VLM Evaluation

VLM: Vision-Language Model

Ask about this topic

Overview

Vision-language model evaluation studies what these systems can infer from images and video, and whether their answers are supported by visual evidence. Researchers use broad benchmarks alongside controlled probes, counterfactual edits, and stage-wise tests covering tasks such as spatial reasoning, charts and documents, medical imaging, and long-video understanding. They assess not only accuracy but also grounding, hallucination, calibration, robustness, bias, and abstention, often comparing models with human or specialist judgments. A central challenge is that high benchmark scores can reflect language priors or shortcuts rather than reliable visual understanding, and performance often falters on fine-grained, temporal, compositional, or unfamiliar tasks.

Latest papers

CardsList

999 papers122 papers in the last four weeks, up 12% on the four weeks before. 0.7% of all new papers.

Mar 10, 2026

OddGridBench: Exposing the Lack of Fine-Grained Visual Discrepancy Sensitivity in Multimodal Large Language Models

Multimodal large language models (MLLMs) have achieved remarkable performance across a wide range of vision language tasks. However, their ability in low-level visual perception, particularly in detecting fine-grained visual discrepancies, remains underexplored and lacks systematic analysis. In this work, we introduce OddGridBench, a controllable benchmark for evaluating the visual discrepancy sensitivity of MLLMs. OddGridBench comprises over 1,400 grid-based images, where a single element differs from all others by one or multiple visual attributes such as color, size, rotation, or position. Experiments reveal that all evaluated MLLMs, including open-source families such as Qwen3-VL and InternVL3.5, and proprietary systems like Gemini-2.5-Pro and GPT-5, perform far below human levels in visual discrepancy detection. We further propose OddGrid-GRPO, a reinforcement learning framework that integrates curriculum learning and distance-aware reward. By progressively controlling the difficulty of training samples and incorporating spatial proximity constraints into the reward design, OddGrid-GRPO significantly enhances the model's fine-grained visual discrimination ability. We hope OddGridBench and OddGrid-GRPO will lay the groundwork for advancing perceptual grounding and visual discrepancy sensitivity in multimodal intelligence. Code and dataset are available at https://wwwtttjjj.github.io/OddGridBench/.

Tengjin Weng, Wenhao Jiang, Jingyi Wang +3

College of Computer Science and Software Engineering, Shenzhen University · Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ) · Tsinghua Shenzhen International Graduate School +2

Fine-Grained Visual PerceptionMultimodal Large Language ModelsVLM EvaluationRL for VLMs
AskPDF
Mar 9, 2026

Using Vision Language Foundation Models to Generate Plant Simulation Configurations via In-Context Learning

This paper introduces a benchmark for evaluating whether vision-language models (VLMs) can generate plant simulation configurations from imagery using in-context learning. We study this benchmark for cowpea plot reconstruction for plant simulations, where the VLM needs to generate structured JSON configurations that include field and plant information. Open-source multimodal models from Gemma 4 and Qwen3.5 families are evaluated on a synthetic cowpea dataset with known JSON ground truth and on a real drone orthophoto dataset with field-collected JSON. Five in-context learning methods are used, from format restriction instruction to few-shot image examples with auxiliary grounding information. The results show that VLMs can generate valid JSON outputs, can generally estimate days after planting (DAP), plant counts, plant locations, sun angles, and leaf chlorophyll content, and can render approximate simulations of cowpea plots. Error metrics fluctuate across model families and often remain worse than dataset baselines, particularly when VLMs' pretrained knowledge dominates over weak visual evidence. These results position image-to-simulation JSON generation as a promising but currently challenging task, and establish a benchmark for studying how multimodal reasoning, prompt design, and the sim-to-real domain gap affect plant phenotyping tasks.

Heesup Yun, Isaac Kazuo Uyehara, Earl Ranario +4

University of California, Davis

VLM EvaluationAgricultural Image AnalysisPlant PhenotypingSim-to-Real TransferIn-Context Learning
AskPDF
Mar 9, 2026

MEGC2026: Micro-Expression Grand Challenge on Visual Question Answering

Facial micro-expressions (MEs) are involuntary movements of the face that occur spontaneously when a person experiences an emotion but attempts to suppress or repress the facial expression, typically found in a high-stakes environment. In recent years, substantial advancements have been made in the areas of ME recognition, spotting, and generation. The emergence of multimodal large language models (MLLMs) and large vision-language models (LVLMs) offers promising new avenues for enhancing ME analysis through their powerful multimodal reasoning capabilities. The ME grand challenge (MEGC) 2026 introduces two tasks that reflect these evolving research directions: (1) ME video question answering (ME-VQA), which explores ME understanding through visual question answering on relatively short video sequences, leveraging MLLMs or LVLMs to address diverse question types related to MEs; and (2) ME long-video question answering (ME-LVQA), which extends VQA to long-duration video sequences in realistic settings, requiring models to handle temporal reasoning and subtle micro-expression detection across extended time periods. All participating algorithms are required to submit their results on a public leaderboard. More details are available at https://megc2026.github.io.

