates a 3D plant architecture from an image, creating a functional structural plant model that reflects organ-level geometric and topological parameters and provides a more comprehensive representation of the plant's architecture. Instead of using 3D sensors or processing multi-view images with computer vision to obtain the 3D structure of plants, we proposed a method that generates token sequences that encode a procedural definition of plant architecture. This work used only synthetic images for training and testing, with exact architectural parameters known, allowing testing of the hypothesis that organ-level architectural parameters could be extracted from image data using a vision-language model (VLM). A synthetic dataset of cowpea plant images was generated using the Helios 3D plant simulator, with the detailed plant architecture encoded in XML files. We developed a plant architecture tokenizer for the XML file defining plant architecture, converting it into a token sequence that a language model can predict. The model achieved a token F1 score of 0.73 during teacher-forced training. Evaluation of the model was performed through autoregressive generation, achieving a BLEU-4 score of 94.00% and a ROUGE-L score of 0.5182. This led to the conclusion that such plant architecture model generation and parameter extraction were possible from synthetic images; thus, future work will extend the approach to real imagery data.

Heesup Yun, Isaac Kazuo Uyehara, Ioannis Droutsas +4

Biological and Agricultural Engineering, One Shields Avenue, Davis, 95616, CA, USA · Department of Plant Sciences, One Shields Avenue, Davis, 95616, CA, USA · Wageningen University & Research, P.O. Box 9101, 6700 HB Wageningen, The Netherlands +1

Procedural Content GenerationPlant Phenotyping3D ReconstructionVision-Language Models
AskPDF
Mar 23, 2026

Uncertainty-guided Compositional Alignment with Part-to-Whole Semantic Representativeness in Hyperbolic Vision-Language Models

While Vision-Language Models (VLMs) have achieved remarkable performance, their Euclidean embeddings remain limited in capturing hierarchical relationships such as part-to-whole or parent-child structures, and often face challenges in multi-object compositional scenarios. Hyperbolic VLMs mitigate this issue by better preserving hierarchical structures and modeling part-whole relations (i.e., whole scene and its part images) through entailment. However, existing approaches do not model that each part has a different level of semantic representativeness to the whole. We propose UNcertainty-guided Compositional Hyperbolic Alignment (UNCHA) for enhancing hyperbolic VLMs. UNCHA models part-to-whole semantic representativeness with hyperbolic uncertainty, by assigning lower uncertainty to more representative parts and higher uncertainty to less representative ones for the whole scene. This representativeness is then incorporated into the contrastive objective with uncertainty-guided weights. Finally, the uncertainty is further calibrated with an entailment loss regularized by entropy-based term. With the proposed losses, UNCHA learns hyperbolic embeddings with more accurate part-whole ordering, capturing the underlying compositional structure in an image and improving its understanding of complex multi-object scenes. UNCHA achieves state-of-the-art performance on zero-shot classification, retrieval, and multi-label classification benchmarks. Our code and models are available at: https://github.com/jeeit17/UNCHA.git.

Hayeon Kim, Ji Ha Jang, Junghun James Kim +1

Dept. of Electrical and Computer Engineering, Seoul National University, Republic of Korea · INMC & IPAI, Seoul National University, Republic of Korea

Vision-Language ModelsHierarchical Representation LearningVision-Language AlignmentHyperbolic Representation Learning
AskPDF
Mar 22, 2026

A Large-Scale Remote Sensing Dataset and VLM-based Algorithm for Fine-Grained Road Hierarchy Classification

In this work, we present SYSU-HiRoads, a large-scale hierarchical road dataset, and RoadReasoner, a vision-language-geometry framework for automatic multi-grade road mapping from remote sensing imagery. SYSU-HiRoads is built from GF-2 imagery covering 3631 km2 in Henan Province, China, and contains 1079 image tiles at 0.8 m spatial resolution. Each tile is annotated with dense road masks, vectorized centerlines, and three-level hierarchy labels, enabling the joint training and evaluation of segmentation, topology reconstruction, and hierarchy classification. Building on this dataset, RoadReasoner is designed to generate robust road surface masks, topology-preserving road networks, and semantically coherent hierarchy assignments. We strengthen road feature representation and network connectivity by explicitly enhancing frequency-sensitive cues and multi-scale context. Moreover, we perform hierarchy inference at the skeleton-segment level with geometric descriptors and geometry-aware textual prompts, queried by vision-language models to obtain linguistically interpretable grade decisions. Experiments on SYSU-HiRoads and the CHN6-CUG dataset show that RoadReasoner surpasses state-of-the-art road extraction baselines and produces accurate and semantically consistent road hierarchy maps with 72.6% OA, 64.2% F1 score, and 60.6% SegAcc. The dataset and code will be publicly released to support automated transport infrastructure mapping, road inventory updating, and broader infrastructure management applications.

