VQA
-
LeRF: Learning Reference Coordinate Frames for Perspective Taking Reasoning
2026-09-28 20:17:50Bang Xiao, Wenqi Jia, Ozgur Kara, Tiancheng Shen, Yibo Yang, Bolin Lai, Junho Kim, James Matthew RehgAbstract
Perspective taking is a fundamental component of spatial intelligence, requiring models interpret spatial relations from a specified viewpoint, such as that of another entity or an imagined observer. Despite the increasing spatial reasoning capabilities of Vision-Language Models (VLMs), they still struggle with perspective taking, often defaulting to the camera viewpoint when a query requires reasoning from a different perspective. We introduce Learning Reference Coordinate Frames for Perspective Taking (LeRF), a framework that trains VLMs to construct and use explicit reference frames for viewpoint-dependent reasoning. Given an image and a query, LeRF decides whether a coordinate frame is necessary. If so, it grounds the reference entity and predicts the frame's origin and entity-centered reference frame. A lightweight renderer overlays the frame onto the image, enabling subsequent reasoning over these visual cues without external perception models or explicit 3D reconstruction. To learn this process, we first perform supervised fine-tuning to teach selective tool invocation and reference coordinate frame prediction, followed by reinforcement learning on spatial VQA pairs to improve frame-guided reasoning. Across diverse perspective-taking benchmarks, LeRF consistently improves over its backbone and achieves strong performance against existing open-source methods. Further evaluations also show improved reference-frame grounding and orientation estimation, supporting the effectiveness of learned reference frames for viewpoint-dependent reasoning.
Abstract (translated)
URL
https://arxiv.org/abs/2609.36219
PDF
-
Xiaomi-OCR-0 Technical Report
2026-09-28 19:10:10Xin Chen, Anan Du, Feng Feng, Pei Fu, Jian Luan, Longwei Xu, Shaojie Zhang, Hang Li, Heng Qu, Cheng TanAbstract
Compact OCR-specific vision-language models achieve strong document parsing performance, but often rely on costly supervision and focus primarily on visual-text reconstruction. We introduce Xiaomi-OCR-0, a unified 0.8B model for document parsing and OCR-centric understanding. We build an approximately 170M-sample OCR-centric corpus using an automated data engine that combines expert consensus, render-based verification, and targeted synthesis. Starting from Qwen3.5-0.8B, our progressive training recipe combines Q-Mask-based text anchoring, continued pretraining, and mixed-task reinforcement learning (Mix-RL). Xiaomi-OCR-0 achieves 95.24 on Real5-OmniDocBench, 96.83 on OmniDocBench v1.6, and 87.94 on Wild-OmniDocBench, while reaching an average score of 83.2 across five OCR-oriented VQA benchmarks. Ablations further show that, with sufficient parsing training, OCR-centric understanding supervision provides additional gains for document parsing. Homepage: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.36136
PDF
-
Question-Specific Knowledge Graphs for Efficient Visual Reasoning
2026-09-28 16:21:13Ting-Chih Chen, Emile van Krieken, Shujian Yu, Filip IlievskiAbstract
Recent work in visual question answering has shown that vision-language models can exhibit strong reasoning capabilities by translating visual inputs into textual representations. The effectiveness of this translation depends on how well visual details are retained; models need to surface and align both explicit and implicit knowledge sufficient to support reasoning, without introducing spurious assumptions. Existing methods that leverage detailed image captions introduce visual details unrelated to the reasoning task, inflating input token counts and increasing computational cost. To address these challenges, we propose VisKG, a reinforcement learning (RL) framework in which models learn to translate visual content into question-specific knowledge graph (KG) representations. This process filters out perceptual noise while preserving the entity-relation structure needed for chain-of-thought reasoning, following the principle of minimum sufficient information. To ensure stable RL post-training, VisKG adopts Group reward-Decoupled Normalization Policy Optimization (GDPO). In addition, we strengthen the supervision stage with negative rationale samples, exposing the model to incorrect reasoning paths before RL post-training. Experimental results across science, mathematics, and general visual understanding benchmarks show that VisKG achieves performance comparable to or better than baselines, while requiring fewer tokens than caption-based representations. Moreover, training VisKG with GDPO improves accuracy by 2% over its GRPO-trained counterpart on average. These results suggest that KG representations are a promising approach for supporting multi-step reasoning and open up future work on adaptively selecting the most suitable representation for a given task.