Xinqi Fan, Jingting Li, John See +3

Department of Computing and Mathematics, Manchester Metropolitan University · State Key Laboratory of Cognitive Science and Mental Health, Institute of Psychology, CAS & Department of Psychology, University of the Chinese Academy of Sciences · School of Mathematical and Computer Sciences, Heriot-Watt University Malaysia

Micro-Expression RecognitionVideo QAMultimodal QAFacial Expression RecognitionVLM EvaluationTemporal Video Understanding
AskPDF
Mar 9, 2026

Reading ≠\neq Seeing: Diagnosing and Closing the Typography Gap in Vision-Language Models

Vision-Language Models achieve near-perfect accuracy at reading text in images, yet prove largely typography-blind: capable of recognizing what text says, but not how it looks. We systematically investigate this gap by evaluating font family, size, style, and color recognition across 26 fonts, four scripts, and three difficulty levels. Our evaluation of 15 state-of-the-art VLMs reveals a striking perception hierarchy: color recognition is near-perfect, yet font style detection remains universally poor. We further find that model scale fails to predict performance and that accuracy is uniform across difficulty levels, together pointing to a training-data omission rather than a capacity ceiling. LoRA fine-tuning on a small set of synthetic samples substantially improves an open-source model, narrowing the gap to the best closed-source system and surpassing it on font size recognition. Font style alone remains resistant to fine-tuning, suggesting that relational visual reasoning may require architectural innovation beyond current patch-based encoders. We release our evaluation framework, data, and fine-tuning recipe to support progress in closing the typographic gap in vision-language understanding.

Heng Zhou, Ao Yu, Li Kang +5

Vision-Language ModelsVLM EvaluationVLM Adaptation
AskPDF
Mar 9, 2026

Can Vision-Language Models Solve the Shell Game?

Visual entity tracking is an innate cognitive ability in humans, yet it remains a critical bottleneck for Vision-Language Models (VLMs). This deficit is often obscured in existing video benchmarks by visual shortcuts. We introduce VET-Bench, a synthetic diagnostic testbed featuring visually identical objects that necessitate tracking exclusively through spatiotemporal continuity. Our experiments reveal that current state-of-the-art VLMs perform at or near chance level on VET-Bench, exposing a fundamental limitation: an over-reliance on static frame-level features and a failure to maintain entity representations over time. We provide a theoretical analysis drawing connections to the state-tracking problem, proving that fixed-depth transformer-based VLMs are fundamentally limited in tracking indistinguishable objects without intermediate supervision due to expressivity constraints. To address this, we propose Spatiotemporal Grounded Chain-of-Thought (SGCoT): generating object trajectories as explicit intermediate states. Leveraging Molmo2's object tracking ability, we elicit SGCoT reasoning by fine-tuning on synthesized text-only data for alignment. Our method achieves state-of-the-art accuracy exceeding 90% on VET-Bench, demonstrating that VLMs can reliably solve the video shell-game task end-to-end without external tools. Our code and data are available at https://vetbench.github.io .

Tiedong Liu, Wee Sun Lee

National University of Singapore

Visual Object TrackingTemporal Video UnderstandingVLM EvaluationState Tracking
AskPDF
Mar 9, 2026

Text to Automata Diagrams: Comparing TikZ Code Generation with Direct Image Synthesis

Diagrams are widely used in teaching computer science courses. They are useful in subjects such as automata and formal languages, data structures, etc. These diagrams, often drawn by students during exams or assignments, vary in structure, layout, and correctness. This study examines whether current vision-language and large language models can process such diagrams and produce accurate textual and digital representations. In this study, scanned student-drawn diagrams are used as input. Then, textual descriptions are generated from these images using a vision-language model. The descriptions are checked and revised by human reviewers to make them accurate. Both the generated and the revised descriptions are then fed to a large language model to generate TikZ code. The resulting diagrams are compiled and then evaluated against the original scanned diagrams. We found descriptions generated directly from images using vision-language models are often incorrect and human correction can substantially improve the quality of vision language model generated descriptions. This research can help computer science education by paving the way for automated grading and feedback and creating more accessible instructional materials.

Ethan Young, Zichun Wang, Aiden Taylor +6

Marshall University · West Virginia State University

Code GenerationVLM Evaluation
AskPDF
Mar 9, 2026

VLM-SubtleBench: How Far Are VLMs from Human-Level Subtle Comparative Reasoning?

The ability to distinguish subtle differences between visually similar images is essential for diverse domains such as industrial anomaly detection, medical imaging, and aerial surveillance. While comparative reasoning benchmarks for vision-language models (VLMs) have recently emerged, they primarily focus on images with large, salient differences and fail to capture the nuanced reasoning required for real-world applications. In this work, we introduce VLM-SubtleBench, a benchmark designed to evaluate VLMs on subtle comparative reasoning. Our benchmark covers ten difference types - Attribute, State, Emotion, Temporal, Spatial, Existence, Quantity, Quality, Viewpoint, and Action - and curate paired question-image sets reflecting these fine-grained variations. Unlike prior benchmarks restricted to natural image datasets, our benchmark spans diverse domains, including industrial, aerial, and medical imagery. Through extensive evaluation of both proprietary and open-source VLMs, we reveal systematic gaps between model and human performance across difference types and domains, and provide controlled analyses highlighting where VLMs' reasoning sharply deteriorates. Together, our benchmark and findings establish a foundation for advancing VLMs toward human-level comparative reasoning.