Ting Han, Xiangyi Xie, Yiping Chen +5

School of Geospatial Engineering and Science, Sun Yat-Sen University, Zhuhai, 519082, China · Transport Bureau of Nanhai District, Foshan, 528200, China · Laboratory of Intelligent Collaborative Computing, University of Electronic Science and Technology of China, Chengdu, 611731, China +2

Remote Sensing VLMsClassificationRemote Sensing BenchmarksRemote Sensing Image SegmentationVision-Language ModelsTopology-Preserving Image Segmentation
AskPDF
Mar 22, 2026

Many Dialects, Many Languages, One Cultural Lens: Evaluating Multilingual VLMs for Bengali Culture Understanding Across Historically Linked Languages and Regional Dialects

Bangla culture is richly expressed through region, dialect, history, food, politics, media, and everyday visual life, yet it remains underrepresented in multimodal evaluation. To address this gap, we introduce BanglaVerse, a culturally grounded benchmark for evaluating multilingual vision-language models (VLMs) on Bengali culture across historically linked languages and regional dialects. Built from 1,152 manually curated images across nine domains, the benchmark supports visual question answering and captioning, and is expanded into four languages and five Bangla dialects, yielding ~32.2K artifacts. Our experiments show that evaluating only standard Bangla overestimates true model capability: performance drops under dialectal variation, especially for caption generation, while historically linked languages such as Hindi and Urdu retain some cultural meaning but remain weaker for structured reasoning. Across domains, the main bottleneck is missing cultural knowledge rather than visual grounding alone, with knowledge-intensive categories. These findings position BanglaVerse as a more realistic test bed for measuring culturally grounded multimodal understanding under linguistic variation.

Nurul Labib Sayeedi, Md. Faiyaz Abdullah Sayeedi, Shubhashis Roy Dipta +6

VLM EvaluationMultilingual VLM EvaluationVision-Language ModelsVLM Robustness
AskPDF
Mar 21, 2026

ME-IQA: Memory-Enhanced Image Quality Assessment via Re-Ranking

Reasoning-induced vision-language models (VLMs) advance image quality assessment (IQA) with textual reasoning, yet their scalar scores often lack sensitivity and collapse to a few values, so-called discrete collapse. We introduce ME-IQA, a plug-and-play, test-time memory-enhanced re-ranking framework. It (i) builds a memory bank and retrieves semantically and perceptually aligned neighbors using reasoning summaries, (ii) reframes the VLM as a probabilistic comparator to obtain pairwise preference probabilities and fuse this ordinal evidence with the initial score under Thurstone's Case V model, and (iii) performs gated reflection and consolidates memory to improve future decisions. This yields denser, distortion-sensitive predictions and mitigates discrete collapse. Experiments across multiple IQA benchmarks show consistent gains over strong reasoning-induced VLM baselines, existing non-reasoning IQA methods, and test-time scaling alternatives.

Kanglong Fan, Tianhe Wu, Wen Wen +6

City University of Hong Kong, Hong Kong SAR, China · ByteDance Inc., China

Memory-Augmented VLMsVision-Language ModelsImage Quality Assessment
AskPDF
Mar 21, 2026

Satellite-to-Street: Synthesizing Post-Disaster Views from Satellite Imagery via Generative Vision Models

In the immediate aftermath of natural disasters, rapid situational awareness is critical. Traditionally, satellite observations are widely used to estimate damage extent. However, they lack the ground-level perspective essential for characterizing specific structural failures and impacts. Meanwhile, ground-level data (e.g., street-view imagery) remains largely inaccessible during time-sensitive events. This study investigates Satellite-to-Street View Synthesis to bridge this data gap. We introduce two generative strategies to synthesize post-disaster street views from satellite imagery: a Vision-Language Model (VLM)-guided approach and a damage-sensitive Mixture-of-Experts (MoE) method. We benchmark these against general-purpose baselines (Pix2Pix, ControlNet) using a proposed Structure-Aware Evaluation Framework. This multi-tier protocol integrates (1) pixel-level quality assessment, (2) ResNet-based semantic consistency verification, and (3) a novel VLM-as-a-Judge for perceptual alignment. Experiments on 300 disaster scenarios reveal a critical realism--fidelity trade-off: while diffusion-based approaches (e.g., ControlNet) achieve high perceptual realism, they often hallucinate structural details. Quantitative results show that standard ControlNet achieves the highest semantic accuracy, 0.71, whereas VLM-enhanced and MoE models excel in textural plausibility but struggle with semantic clarity. This work establishes a baseline for trustworthy cross-view synthesis, emphasizing that visually realistic generations may still fail to preserve critical structural information required for reliable disaster assessment.