Abstract (translated)
URL
https://arxiv.org/abs/2609.35942
PDF
-
From Reward Signal to Visual Utility: A Controlled Audit of Medical VLM Post-Training
2026-09-25 16:06:54Wang JingxinAbstract
Medical vision-language model (VLM) post-training is commonly evaluated through answer accuracy. We examine how changes in accuracy and training objectives relate to image-conditioned decisions in a controlled Qwen2.5-VL-3B study on PMC-VQA. We compare supervised fine-tuning (SFT) with low-rank adaptation (LoRA) restricted to the language model, expanded multimodal adaptation scopes, standard answer-only Group Relative Policy Optimization (GRPO), and a counterfactual evidence objective. On 2,000 clean-test questions, language model LoRA SFT changes correct-image accuracy by +1.10 percentage points (95% paired bootstrap CI:-0.85 to +3.05), while visual-benefit events decrease by 2.40 points and image sensitivity decreases by 5.60 points. Paired records reveal 155 acquired and 203 lost visual-benefit events. Broader adaptation yields lower correct-image accuracy than language-model LoRA SFT. Standard GRPO produces mixed-reward groups and parameter updates, with an uncertain clean test accuracy change. A generation audit reveals that canonical option scores can follow a different token path from generated answers. With scores taken along the greedy generation path, the evidence target improves on the training set; its gains over standard GRPO remain inconsistent on validation data at matched training doses. Sample-level analyses trace how evidence scores, decision margins, and generated answers change during post-training. This empirical and measurement audit identifies gaps between optimization activity, target acquisition, and useful held-out visual behavior.
Abstract (translated)
URL
https://arxiv.org/abs/2609.31450
PDF
-
DepthEvidence: Unifying Metric Depth Prediction and Geometric Reasoning in Multimodal Language Models
2026-09-25 10:42:20Jiangning Wei, Yuan Yao, Miaomiao Cui, Mingsheng Li, Humen Zhong, Shuai Bai, Zhibo YangAbstract
Spatial reasoning with metric constraints requires linking objects to geometric measurements and preserving their numerical content during language reasoning. We present DepthEvidence, a 4B model that uses its own dense metric predictions as object-grounded evidence for language generation. A camera-conditioned decoder predicts full-resolution metric depth using multi-scale visual features and high-resolution RGB refinement. A dense-to-language interface converts predicted depths and decoder features into object-aligned continuous geometry tokens anchored to object identifiers. Geometric supervision encourages metric information to remain recoverable before and after language-context interaction, while instruction tuning supports object measurement and compositional reasoning. We introduce a Depth-VQA benchmark evaluating object-depth queries, relative comparisons, and decisions combining spatial and numerical constraints. Across nine datasets, DepthEvidence achieves the highest average dense $\delta_1$ among evaluated methods, competitive with specialized estimators. It also leads the evaluated methods in instance-level metric depth estimation and overall accuracy on both relative and metric reasoning tracks, while broadly preserving general VQA performance and improving spatial understanding relative to the base model.