Minkyu Kim, Sangheon Lee, Dongmin Park

KRAFTON · KAIST

VLM EvaluationVLM ReasoningFine-Grained Visual Perception
AskPDF
Mar 9, 2026

Hospitality-VQA: Decision-Oriented Informativeness Evaluation for Vision-Language Models

Recent advances in Vision-Language Models (VLMs) have demonstrated impressive multimodal understanding in general domains. However, their applicability to decision-oriented domains such as hospitality remains largely unexplored. In this work, we investigate how well VLMs can perform visual question answering (VQA) about hotel and facility images that are central to consumer decision-making. While many existing VQA benchmarks focus on factual correctness, they rarely capture what information users actually find useful. To address this, we first introduce Informativeness as a formal framework to quantify how much hospitality-relevant information an image-question pair provides. Guided by this framework, we construct a new hospitality-specific VQA dataset that covers various facility types, where questions are specifically designed to reflect key user information needs. Using this benchmark, we conduct experiments with several state-of-the-art VLMs, revealing that VLMs are not intrinsically decision-aware-key visual signals remain underutilized, and reliable informativeness reasoning emerges only after modest domain-specific finetuning.

Jeongwoo Lee, Baek Duhyeong, Eungyeol Han +5

Yonsei University · Yanolja NEXT

VLM EvaluationVisual Question Answering
AskPDF
Mar 8, 2026

OrdinalBench: A Benchmark Dataset for Diagnosing Generalization Limits in Ordinal Number Understanding of Vision-Language Models

Vision-Language Models (VLMs) have advanced across multimodal benchmarks but still show clear gaps in ordinal number understanding, i.e., the ability to track relative positions and generalize to large indices. We present OrdinalBench, a diagnostic benchmark that standardizes ordinal number understanding as an evaluation task for VLMs. The core task is N-th object identification, defined by a starting reference and traversal rule. Task difficulty is controlled along three axes: (i) ordinal magnitude, from small numbers to extreme cases up to 300; (ii) arrangement complexity, from single loops to maze-like paths; and (iii) object count. The benchmark provides 39,000 question-answer pairs, each annotated with a ground-truth reasoning trajectory and balanced across difficulty levels for controlled large-scale testing. Beyond answer-only evaluation, our framework requires models to generate structured stepwise traces of the counting process and provides an open evaluation toolkit that measures both final accuracy and step-level path consistency. Zero-shot evaluations of GPT-5, Gemini 2.5 Flash Lite, Qwen2.5-VL, InternVL3.5, and Molmo reveal sharp degradation under large-ordinal and complex-path conditions, highlighting weak generalization despite strong scores on standard multimodal tasks. By framing ordinal number understanding as a core target, OrdinalBench provides a reproducible benchmark and diagnostic framework for developing VLMs with stronger sequential reasoning. All data and code are available at https://ordinalbench.github.io/

Yusuke Tozaki, Hisashi Miyamori

Division of Frontier Informatics, Kyoto Sangyo University, Kyoto, Japan

VLM EvaluationVLM Reasoning
AskPDF
Mar 8, 2026

Scaling Test-Time Robustness of Vision-Language Models via Self-Critical Inference Framework

The emergence of Large Language Models (LLMs) has driven rapid progress in multi-modal learning, particularly in the development of Large Vision-Language Models (LVLMs). However, existing LVLM training paradigms place excessive reliance on the LLM component, giving rise to two critical robustness challenges: language bias and language sensitivity. To address both issues simultaneously, we propose a novel Self-Critical Inference (SCI) framework that extends Visual Contrastive Decoding by conducting multi-round counterfactual reasoning through both textual and visual perturbations. This process further introduces a new strategy for improving robustness by scaling the number of counterfactual rounds. Moreover, we also observe that failure cases of LVLMs differ significantly across models, indicating that fixed robustness benchmarks may not be able to capture the true reliability of LVLMs. To this end, we propose the Dynamic Robustness Benchmark (DRBench), a model-specific evaluation framework targeting both language bias and sensitivity issues. Extensive experiments show that SCI consistently outperforms baseline methods on DRBench, and that increasing the number of inference rounds further boosts robustness beyond existing single-step counterfactual reasoning methods.

Kaihua Tang, Jiaxin Qi, Jinli Ou +2

Tongji University, China · Computer Network Information Center, CAS, China · University of Chinese Academy of Sciences, China +1

VLM RobustnessTest-Time Scaling for VLMsVLM ReasoningCounterfactual Evaluation of VLMsVLM EvaluationContrastive Decoding
AskPDF
Mar 8, 2026

Prompt-Based Caption Generation for Single-Tooth Dental Images Using Vision-Language Models

Digital dentistry has made significant advances with the advent of deep learning. However, the majority of these deep learning-based dental image analysis models focus on very specific tasks such as tooth segmentation, tooth detection, cavity detection, and gingivitis classification. There is a lack of a specialized model that has holistic knowledge of teeth and can perform dental image analysis tasks based on that knowledge. Datasets of dental images with captions can help build such a model. To the best of our knowledge, existing dental image datasets with captions are few in number and limited in scope. In many of these datasets, the captions describe the entire mouth, while the images are limited to the anterior view. As a result, posterior teeth such as molars are not clearly visible, limiting the usefulness of the captions for training vision-language models. Additionally, the captions focus only on a specific disease (gingivitis) and do not provide a holistic assessment of each tooth. Moreover, tooth disease scores are typically assigned to individual teeth, and each tooth is treated as a separate entity in orthodontic procedures. Therefore, it is important to have captions for single-tooth images. As far as we know, no such dataset of single-tooth images with dental captions exists. In this work, we aim to bridge that gap by assessing the possibility of generating captions for dental images using Vision-Language Models (VLMs) and evaluating the extent and quality of those captions. Our findings suggest that guided prompts help VLMs generate meaningful captions. We show that the prompts generated by our framework are better anchored in describing the visual aspects of dental images. We selected RGB images as they have greater potential in consumer scenarios.