Yifan Yang, Lei Zou, Wendy Jepson

Department of Geography, Texas A&M University, College Station, TX, USA

Image GenerationDisaster Damage AssessmentVision-Language Models
AskPDF
Mar 20, 2026

PEARL: Personalized Streaming Video Understanding Model

Human cognition of new concepts is inherently a streaming process: we continuously recognize new objects or identities and update our memories over time. However, current multimodal personalization methods are largely limited to static images or offline videos. This disconnects continuous visual input from instant real-world feedback, limiting their ability to provide the real-time, interactive personalized responses essential for future AI assistants. To bridge this gap, we first propose and formally define the novel task of Personalized Streaming Video Understanding (PSVU). To facilitate research in this new direction, we introduce PEARL-Bench, the first comprehensive benchmark designed specifically to evaluate this challenging setting. It evaluates a model's ability to respond to personalized concepts at exact timestamps under two modes: (1) Frame-level, focusing on a specific person or object in discrete frames, and (2) a novel Video-level, focusing on personalized actions unfolding across continuous frames. PEARL-Bench comprises 132 unique videos and 2,173 fine-grained annotations with precise timestamps. Concept diversity and annotation quality are strictly ensured through a combined pipeline of automated generation and human verification. To tackle this challenging new setting, we further propose PEARL, a plug-and-play, training-free strategy that serves as a strong baseline. Extensive evaluations across 8 offline and online models demonstrate that PEARL achieves state-of-the-art performance. Notably, it brings consistent PSVU improvements when applied to 3 distinct architectures, proving to be a highly effective and robust strategy. We hope this work advances vision-language model (VLM) personalization and inspires further research into streaming personalized AI assistants. Code is available at https://github.com/Yuanhong-Zheng/PEARL.

Yuanhong Zheng, Ruichuan An, Xiaopeng Lin +10

Peking University · CASIA · Stepfun +3

Streaming Video UnderstandingVision-Language Models
AskPDF
Mar 20, 2026

Contextual inference from single objects in Vision-Language models

How much scene context a single object carries is a well-studied question in human scene perception, yet how this capacity is organized in vision-language models (VLMs) remains poorly understood, with direct implications for the robustness of these models. We investigate this question through a systematic behavioral and mechanistic analysis of contextual inference from single objects. Presenting VLMs with single objects on masked backgrounds, we probe their ability to infer both fine-grained scene category and coarse superordinate context (indoor vs. outdoor). We found that single objects support above-chance inference at both levels, with performance modulated by the same object properties that predict human scene categorization. Object identity, scene, and superordinate predictions are partially dissociable: accurate inference at one level neither requires nor guarantees accurate inference at the others, and the degree of coupling differs markedly across models. Mechanistically, object representations that remain stable when background context is removed are more predictive of successful contextual inference. Scene and superordinate schemas are grounded in fundamentally different ways: scene identity is encoded in image tokens throughout the network, while superordinate information emerges only late or not at all. Together, these results reveal that the organization of contextual inference in VLMs is more complex than accuracy alone suggests, with behavioral and mechanistic signatures

Martina G. Vilas, Timothy Schaumlöffel, Gemma Roig

Goethe University Frankfurt · The Hessian Center for AI

Vision-Language ModelsVLM ReasoningVLM Interpretability
AskPDF
Mar 20, 2026

Geometric Risk Control for Vision-Language Model OCR

Vision-language models (VLMs) enable flexible generative optical character recognition (OCR), while their open-ended decoders can expose wrong but fluent text with weak visual support. In audit-sensitive records, such an output can be more costly than abstention. Frozen or externally served VLMs therefore require an external decision layer that can determine whether a transcription has sufficient visual evidence for release. We introduce the Geometric Risk Controller (GRC), a model-agnostic controller that treats controlled geometric transformations as repeatable black-box probes, screens structurally implausible continuations, and releases the unique candidate supported by coherent cross-view evidence. The protocol provides empirical selective exposure control with explicit coverage and query cost under a reproducible fixed decision rule. Experiments across frozen VLMs and standard scene-text benchmarks consistently reduce mean, upper-tail, and catastrophic error among released outputs while retaining high coverage.