Abstract (translated)
URL
https://arxiv.org/abs/2609.31103
PDF
-
UltraG-Bench: A Multi-task Benchmark for assessing Large Vision-Language Models on Pixel-level Evidence Grounding in Ultrasound
2026-09-25 07:43:12Quanhao Zhu, Bo Xu, Rui Lin, Chenyuan Wang, Yu Shao, Boling Zhu, Jiuyan Sun, Liang Zhao, Hongfei Lin, Feng XiaAbstract
Ultrasound is one of the most widely used medical imaging modalities, and recent large vision-language models(VLMs) have shown increasing capabilities in ultrasound image understanding. However, these models fail to provide pixel-level visual evidence aligned with their semantic predictions, and their fine-grained grounding capability in ultrasound remains largely unclear. We introduce UltraG-Bench, a large-scale multi-task benchmark for evaluating pixel-level evidence grounding in ultrasound. UltraG-Bench is built by annotating 40 public ultrasound segmentation datasets spanning 13 anatomical categories, and comprises three progressive tasks: instruction-guided segmentation, evidence-grounded VQA, and evidence-grounded report generation, with 331125, 666779, and 138832 annotations, respectively. Comprehensive evaluation of 14 state-of-the-art models reveals a substantial gap between semantic understanding and fine-grained pixel-level localization. We further propose UltraG-Agent, which combines the semantic reasoning capabilities of a VLM with the ultrasound-specific segmentation capability of UltraSAM3. Experiments show that UltraG-Agent substantially improves both semantic prediction and pixel-level visual grounding. Our dataset and code are available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.30928
PDF
-
TrafficImag: A Benchmark for Counterfactual Roadside Traffic Video Generation
2026-09-25 02:49:06Xiangyu Li, Tianyi Wang, Zhihao Dou, Christian Claudel, Zhaomiao GuoAbstract
Existing roadside traffic datasets support perception, forecasting, and visual question answering, but they do not evaluate counterfactual video generation, in which a selected actor is modified and the generated future should remain consistent with road topology and unrelated traffic. We introduce TrafficImag, the first benchmark for counterfactual roadside traffic video generation. TrafficImag combines a large-scale roadside dataset (9,022 annotated images, 7,043 deduplicated video clips, and 31,145 actor-centered history-future samples) with an executable protocol that supports behavior reasoning, intervention-aware image editing, and conditional video generation. Each intervention is represented as an actor-level program describing the target actor, intended behavior, legal route, interaction order, and temporal constraints, enabling a unified evaluation interface across heterogeneous foundation models. TrafficImag evaluates four complementary validity dimensions: initial-state correctness, route and behavior validity, interaction consistency, and non-target preservation, and considers an end-to-end counterfactual successful only when all four are satisfied. Across state-of-the-art foundation models, the strongest reasoner reaches 80.4% macro F1, the complete condition interface raises end-to-end success from 23.3% to 55.0% for the best generator. Oracle studies further show that conditional video execution is the primary remaining bottleneck. TrafficImag provides a reproducible benchmark for evaluating and diagnosing counterfactual traffic video generation beyond perceptual video quality.
Abstract (translated)
URL
https://arxiv.org/abs/2609.30722
PDF
-
From Text Decisions to Pixels: An Study of Jev-Style Visual Choice Model
2026-09-24 09:19:05Xunlan Zhou, Xianliang Yang, Li ZhaoAbstract
Visual software often needs a decision over supplied alternatives rather than a generated explanation. We present PixelJev, a native-image decision interface that maps an image, a task instruction, and a runtime candidate set to a structured choice and candidate-conditioned probabilities using small open multimodal models. Its initial realization unifies recognition and multiplechoice visual question answering through an existing language-model readout, with separately evaluated options for frozen inference, language-side adaptation, and held-out calibration. Across seven benchmark evaluations, 64-shot source adaptation raises Pets accuracy from 60.13% to 92.40% across optimization seeds and transfers to natural resampling, new texture labels, and A-OKVQA without target fitting, while frozen inference already supports both VQA tasks. A matched prompt-only follow-up on Pets and ScienceQA attributes the large Pets gain to adaptation and identifies a narrower output validity benefit of candidate readout in adapted VQA. Specialist DINOv2 probes remain stronger on source recognition, frozen 4B is stronger than adapted 2B on DTD and ScienceQA, and accuracy gains do not ensure calibrated target probabilities. These findings establish a working starting point for general-purpose visual decision models and identify the remaining requirements: schema robustness, cross-family transfer, and reliable use of visual evidence.
Abstract (translated)
URL
https://arxiv.org/abs/2609.29283
PDF
-
Looks the Same, Answers Differently: Flip-Direction Steering for Robust Vision-Language Reasoning
2026-09-23 23:43:26Yeonsung Jung, Joonhyun Jeong, Hoang Pham, Joowon Kim, Yoonsik Park, Viet Dac Lai, Eunho YangAbstract
Vision-language models (VLMs) achieve strong visual reasoning performance, yet subtle changes from routine image capture and processing can alter their reasoning trajectories even when images appear nearly identical. In long-horizon generation, the resulting activation shifts may accumulate across decoding steps, progressively altering reasoning tokens and ultimately changing the final answer, a phenomenon referred to as answer flips. To address this instability, we propose FlipDir (Flip-Direction Steering), a training-free inference-time method that estimates a low-rank flip-inducing activation subspace from contrastive pairs of original and answer-flipping inputs and selectively steers hidden states during decoding. A margin-based gate limits subspace attenuation to uncertain decoding steps, recovering original predictions while preserving stable ones. To evaluate robustness beyond accuracy or consistency on fixed test sets, we introduce VisFlip, a benchmark framework that constructs evaluation groups for a target model and visual variation setting to separately assess recovery of original predictions and preservation of stable ones. VisFlip spans nine dataset-variation combinations across scientific reasoning, robot-scene understanding, and medical VQA, covering subtle visual variations common in each domain. Experiments across 18 settings demonstrate that FlipDir consistently outperforms existing methods on the combined recovery and preservation metric. We will make our code publicly available.