Anastasiia Sukhanova, Aiden Taylor, Julian Myers +5

Marshall University · West Virginia State University

Dental ImagingImage CaptioningMedical VLMsMedical ImagingVLM EvaluationPrompt Engineering
AskPDF
Mar 7, 2026

Vision Language Models Cannot Reason About Physical Transformation

Understanding physical transformations is fundamental for reasoning in dynamic environments. While Vision Language Models (VLMs) show promise in embodied applications, whether they genuinely understand physical transformations remains unclear. We introduce ConservationBench evaluating conservation -- whether physical quantities remain invariant under transformations. Spanning four properties with paired conserving/non-conserving scenarios, we generate and evaluate 23,040 questions across 112 VLMs. Results reveal systematic failure: performance remains near chance with improvements on conservation tasks accompanied by drops on controls. Control experiments show strong textual priors favoring invariance, yet models perform worse with actual visual content when performance is balanced across conserving and non-conserving scenarios. Neither temporal resolution, prompting, nor curated sampling helps. These findings show that current VLMs fail to maintain transformation-invariant representations of physical properties across dynamic scenes.

Dezhi Luo, Yijiang Li, Maijunxian Wang +7

University of Michigan · University of California San Diego · University of California Berkeley +6

Physical ReasoningVision-Language ModelsVLM ReasoningVLM Evaluation
AskPDF
Mar 7, 2026

VirtueBench: Evaluating Trustworthiness under Uncertainty in Long Video Understanding

Recent Vision-Language Models (VLMs) have made remarkable progress in multimodal understanding tasks, yet their evaluation on long video understanding remains unreliable. Due to limited frame inputs, key frames necessary for answering the question may be missing from the model's input. However, models that truthfully refuse to answer under such uncertainty are marked as incorrect, while those that guess may coincidentally produce the correct answer and thus obtain deceptively higher accuracy, leading to misleading evaluation results and encouraging models to guess rather than respond honestly. To address this issue, we introduce VirtueBench, a benchmark explicitly designed to assess model trustworthiness under uncertainty. VirtueBench constructs multiple frame-sampling levels for each video and provides ground truths that distinguish between answerable and unanswerable cases. Evaluations on 25 open-source and commercial VLMs reveal distinct refusal behaviors across different model families, with refusal accuracy ranging from over 70% in the best models to nearly 0% in the worst. Moreover, most models exhibit a substantial drop in refusal when the prompt does not explicitly require them to do so. These findings highlight the need for developing trustworthy VLMs for multimodal understanding, guided by benchmarks and leaderboards that emphasize reliability and trustworthiness.

Xueqing Yu, Bohan Li, Yan Li +1

Peking University · ByteDance

Abstention in VLMsVLM EvaluationVideo-Language Model EvaluationVision-Language ModelsLong-Video Understanding
AskPDF
Mar 6, 2026

VLM-RobustBench: A Comprehensive Benchmark for Robustness of Vision-Language Models

Vision-language models (VLMs) achieve strong performance on standard, high-quality datasets, but we still do not fully understand how they perform under real-world image distortions. We present VLM-RobustBench, a benchmark spanning 49 augmentation types across noise, blur, weather, digital, and geometric perturbations, evaluated under graded severities (low/mid/high) and binary transforms, yielding 133 corrupted settings. We evaluate VLMs from four families (Qwen, InternVL, Molmo, Gemma) on two complementary benchmarks: MMBench (visually grounded) and MMMU-Pro (reasoning-oriented). Our results reveal that visual severity is a weak predictor of difficulty: low-severity spatial perturbations often degrade performance more than visually severe photometric corruptions. In particular, low-severity glass_blur reduces MMBench accuracy by about 8 pp on average across models, while the largest drops arise from resampling and geometric distortions (e.g., upsample, elastic_transform), reaching up to 34 pp. Overall, our findings suggest current VLMs are semantically strong but spatially fragile, motivating the definition of novel robustness evaluation protocols and training regimes that emphasize resampling and geometric invariances.

Rohit Saxena, Alessandro Suglia, Pasquale Minervini

University of Edinburgh, UK

VLM RobustnessVLM Evaluation
AskPDF
Mar 6, 2026

Spatial Colour Mixing Illusions as a Perception Stress Test for Vision-Language Models

Vision-language models (VLMs) achieve strong benchmark results, yet can exhibit systematic perceptual weaknesses: structured, large changes to pixel values can cause confident yet nonsensical predictions, even when the underlying scene remains easily recognizable to humans. We study this gap using Spatial Colour Mixing, a programmatic family of colour distortions that overlays structured patterns (in both RGB and Ostwald colour systems) onto natural images. We introduce a framework of eight spatial colour mixing variants and evaluate nine VLMs across three model families on four datasets. Across models and datasets, accuracy degrades sharply with increasing distortion, and scaling the language model does not reliably mitigate the failure. In a human study with 61 participants on an animal recognition dataset, humans substantially outperform VLMs under the same distortions. Finally, we show that a simple human-inspired preprocessing step recovers a meaningful portion of performance for several distortion types, motivating perception-aware preprocessing and tool-use as practical strategies for improving VLM robustness.