Weile Gong, Zijian Lu, Mingcai Chen +3

Nanjing University of Posts and Telecommunications, Nanjing, China

Selective Prediction for VLMsVLM Hallucination MitigationOptical Character RecognitionVLM RobustnessVision-Language Models
AskPDF
Mar 19, 2026

dinov3.seg: Open-Vocabulary Semantic Segmentation with DINOv3

Open-Vocabulary Semantic Segmentation (OVSS) assigns pixel-level labels from an open set of text-defined categories, demanding reliable generalization to unseen classes at inference. Although modern vision-language models (VLMs) support strong open-vocabulary recognition, their representations learned through global contrastive objectives remain suboptimal for dense prediction, prompting many OVSS methods to depend on limited adaptation or refinement of image-text similarity maps. This, in turn, restricts spatial precision and robustness in complex, cluttered scenes. We introduce dinov3.seg, extending dinov3.txt into a dedicated framework for OVSS. Our contributions are four-fold. First, we design a task-specific architecture tailored to this backbone, systematically adapting established design principles from prior open-vocabulary segmentation work. Second, we jointly leverage text embeddings aligned with both the global [CLS] token and local patch-level visual features from ViT-based encoder, effectively combining semantic discrimination with fine-grained spatial locality. Third, unlike prior approaches that rely primarily on post hoc similarity refinement, we perform early refinement of visual representations prior to image-text interaction, followed by late refinement of the resulting image-text correlation features, enabling more accurate and robust dense predictions in cluttered scenes. Finally, we propose a high-resolution local-global inference strategy based on sliding-window aggregation, which preserves spatial detail while maintaining global context. We conduct extensive experiments on five widely adopted OVSS benchmarks to evaluate our approach. The results demonstrate its effectiveness and robustness, consistently outperforming current state-of-the-art methods.

Saikat Dutta, Biplab Banerjee, Hamid Rezatofighi

IITB-Monash Research Academy · IIT Bombay · Monash University

Open-Vocabulary Semantic SegmentationVision-Language Models
AskPDF
Mar 19, 2026

Do VLMs Need Vision Transformers? Evaluating State Space Models as Vision Encoders

Large vision--language models (VLMs) often use a frozen vision backbone, whose image features are mapped into a large language model through a lightweight connector. While transformer-based encoders are the standard visual backbone, we ask whether state space model (SSM) vision backbones can be a strong alternative. We systematically evaluate SSM vision backbones for VLMs in a controlled setting. Under matched ImageNet-1K initialization, the SSM backbone achieves the strongest overall performance across both VQA and grounding/localization. We further adapt both SSM and ViT-family backbones with detection or segmentation training and find that dense-task tuning generally improves performance across families; after this adaptation, the SSM backbone remains competitive while operating at a substantially smaller model scale. We further observe that (i) higher ImageNet accuracy or larger backbones do not reliably translate into better VLM performance, and (ii) some visual backbones are unstable in localization. Based on these findings, we propose stabilization strategies that improve robustness for both backbone families and highlight SSM backbones as a strong alternative to transformer-based vision encoders in VLMs.

Shang-Jui Ray Kuo, Paola Cascante-Bonilla

Stony Brook University

Vision-Language ModelsState Space ModelsVLM EvaluationVLM Robustness
AskPDF
Mar 19, 2026

Tinted Frames: Question Framing Blinds Vision-Language Models

Vision-Language Models (VLMs) have been shown to be blind, often underutilizing their visual inputs even on tasks that require visual reasoning. In this work, we demonstrate that VLMs are selectively blind. They modulate the amount of attention applied to visual inputs based on linguistic framing even when alternative framings demand identical visual reasoning. Using visual attention as a probe, we quantify how framing alters both the amount and distribution of attention over the image. Constrained framings, such as multiple choice and yes/no, induce substantially lower attention to image context compared to open-ended, reduce focus on task-relevant regions, and shift attention towards uninformative tokens. We further demonstrate that this attention misallocation is the principal cause of degraded accuracy and cross-framing inconsistency. Building on this mechanistic insight, we introduce a lightweight prompt-tuning method using learnable tokens that encourages the robust, visually grounded attention patterns observed in open-ended settings, improving visual grounding and improving performance across framings.

Wan-Cyuan Fan, Jiayun Luo, Declan Kutscher +2

University of British Columbia · Vector Institute for AI · University of California, Berkeley

Vision-Language ModelsVLM RobustnessVisual Evidence Sensitivity in VLMsReasoning for VLMsVision-Language Grounding
AskPDF
Mar 19, 2026

SAVeS: Steering Safety Judgments in Vision-Language Models via Semantic Cues

Vision-language models (VLMs) are increasingly deployed in real-world and embodied settings where safety decisions depend on visual context. However, it remains unclear which visual evidence drives these judgments. We study whether multimodal safety behavior in VLMs can be steered by simple semantic cues. We introduce a semantic steering framework that applies controlled textual, visual, and cognitive interventions without changing the underlying scene content. To evaluate these effects, we propose SAVeS, a benchmark for situational safety under semantic cues, together with an evaluation protocol that separates behavioral refusal, grounded safety reasoning, and false refusals. Experiments across multiple VLMs and an additional state-of-the-art benchmark show that safety decisions are highly sensitive to semantic cues, indicating reliance on learned visual-linguistic associations rather than grounded visual understanding. We further demonstrate that automated steering pipelines can exploit these mechanisms, highlighting a potential vulnerability in multimodal safety systems.