Abstract (translated)
URL
https://arxiv.org/abs/2609.28851
PDF
-
CinematicVQA: Benchmarking Film-Grammar Reasoning in Large Vision-Language Models
2026-09-23 21:47:34Shuo Xing, Pooja Verlani, Balu Adsumilli, Zhengzhong TuAbstract
Cinematography, the craft of visual storytelling through framing, lighting, and camera operation, fundamentally shapes how audiences perceive and emotionally engage with video content. While Large Vision Language Models (LVLMs) have made remarkable progress in video question answering, existing benchmarks primarily focus on identifying low-level techniques rather than understanding their storytelling impact. To address this, we introduce CinematicVQA, the first-of-its-kind benchmark for cinematic video understanding that goes beyond technique recognition to evaluate film-grammar reasoning, utilizing our introduced Cinematic Scene Graph (CSG), a structured representation that links filming techniques to their perceptual effects and narrative functions. Through comprehensive evaluation of state-of-the-art LVLMs, we reveal a striking semantic gap: models consistently perform higher on describing visual presentations than on identifying the underlying techniques. Surprisingly, Chain-of-Thought prompting fails to provide consistent gains and degrades performance for most models, suggesting that current LVLMs lack sufficient cinematic domain knowledge to benefit from step-by-step reasoning. Fine-tuning on \textsc{CinematicVQA-train} yields consistent improvements, particularly for narrative function and multi-hop reasoning. Overall, \textsc{CinematicVQA} serves both as a rigorous benchmark for cinematic evaluation in LVLMs and as a practical dataset for training more film-aware video models.
Abstract (translated)
URL
https://arxiv.org/abs/2609.28813
PDF
-
Small yet Assistive: Spatially-Aware Post-Training for Low Vision
2026-09-23 20:10:25Rishabh Choudhary, Shreyansh Raj, Umesh Goyal, Shubh Kashyap, Shrestha Kumar, Sushovan Jena, Komal Kumar, Hisham Cholakkal, Aditya NigamAbstract
An estimated 1 billion people worldwide live with vision impairment, yet current vision-language models (VLMs) produce descriptions too vague for safe navigation by blind and low-vision (BLV) users. Large VLMs can generate high-quality audio-description-compliant narrations but cannot run on mobile devices; small VLMs offer competitive latency but lack spatial detail, directional cues, and hazard awareness for navigational assistance. We present Smol-VL-BLV, a compact VLM for blind and low-vision users that closes this gap using a 500M decoder transformer model and two post-training mechanisms: (1) teacher-student distillation and (2) Group Relative Policy Optimization (GRPO) with a composite BLV reward targeting directional language, metric distances, and hazard detection. Because multi-stage post-training can induce catastrophic forgetting, we add a lightweight finetuning stage after the last stage GRPO finetuning to recover general descriptive quality while preserving BLV-specific spatial grounding. Our best model substantially outperforms the baseline across various benchmarks, including tasks: VQA, BLV captioning, OCR, and latency. Compared with the baseline for relative improvement, it improves the Spatial score gain of 19.3%, and the Social score gain of 14.8%. It also increases OCR-Bench by 101.5%, and raises TextVQA accuracy by 44.2%. These results show that BLV-focused post-training improves both accessibility-specific spatial grounding and general visual-text reasoning. Deployed on a mid-range Android smartphone via Mixed-Precision Quantization, the model remains approx. 450 MB and runs entirely on-device, offline and without network dependency, generating descriptions with latency dependent on host hardware capabilities. Our model, dataset, and code is publicly released at this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2609.28757
PDF
-
GeoNLI - A Natural Language Interpreter for Satellite Imagery
2026-09-23 19:31:23Ashutosh Gandhe, Anupam Rawat, Geet Sethi, Kabir Nasiruddin, Madhav Kotecha, Panav Shah, Rakshit Sawarn, Soumitra NayakAbstract