Nicoleta-Nina Basoc, Adrian Cosma, Emilian Radoi

National University of Science and Technology POLITEHNICA Bucharest, Bucharest, Romania · IDSIA, The Dalle Molle Institute for Artificial Intelligence, Lugano, Switzerland

VLM RobustnessVision-Language ModelsVLM Evaluation
AskPDF
Mar 4, 2026

Underrepresented in Foundation Model Pretraining Data? A One-Shot Probe

Large-scale Vision-Language Foundation Models (VLFMs), such as CLIP, now underpin a wide range of computer vision research and applications. VLFMs are often adapted to various domain-specific tasks. However, VLFM performance on novel, specialised, or underrepresented domains remains inconsistent. Evaluating VLFMs typically requires labelled test sets, which are often unavailable for niche domains of interest, particularly those from the Global South. We address this gap by proposing a highly data-efficient method to predict a VLFM's zero-shot accuracy on a target domain using only a single labelled image per class. Our approach uses a Large Language Model to generate plausible counterfactual descriptions of a given image. By measuring the VLFM's ability to distinguish the correct description from these hard negatives, we engineer features that capture the VLFM's discriminative power in its shared embedding space. A linear regressor trained on these similarity scores estimates the VLFM's zero-shot test accuracy across various visual domains with a Pearson-r correlation of 0.96. We demonstrate our method's performance across five diverse datasets, including standard benchmark datasets and underrepresented datasets from Africa. Our work provides a low-cost, reliable tool for probing VLFMs, enabling researchers and practitioners to make informed decisions about data annotation efforts before committing significant resources. The model training code, generated captions and counterfactuals are released here: https://github.com/chris-vorster/PreLabellingProbe.

Chris Vorster, Mayug Maniparambil, Noel E. O'Connor +2

ML-Labs, Dublin City University, Dublin, Ireland

VLM EvaluationZero-Shot Image Classification
AskPDF
Mar 4, 2026

Real5-OmniDocBench: A Full-Scale Physical Reconstruction Benchmark for Robust Document Parsing in the Wild

While Vision-Language Models (VLMs) achieve near-perfect scores on digital document benchmarks like OmniDocBench, their performance in the unpredictable physical world remains largely unknown due to the lack of controlled yet realistic evaluations. We introduce Real5-OmniDocBench, the first benchmark that performs a full-scale, one-to-one physical reconstruction of the entire OmniDocBench v1.5 (1,355 images) across five critical real-world scenarios: Scanning, Warping, Screen-Photography, Illumination, and Skew. Unlike prior benchmark that either lack digital correspondence or employ partial sampling, our complete ground-truth mapping enables, for the first time, rigorous factor-wise attribution of performance degradation-allowing us to pinpoint whether failures stem from geometric distortions, optical artifacts, or model limitations. Our benchmark establishes a challenging new standard for the community, demonstrating that the 'reality gap' in document parsing is far from closed, and provides a diagnostic tool to guide the development of truly resilient document intelligence.

Changda Zhou, Ziyue Gao, Xueqing Wang +4

PaddlePaddle Team, Baidu Inc. · The Hong Kong University of Science and Technology (Guangzhou)

VLM EvaluationDocument Parsing BenchmarksDocument ParsingDocument Image Understanding
AskPDF
Mar 4, 2026

DISCO: Document Intelligence Suite for COmparative Evaluation

Document intelligence requires accurate text extraction and reliable reasoning over document content. We introduce DISCO, a Document Intelligence Suite for COmparative Evaluation, that evaluates optical character recognition (OCR) pipelines and vision-language models (VLMs) separately on parsing and question answering across diverse document types, including handwritten text, multilingual scripts, medical forms, infographics, and multi-page documents. Our evaluation shows that performance varies substantially across tasks and document characteristics, underscoring the need for complexity-aware approach selection. OCR pipelines are generally more reliable for handwriting and for long or multi-page documents, where explicit text grounding supports text-heavy reasoning, while VLMs perform better on multilingual text and visually rich layouts. Task-aware prompting yields mixed effects, improving performance on some document types while degrading it on others. These findings provide empirical guidance for selecting document processing strategies based on document structure and reasoning demands.

Kenza Benkirane, Dan Goldwater, Martin Asenov +1

Parexel AI Labs, London, United Kingdom · Parexel AI Labs, San Francisco, United-States

Document UnderstandingDocument Image UnderstandingVLM EvaluationOptical Character RecognitionMultilingual VLM Evaluation
AskPDF
Mar 4, 2026

When Visual Evidence is Ambiguous: Pareidolia as a Diagnostic Probe for Vision Models

When visual evidence is ambiguous, vision models must decide how to interpret face-like patterns. Face pareidolia, the perception of faces in non-face objects, provides a controlled probe of such decisions. We introduce a diagnostic framework that analyzes detection, localization, uncertainty and bias across class, difficulty and emotion. We evaluate six models spanning four representational regimes: vision-language models (VLMs; CLIP-B/32, CLIP-L/14, LLaVA-1.5-7B), pure vision classification (ViT), object detection (YOLOv8), and face detection (RetinaFace). Our results reveal that uncertainty and bias are decoupled: low uncertainty can signal either safe suppression, as in detectors, or extreme over-interpretation, as in VLMs. VLMs exhibit semantic overactivation, systematically interpreting ambiguous non-human regions as Human, with LLaVA over-calling on 73% of non-human pareidolic images, especially for negative emotions. ViT instead follows an uncertainty-as-abstention strategy, remaining diffuse yet largely unbiased. Detection-based models achieve low bias through conservative priors that suppress pareidolia responses even when localization is controlled. Together, these results show that behavior under ambiguity is governed more by representation than thresholds, establishing face pareidolia as a diagnostic of semantic robustness and a source of ambiguity-aware hard negatives for vision models. Code will be released upon publication.