Carlos Hinojosa, Clemens Grange, Bernard Ghanem

King Abdullah University of Science and Technology (KAUST), Saudi Arabia · Technical University of Munich (TUM), Germany

Cultural Safety Evaluation of VLMsVision-Language ModelsVLM EvaluationVLM Robustness
AskPDF
Mar 19, 2026

SEM: Sparse Embedding Modulation for Post-Hoc Debiasing of Vision-Language Models

Models that bridge vision and language, such as CLIP, are key components of multimodal AI, yet their large-scale, uncurated training data introduce severe social and spurious biases. Existing post-hoc debiasing methods often operate directly in the dense CLIP embedding space, where bias and task-relevant information are highly entangled. This entanglement limits their ability to remove bias without degrading semantic fidelity. In this work, we propose Sparse Embedding Modulation (SEM), a post-hoc, zero-shot debiasing framework that operates in a Sparse Autoencoder (SAE) latent space. By decomposing CLIP text embeddings into disentangled features, SEM identifies and modulates bias-relevant neurons while preserving query-relevant ones. This enables more precise, non-linear interventions. Across four benchmark datasets and two CLIP backbones, SEM achieves substantial fairness gains in retrieval and zero-shot classification. Our results demonstrate that sparse latent representations provide an effective foundation for post-hoc debiasing of vision-language models.

Quentin Guimard, Federico Bartsch, Simone Caldarella +3

University of Trento · Toyota Motor Europe · Fondazione Bruno Kessler

Cultural Bias in VLMsVision-Language Models
AskPDF
Mar 19, 2026

Balanced Thinking: Improving Chain of Thought Training in Vision Language Models

Multimodal reasoning in vision-language models (VLMs) typically relies on a two-stage process: supervised fine-tuning (SFT) and reinforcement learning (RL). In standard SFT, all tokens contribute equally to the loss, even though reasoning data are inherently token-imbalanced. Long <think> traces overshadow short but task-critical <answer> segments, leading to verbose reasoning and inaccurate answers. We propose SCALe (Scheduled Curriculum Adaptive Loss), which explicitly separates supervision over reasoning and answer segments using dynamic, length-independent weighting. Unlike vanilla SFT, which overweights the <think> segment, SCALe-SFT gradually shifts the focus from <think> to <answer> throughout training via a cosine scheduling policy, encouraging concise and well-grounded reasoning. We evaluate SCALe across diverse benchmarks and architectures. Results show that SCALe consistently improves accuracy over vanilla SFT and matches the performance of the full two-phase SFT + GRPO pipeline while requiring only about one-seventh of the training time, making it a lightweight yet effective alternative. When combined with GRPO, SCALe achieves the best overall performance, highlighting its value both as a standalone method and as a strong foundation for reinforcement refinement.

Shaked Perek, Ben Wiesel, Avihu Dekel +2

IBM Research

PEFT of VLMsVLM ReasoningVision-Language ModelsSupervised Fine-TuningCurriculum Learning
AskPDF
Mar 19, 2026

Complementary Text-Guided Attention for Zero-Shot Adversarial Robustness

Due to the impressive zero-shot capabilities, pre-trained vision-language models (e.g., CLIP), have attracted widespread attention and adoption across various domains. Nonetheless, CLIP has been observed to be susceptible to adversarial examples. Through experimental analysis, we have observed a phenomenon wherein adversarial perturbations induce shifts in text-guided attention. Building upon this observation, we propose a simple yet effective strategy: Text-Guided Attention for Zero-Shot Robustness (TGA-ZSR). This framework incorporates two components: Local Attention Refinement Module and Global Attention Constraint Module. Our goal is to maintain the generalization of the CLIP model and enhance its adversarial robustness. Additionally, the Global Attention Constraint Module acquires text-guided attention from both the target and original models using clean examples. Its objective is to maintain model performance on clean samples while enhancing overall robustness. However, we observe that the method occasionally focuses on irrelevant or spurious features, which can lead to suboptimal performance and undermine its robustness in certain scenarios. To overcome this limitation, we further propose a novel approach called Complementary Text-Guided Attention (Comp-TGA). This method integrates two types of foreground attention: attention guided by the class prompt and reversed attention driven by the non-class prompt. These complementary attention mechanisms allow the model to capture a more comprehensive and accurate representation of the foreground. The experiments validate that TGA-ZSR and Comp-TGA yield 9.58% and 11.95% improvements respectively, in zero-shot robust accuracy over the current state-of-the-art techniques across 16 datasets.

Lu Yu, Haiyang Zhang, Changsheng Xu

School of Computer Science and Engineering, Tianjin University of Technology, 300384, China · State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China · School of Artificial Intelligence, University of the Chinese Academy of Sciences, Beijing, 100049, China

Attention-Based Adversarial AttacksContrastive Learning for VLMsVLM RobustnessAdversarial RobustnessVision-Language Models
AskPDF
Mar 19, 2026

To See or To Please: Uncovering Visual Sycophancy and Split Beliefs in VLMs

When VLMs answer correctly, do they genuinely rely on visual information? We introduce a Tri-Layer Diagnostic Framework with three per-sample metrics: Latent Anomaly Detection, Visual Necessity Score, and Competition Score, which disentangle perception, dependency, and alignment failures. Across 9 VLMs and 9,000 model-sample pairs under counterfactual blind, noise, and conflict interventions, 72.9% of samples exhibit Visual Sycophancy, a Split Beliefs pattern in which internal evidence is preserved yet a hallucinated answer is decoded, while zero samples show Robust Refusal, indicating that current alignment training has eliminated refusal as a decoding outcome. Scaling within the Qwen-VL family, both within- and across-generation, monotonically reduces Language Shortcuts but amplifies Visual Sycophancy, showing that scale and newer post-training alone cannot resolve the grounding problem. Diagnostic scores further enable a training-free selective-prediction strategy yielding up to +9.5 percentage points accuracy at 50% coverage.

Rui Hong, Shuxue Quan

George Mason University, Fairfax, VA, USA · Independent Researcher

Selective Prediction for VLMsSycophancy in VLMsVision-Language ModelsVision-Language GroundingVisual Evidence Sensitivity in VLMsHallucination Detection in VLMs+1
AskPDF
Mar 18, 2026

CycleCap: Improving VLMs Captioning Performance via Self-Supervised Cycle Consistency Fine-Tuning

Visual-Language Models (VLMs) have achieved remarkable progress in image captioning, visual question answering, and visual reasoning. Yet they remain prone to vision-language misalignment, often producing overly generic or hallucinated descriptions. Existing approaches address this via instruction tuning-requiring costly, large-scale annotated datasets or via complex test-time frameworks for caption refinement. In this work, we revisit image-text alignment through the lens of cycle consistency: given an image and a caption generated by an image-to-text model, the backward mapping through a text-to-image model should reconstruct an image that closely matches the original. In our setup, a VLM serves as the image-to-text component, while a pre-trained text-to-image model closes the loop by reconstructing the image from the generated caption. Building on this, we introduce CycleCap, a fine-tuning scheme to improve image captioning using Group Relative Policy Optimization (GRPO) with a reward based on the similarity between the original and reconstructed images, computed on-the-fly. Unlike previous work that uses cycle consistency loss for preference dataset construction, our method leverages cycle consistency directly as a self-supervised training signal. This enables the use of raw images alone, eliminating the need for curated image-text datasets, while steering the VLM to produce more accurate and grounded text descriptions. Applied to four VLMs ranging from 1B to 7B parameters, CycleCap yields consistent improvements across captioning and hallucination benchmarks, surpassing state-of-the-art methods that rely on supervised cycle consistency training.

Marios Krestenitis, Christos Tzelepis, Konstantinos Ioannidis +5

Queen Mary, University of London · Centre for Research and Technology Hellas · City St George’s, University of London

Cycle ConsistencyImage CaptioningVision-Language ModelsVision-Language AlignmentVLM Hallucination Mitigation
AskPDF
Mar 18, 2026

VLM-AutoDrive: Post-Training Vision-Language Models for Safety-Critical Autonomous Driving Events

The rapid growth of ego-centric dashcam footage presents a major challenge for detecting safety-critical events such as collisions and near-collisions, scenarios that are brief, rare, and difficult for generic vision models to capture. While multimodal large language models (MLLMs) demonstrate strong general reasoning ability, they underperform in driving contexts due to domain and temporal misalignment. We introduce VLM-AutoDrive, a modular post-training framework for adapting pretrained Vision-Language Models (VLMs) to high-fidelity anomaly detection. The framework integrates metadata-derived captions, LLM-generated descriptions, visual question answering (VQA) pairs, and chain-of-thought (CoT) reasoning supervision to enable domain-aligned and interpretable learning. Off-the-shelf VLMs such as NVIDIA's Cosmos-Reason1 7B (CR1) exhibit near-zero Collision recall in zero-shot settings; fine-tuning with VLM-AutoDrive improves Collision F1 from 0.00 to 0.69 and overall accuracy from 35.35% to 77.27%. VLM-AutoDrive offers a scalable recipe for adapting general-purpose VLMs to safety-critical, temporally localized perception tasks. Evaluated on real-world Nexar dashcam videos, it achieves substantial gains in Collision and Near-Collision detection while producing interpretable reasoning traces, bridging the gap between perception, causality, and decision reasoning in autonomous driving.