Qianpu Chen, Derya Soydaner, Rob Saunders

LIACS (Leiden Institute of Advanced Computer Science), Leiden University, Leiden, The Netherlands

VLM RobustnessVLM EvaluationVision-Language Models
AskPDF
Mar 3, 2026

VB: Visibility Benchmark for Visibility and Perspective Reasoning in Images

We present VB, a benchmark that tests whether vision-language models can determine what is and is not visible in a photograph, and abstain when a human viewer cannot reliably answer. Each item pairs a single photo with a short yes/no visibility claim; the model must output VISIBLY_TRUE, VISIBLY_FALSE, or ABSTAIN, together with a confidence score. Items are organized into 100 families using a 2x2 design that crosses a minimal image edit with a minimal text edit, yielding 300 headline evaluation cells. Unlike prior unanswerable-VQA benchmarks, VB tests not only whether a question is unanswerable but why (via reason codes tied to specific visibility factors), and uses controlled minimal edits to verify that model judgments change when and only when the underlying evidence changes. We score models on confidence-aware accuracy with abstention (CAA), minimal-edit flip rate (MEFR), confidence-ranked selective prediction (SelRank), and second-order perspective reasoning (ToMAcc); all headline numbers are computed on the strict XOR subset (three cells per family, 300 scored items per model). We evaluate nine models spanning flagship and prior-generation closed-source systems, and open-source models from 8B to 12B parameters. GPT-4o and Gemini 3.1 Pro effectively tie for the best composite score (0.728 and 0.727), followed by Gemini 2.5 Pro (0.678). The best open-source model, Gemma 3 12B (0.505), surpasses one prior-generation closed-source system. Text-flip robustness exceeds image-flip robustness for six of nine models, and confidence calibration varies substantially: GPT-4o and Gemini 2.5 Pro achieve similar accuracy yet differ sharply in selective prediction quality.

Neil Tripathi

New York University

VLM EvaluationVLM ReasoningCounterfactual Evaluation of VLMs
AskPDF
Mar 3, 2026

Logit-Level Uncertainty Quantification in Vision-Language Models for Histopathology Image Analysis

Vision-Language Models (VLMs) with their multimodal capabilities have demonstrated remarkable success in almost all domains, including education, transportation, healthcare, energy, finance, law, and retail. Nevertheless, the utilization of VLMs in healthcare applications raises crucial concerns due to the sensitivity of large-scale medical data and the trustworthiness of these models (reliability, transparency, and security). This study proposes a logit-level uncertainty quantification (UQ) framework for histopathology image analysis using VLMs to deal with these concerns. UQ is evaluated for three VLMs using metrics derived from temperature-controlled output logits. The proposed framework demonstrates a critical separation in uncertainty behavior. While VLMs show high stochastic sensitivity (cosine similarity (CS) <0.71<0.71 and <0.84<0.84, Jensen-Shannon divergence (JS) <0.57<0.57 and <0.38<0.38, and Kullback-Leibler divergence (KL) <0.55<0.55 and <0.35<0.35, respectively for mean values of VILA-M3-8B and LLaVA-Med v1.5), near-maximal temperature impacts (ΔT≈1.00Δ_T \approx 1.00), and displaying abrupt uncertainty transitions, particularly for complex diagnostic prompts. In contrast, the pathology-specific PRISM model maintains near-deterministic behavior (mean CS >0.90>0.90, JS <0.10<0.10, KL <0.09<0.09) and significantly minimal temperature effects across all prompt complexities. These findings emphasize the importance of logit-level uncertainty quantification to evaluate trustworthiness in histopathology applications utilizing VLMs.

Betul Yurdem, Ferhat Ozgur Catak, Murat Kuzlu +1

Department of Electrical and Electronics Engineering, Izmir Bakircay University, 35665 Izmir, Turkey · Department of Electrical Engineering and Computer Science, University of Stavanger, 4021 Stavanger, Norway · Batten College of Engineering and Technology, Old Dominion University, Norfolk, VA 23529, USA

Uncertainty QuantificationMedical VLMsVLM EvaluationHistopathology Image Classification
AskPDF
Mar 3, 2026

Structure-Aware Text Recognition for Ancient Greek Critical Editions

Recent advances in visual language models (VLMs) have transformed end-to-end document understanding. However, their ability to interpret the complex layout semantics of historical scholarly texts remains limited. This paper investigates structure-aware text recognition for Ancient Greek critical editions, which have dense reference hierarchies and extensive marginal annotations. We introduce two novel resources: (i) a large-scale synthetic corpus of 185,000 page images generated from TEI/XML sources with controlled typographic and layout variation, and (ii) a curated benchmark of real scanned editions spanning more than a century of editorial and typographic practices. Using these datasets, we evaluate three state-of-the-art VLMs under both zero-shot and fine-tuning regimes. Our experiments reveal substantial limitations in current VLM architectures when confronted with highly structured historical documents. In zero-shot settings, most models significantly underperform compared to established off-the-shelf software. Nevertheless, the Qwen3VL-8B model achieves state-of-the-art performance, reaching a median Character Error Rate of 1.0% on real scans. These results highlight both the current shortcomings and the future potential of VLMs for structure-aware recognition of complex scholarly documents.

Nicolas Angleraud, Antonia Karamolegkou, Benoît Sagot +1

Inria, Paris, France

Optical Character RecognitionVLM EvaluationDocument Layout Analysis
AskPDF
Mar 2, 2026

Cultural Counterfactuals: Evaluating Cultural Biases in Large Vision-Language Models with Counterfactual Examples

Large Vision-Language Models (LVLMs) have grown increasingly powerful in recent years, but can also exhibit harmful biases. Prior studies investigating such biases have primarily focused on demographic traits related to the visual characteristics of a person depicted in an image, such as their race or gender. This has left biases related to cultural differences (e.g., religion, socioeconomic status), which cannot be readily discerned from an individual's appearance alone, relatively understudied. A key challenge in measuring cultural biases is that determining which group an individual belongs to often depends upon cultural context cues in images, and datasets annotated with cultural context cues are lacking. To address this gap, we introduce Cultural Counterfactuals: a high-quality synthetic dataset containing nearly 60k counterfactual images for measuring cultural biases related to religion, nationality, and socioeconomic status. To ensure that cultural contexts are accurately depicted, we generate our dataset using an image-editing model to place people of different demographics into real cultural context images. This enables the construction of counterfactual image sets which depict the same person in multiple different contexts, allowing for precise measurement of the impact that cultural context differences have on LVLM outputs. We demonstrate the utility of Cultural Counterfactuals for quantifying cultural biases in popular LVLMs.

Phillip Howard, Xin Su, Kathleen C. Fraser

Thoughtworks · University of Ottawa

VLM EvaluationCounterfactual Evaluation of VLMsLarge Vision-Language ModelsCultural Bias in VLMs
AskPDF
Mar 2, 2026

Semantic Similarity is a Spurious Measure of Comic Understanding: Lessons Learned from Hallucinations in a Benchmarking Experiment

A system that enables blind or visually impaired users to access comics/manga would introduce a new medium of storytelling to this community. However, no such system currently exists. Generative vision-language models (VLMs) have shown promise in describing images and understanding comics, but most research on comic understanding is limited to panel-level analysis. To fully support blind and visually impaired users, greater attention must be paid to page-level understanding and interpretation. In this work, we present a preliminary benchmark of VLM performance on comic interpretation tasks. We identify and categorize hallucinations that emerge during this process, organizing them into generalized object-hallucination taxonomies. We conclude with guidance on future research, emphasizing hallucination mitigation and improved data curation for comic interpretation.

Christopher Driggers-Ellis, Nachiketh Tibrewal, Rohit Bogulla +4

University of Florida, Gainesville, FL 32611

VLM EvaluationVLM Hallucination
AskPDF
Mar 2, 2026

Measuring What VLMs Don't Say: Validation Metrics Hide Clinical Terminology Erasure in Radiology Report Generation

Reliable deployment of Vision-Language Models (VLMs) in radiology requires validation metrics that go beyond surface-level text similarity to ensure clinical fidelity and demographic fairness. This paper investigates a critical blind spot in current model evaluation: the use of decoding strategies that lead to high aggregate token-overlap scores despite succumbing to template collapse, in which models generate only repetitive, safe generic text and omit clinical terminology. Unaddressed, this blind spot can lead to metric gaming, where models that perform well on benchmarks prove clinically uninformative. Instead, we advocate for lexical diversity measures to check model generations for clinical specificity. We introduce Clinical Association Displacement (CAD), a vocabulary-level framework that quantifies shifts in demographic-based word associations in generated reports. Weighted Association Erasure (WAE) aggregates these shifts to measure the clinical signal loss across demographic groups. We show that deterministic decoding produces high levels of semantic erasure, while stochastic sampling generates diverse outputs but risks introducing new bias, motivating a fundamental rethink of how "optimal" reporting is defined.

Aditya Parikh, Aasa Feragen, Sneha Das +1

DTU Compute, Technical University of Denmark

VLM EvaluationRadiology Report EvaluationRadiology Report GenerationDemographic Bias in Language Models
AskPDF
Mar 1, 2026

Towards Policy-Adaptive Image Guardrail: Benchmark and Method

Accurate rejection of sensitive or harmful visual content, i.e., harmful image guardrail, is critical in many application scenarios. This task must continuously adapt to the evolving safety policies and content across various domains and over time. However, traditional classifiers, confined to fixed categories, require frequent retraining when new policies are introduced. Vision-language models (VLMs) offer a more adaptable and generalizable foundation for dynamic safety guardrails. Despite this potential, existing VLM-based safeguarding methods are typically trained and evaluated under only a fixed safety policy. We find that these models are heavily overfitted to the seen policy, fail to generalize to unseen policies, and even lose the basic instruction-following ability and general knowledge. To address this issue, in this paper we make two key contributions. First, we benchmark the cross-policy generalization performance of existing VLMs with SafeEditBench, a new evaluation suite. SafeEditBench leverages image-editing models to convert unsafe images into safe counterparts, producing policy-aligned datasets where each safe-unsafe image pair remains visually similar except for localized regions violating specific safety rules. Human annotators then provide accurate safe/unsafe labels under five distinct policies, enabling fine-grained assessment of policy-aware generalization. Second, we introduce SafeGuard-VL, a reinforcement learning-based method with verifiable rewards (RLVR) for robust unsafe-image guardrails. Instead of relying solely on supervised fine-tuning (SFT) under fixed policies, SafeGuard-VL explicitly optimizes the model with policy-grounded rewards, promoting verifiable adaptation across evolving policies. Extensive experiments verify the effectiveness of our method for unsafe image guardrails across various policies.