Mohammad Qazim Bhat, Yufan Huang, Niket Agarwal +7

NVIDIA, California, USA

VLM Post-TrainingVLMs for Autonomous DrivingVision-Language ModelsAutonomous Driving PerceptionVideo Anomaly DetectionTraffic Accident Anticipation
AskPDF
Mar 18, 2026

Learning Transferable Temporal Primitives for Video Reasoning via Synthetic Videos

The transition from image to video understanding requires vision-language models (VLMs) to shift from recognizing static patterns to reasoning over temporal dynamics such as motion trajectories, speed changes, and state transitions. Yet current post-training methods fall short due to two critical limitations: (1) existing datasets often lack temporal-centricity, where answers can be inferred from isolated keyframes rather than requiring holistic temporal integration; and (2) training data generated by proprietary models contains systematic errors in fundamental temporal perception, such as confusing motion directions or misjudging speeds. We introduce SynRL, a post-training framework that teaches models temporal primitives, the fundamental building blocks of temporal understanding including direction, speed, and state tracking. Our key insight is that these abstract primitives, learned from programmatically generated synthetic videos, transfer effectively to real-world scenarios. We decompose temporal understanding into short-term perceptual primitives (speed, direction) and long-term cognitive primitives, constructing 7.7K CoT and 7K RL samples with ground-truth frame-level annotations through code-based video generation. Despite training on simple geometric shapes, SynRL achieves substantial improvements across 15 benchmarks spanning temporal grounding, complex reasoning, and general video understanding. Remarkably, our 7.7K synthetic CoT samples outperform Video-R1 with 165K real-world samples. We attribute this to fundamental temporal skills, such as tracking frame by frame changes and comparing velocity, that transfer effectively from abstract synthetic patterns to complex real-world scenarios. This establishes a new paradigm for video post-training: video temporal learning through carefully designed synthetic data provides a more cost efficient scaling path.

Songtao Jiang, Sibo Song, Chenyi Zhou +12

Zhejiang University · Qwen Team, Alibaba Group · Shanghai Jiao Tong University +1

VLM Post-TrainingTemporal Video UnderstandingVision-Language ModelsTemporal Video Grounding
AskPDF
Mar 18, 2026

VLM2Rec: Resolving Modality Collapse in Vision-Language Model Embedders for Multimodal Sequential Recommendation

Sequential Recommendation (SR) in multimodal settings typically relies on small frozen pretrained encoders, which limits semantic capacity and prevents Collaborative Filtering (CF) signals from being fully integrated into item representations. Inspired by the recent success of Large Language Models (LLMs) as high-capacity embedders, we investigate the use of Vision-Language Models (VLMs) as CF-aware multimodal embedders for SR. However, we find that standard contrastive Supervised Fine-Tuning (SFT), used to adapt VLMs for embedding generation and inject CF signals, can amplify inherent modality imbalance: optimization becomes dominated by one modality while the other degrades, ultimately undermining recommendation accuracy. To address this, we propose VLM2Rec, a VLM embedder-based framework for multimodal sequential recommendation designed to promote balanced modality utilization. Specifically, we introduce Weak-modality Penalized Contrastive Learning to mitigate gradient imbalance during optimization and cross-modal Relational Topology Regularization to preserve geometric consistency between modalities. Experiments show that VLM2Rec consistently improves over strong baselines in both accuracy and robustness across diverse scenarios.

Junyoung Kim, Woojoo Kim, Wonbin Kweon +3

Pohang University of Science and Technology · University of Illinois Urbana-Champaign

Multimodal RecommendationMultimodal EmbeddingVision-Language ModelsMultimodal Contrastive Learning
AskPDF
Mar 18, 2026

FMMC: Harnessing the Power of Foundation Models for Accurate Material Classification