Caiyong Piao, Zhiyuan Yan, Haoming Xu +4

Fudan University · Tencent · Peking University

Cultural Safety Evaluation of VLMsVLM Evaluation
AskPDF
Mar 1, 2026

GameVerse: Can Vision-Language Models Learn from Video-based Reflection?

Human gameplay is a visually grounded interaction loop in which players act, reflect on failures, and watch tutorials to refine strategies. Can Vision-Language Models (VLMs) also learn from video-based reflection? We present GameVerse, a comprehensive video game benchmark that enables a reflective visual interaction loop. Moving beyond traditional fire-and-forget evaluations, it uses a novel reflect-and-retry paradigm to assess how VLMs internalize visual experience and improve policies. To facilitate systematic and scalable evaluation, we also introduce a cognitive hierarchical taxonomy spanning 15 globally popular games, dual action space for both semantic and GUI control, and milestone evaluation using advanced VLMs to quantify progress. Our experiments show that VLMs benefit from video-based reflection in varied settings, and perform best by combining failure trajectories and expert tutorials-a training-free analogue to reinforcement learning (RL) plus supervised fine-tuning (SFT).Our project page is available at https://gameverse-bench.github.io/ . Our code is available at https://github.com/THUSI-Lab/GameVerse .

Kuan Zhang, Dongchen Liu, Qiyue Zhao +5

College of AI, Tsinghua University, China.

Vision-Language ModelsVLM Evaluation
AskPDF
Mar 1, 2026

Can Vision Language Models Assess Graphic Design Aesthetics? A Benchmark, Evaluation, and Dataset Perspective

Assessing the aesthetic quality of graphic design is central to visual communication, yet remains underexplored in vision language models (VLMs). We investigate whether VLMs can evaluate design aesthetics in ways comparable to humans. Prior work faces three key limitations: benchmarks restricted to narrow principles and coarse evaluation protocols, a lack of systematic VLM comparisons, and limited training data for model improvement. In this work, we introduce AesEval-Bench, a comprehensive benchmark spanning four dimensions, twelve indicators, and three fully quantifiable tasks: aesthetic judgment, region selection, and precise localization. Then, we systematically evaluate proprietary, open-source, and reasoning-augmented VLMs, revealing clear performance gaps against the nuanced demands of aesthetic assessment. Moreover, we construct a training dataset to fine-tune VLMs for this domain, leveraging human-guided VLM labeling to produce task labels at scale and indicator-grounded reasoning to tie abstract indicators to concrete design regions.Together, our work establishes the first systematic framework for aesthetic quality assessment in graphic design. Our code and dataset will be released at: https://github.com/arctanxarc/AesEval-Bench

Arctanx An, Shizhao Sun, Danqing Huang +5

Peking University · Microsoft Research Asia · Microsoft +1

VLM Evaluation
AskPDF
Mar 1, 2026

The Aftermath of DrawEduMath: Vision Language Models Underperform with Struggling Students and Misdiagnose Errors

Effective mathematics education requires identifying and responding to students' mistakes. For AI to support pedagogical applications, models must perform well across different levels of student proficiency. Our work provides an extensive, year-long snapshot of how 11 vision-language models (VLMs) perform on DrawEduMath, a QA benchmark involving real students' handwritten, hand-drawn responses to math problems. We find that models' weaknesses concentrate on a core component of math education: student error. All evaluated VLMs underperform when describing work from students who require more pedagogical help, and across all QA, they struggle the most on questions related to assessing student error. Thus, while VLMs may be optimized to be math problem solving experts, our results suggest that they require alternative development incentives to adequately support educational use cases.

Li Lucy, Albert Zhang, Nathan Anderson +2

University of Washington · Insource Services · Worcester Polytechnic Institute +1

AI in EducationVLM EvaluationEducational AssessmentMathematics EducationIntelligent Tutoring Systems
AskPDF
Feb 27, 2026

Ref-Adv: Exploring MLLM Visual Reasoning in Referring Expression Tasks

Referring Expression Comprehension (REC) links language to region level visual perception. Standard benchmarks (RefCOCO, RefCOCO+, RefCOCOg) have progressed rapidly with multimodal LLMs but remain weak tests of visual reasoning and grounding: (i) many expressions are very short, leaving little reasoning demand; (ii) images often contain few distractors, making the target easy to find; and (iii) redundant descriptors enable shortcut solutions that bypass genuine text understanding and visual reasoning. We introduce Ref-Adv, a modern REC benchmark that suppresses shortcuts by pairing linguistically nontrivial expressions with only the information necessary to uniquely identify the target. The dataset contains referring expressions on real images, curated with hard distractors and annotated with reasoning facets including negation. We conduct comprehensive ablations (word order perturbations and descriptor deletion sufficiency) to show that solving Ref-Adv requires reasoning beyond simple cues, and we evaluate a broad suite of contemporary multimodal LLMs on Ref-Adv. Despite strong results on RefCOCO, RefCOCO+, and RefCOCOg, models drop markedly on Ref-Adv, revealing reliance on shortcuts and gaps in visual reasoning and grounding. We provide an in depth failure analysis and aim for Ref-Adv to guide future work on visual reasoning and grounding in MLLMs.

Qihua Dong, Kuo Yang, Lin Ju +6

Northeastern University

Referring Expression ComprehensionVLM EvaluationVision-Language GroundingVisually Grounded Reasoning
AskPDF