Material classification has emerged as a critical task in computer vision and graphics, supporting the assignment of accurate material properties to a wide range of digital and real-world applications. While traditionally framed as an image classification task, this domain faces significant challenges due to the scarcity of annotated data, limiting the accuracy and generalizability of trained models. Recent advances in vision-language foundation models (VLMs) offer promising avenues to address these issues, yet existing solutions leveraging these models still exhibit unsatisfying results in material recognition tasks. In this work, we propose a novel framework that effectively harnesses foundation models to overcome data limitations and enhance classification accuracy. Our method integrates two key innovations: (a) a robust image generation and auto-labeling pipeline that creates a diverse and high-quality training dataset with material-centric images, and automatically assigns labels by fusing object semantics and material attributes in text prompts; (b) a prior incorporation strategy to distill information from VLMs, combined with a joint fine-tuning method that optimizes a pre-trained vision foundation model alongside VLM-derived priors, preserving broad generalizability while adapting to material-specific features. Extensive experiments demonstrate significant improvements on multiple datasets. We show that our synthetic dataset effectively captures the characteristics of real world materials, and the integration of priors from vision-language models significantly enhances the final performance. The source code and dataset will be released.

Qingran Lin, Fengwei Yang, Chaolun Zhu

Georgia Institute of Technology · Duke University · Waseda University

Vision-Language ModelsSynthetic Data GenerationVision Foundation Model AdaptationImage Classification
AskPDF
Mar 17, 2026

Hidden Clones: Exposing and Fixing Family Bias in Vision-Language Model Ensembles

Ensembling Vision-Language Models (VLMs) from different providers maximizes benchmark accuracy, yet models from the same architectural family share correlated errors that standard voting ignores. We study this structure across 17 VLMs from 8 families on VQAv2, TextVQA, and GQA. Family-correlated errors reduce effective ensemble dimensionality to 2.5-3.6 independent voters and create a Misleading tier (1.5-6.5% of questions) where correlated majority errors destroy accuracy to 0% despite the best model being correct. We propose three family-aware methods. Hierarchical Family Voting (HFV) aggregates within families before voting across them, recovering +18-26 pp on the Misleading tier. QualRCCV, a training-free method weighting models by calibration, family quality, and inverse family size, is the first to beat calibrated voting on all three benchmarks (p<0.05). Learned Candidate Scoring (LCS) trains a cross-validated classifier to re-rank candidate answers using support breadth, family diversity, and model quality, achieving the largest gains: +0.68% VQAv2, +0.61% TextVQA, +2.45% GQA -- all significant -- and is the only learned method that never degrades any benchmark. On VQAv2 test-standard (EvalAI), LCS reaches 87.83% with 12 models, confirming generalization.

Zacharie Bugaud

Astera Institute

Vision-Language ModelsVisual Question Answering
AskPDF
Mar 17, 2026

Empirical Recipes for Efficient and Compact Vision-Language Models

Deploying vision-language models (VLMs) in resource-constrained settings demands low latency and high throughput, yet existing compact VLMs often fall short of the inference speedups their smaller parameter counts suggest. To explain this discrepancy, we conduct an empirical end-to-end efficiency analysis and systematically profile inference to identify the dominant bottlenecks. Based on these findings, we develop optimization recipes tailored to compact VLMs that substantially reduce latency while preserving accuracy. These techniques cut time to first token (TTFT) by 53% on InternVL3-2B and by 93% on SmolVLM-256M. Our recipes are broadly applicable across both VLM architectures and common serving frameworks, providing practical guidance for building efficient VLM systems. Beyond efficiency, we study how to extend compact VLMs with structured perception outputs and introduce the resulting model family, ArgusVLM. Across diverse benchmarks, ArgusVLM achieves strong performance while maintaining a compact and efficient design.

Jiabo Huang, Zhizhong Li, Sina Sajadmanesh +2

Sony AI.

Vision-Language ModelsEfficient VLM Inference
AskPDF
Mar 17, 2026

How to Utilize Complementary Vision-Text Information for 2D Structure Understanding

LLMs typically linearize 2D tables into 1D sequences to fit their autoregressive architecture, which weakens row-column adjacency and other layout cues. In contrast, purely visual encoders can capture spatial cues, yet often struggle to preserve exact cell text. Our analysis reveals that these two modalities provide highly distinct information to LLMs and exhibit strong complementarity. However, direct concatenation and other fusion methods yield limited gains and frequently introduce cross-modal interference. To address this issue, we propose DiVA-Former, a lightweight architecture designed to effectively integrate vision and text information. DiVA-Former leverages visual tokens as dynamic queries to distill long textual sequences into digest vectors, thereby effectively exploiting complementary vision--text information. Evaluated across 13 table benchmarks, DiVA-Former improves upon the pure-text baseline by 23.9% and achieves consistent gains over existing baselines using visual inputs, textual inputs, or a combination of both.

Jiancheng Dong, Pengyue Jia, Derong Xu +9

City University of Hong Kong · Baidu Inc.

Document UnderstandingVision-Language ModelsMultimodal Fusion
AskPDF