Multi-modal multitasking models have shown strong performance on remote sensing datasets. However, because these models are trained on heterogeneous data and vary across tasks, designing a unified model that performs well in captioning, visual question answering (VQA), and visual grounding remains challenging. In this work, we evaluate several models on the VRS Bench and NWPU-VHR-10 datasets. The EarthMind model demonstrates strong results in both captioning and VQA. For grounding, we propose multiple pipelines - RemoteSAM-SAM-v1, RemoteSAM-SAM-v2, and DiffuSAM - and ultimately adopt a majority-voting ensemble across EarthMind, RemoteSAM, SAM3, Falcon, RemoteSAM-SAM3-v1, RemoteSAM-SAM3-v2, and DiffuSAM predictions. Our unified, modular pipeline integrates advanced SAM variants with multimodal LLMs to jointly perform captioning, VQA, and grounding. It achieves 82% accuracy on captioning and 83.32% on VQA, with 90.94%, 52.04%, and 92.06% for binary, numeric, and semantic question types respectively. For grounding, it attains 64.94% accuracy. By combining diverse VLMs with our custom RemoteSAM-SAM3 models through ensemble majority voting, the system delivers more accurate and consistent remote-sensing understanding than task-specific approaches.
Abstract (translated)
URL
https://arxiv.org/abs/2609.28741
PDF
-
Mizar: A 159M-Parameter Audio-Language Model for Audio Understanding
2026-09-23 16:21:51Kaiyang Li, Shaobo Han, Yue Tian, Shihao JiAbstract
Audio-language models (ALMs) integrate acoustic perception with the knowledge encoded in language models, enabling contextual understanding of auditory events. Making these capabilities practical on devices with limited memory and computation motivates our focus on small ALMs with fewer than 200M parameters. We introduce a recipe that brings together architecture, data, and three-stage training to build Mizar, a 159.3M-parameter ALM. Its architecture connects a compact CED-Small audio encoder to SmolLM2-135M through a frequency-merging mapper. With supervision drawn from ReasonAQA, AudioMCQ, and AVQA, the model undergoes three training stages: audio-language alignment (Stage 1), audio-dependent fine-tuning (Stage 2), and post-training (Stage 3) aimed at strengthening weak skills while retaining learned capabilities. Across five random seeds, Mizar achieves mean accuracies of 52.92% on MMAU, 42.42% on MMAR, and 36.02% on ADQA-clean, surpassing the previous best-performing ALM below 200M parameters on all three benchmarks. It also supports local inference on a single CPU: on questions from the MMAU benchmark, the mean latency from opening the audio file to generating a complete answer is 1.09 seconds. Code and checkpoints are available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.28344
PDF
-
Less Language, More Latents: Annotation-Efficient VLAs for Driving
2026-09-23 12:00:04Alexey Zakharov, Kemal Oksuz, Puneet K. DokaniaAbstract
Vision-language-action models (VLA) promise human-steerable autonomous driving, but their training is bottlenecked by the scarcity of frames paired with natural-language instructions: while camera streams and expert trajectories are logged at scale, language annotations (e.g., turn left at the intersection) remain scarce and expensive to acquire. To address this challenge, we introduce Latent Action Driving Annotations (LADA), a three-stage pipeline that transforms abundant unlabelled observation-trajectory pairs into a substrate for language-conditioned control. First, we train a latent action model with a vector-quantised bottleneck, producing a compact codebook of high-level vehicle intents. Second, a small language-annotated subset is used to train a vision-language translator to map observations and language instructions into this codebook. Third, we train a driving VLA on observation-latent-action pairs over the full unlabelled corpus. Using fewer than 5% of language annotations and without leveraging any auxiliary chain-of-thought reasoning or visual question answering streams, LADA achieves a Driving Score of 87.98 and a Success Rate of 70.46% on the closed-loop Bench2Drive benchmark, matching or surpassing fully supervised baselines.
Abstract (translated)
URL
https://arxiv.org/abs/2609.27747
PDF
-
Gender Bias in Vision-Language In-Context Learning
2026-09-23 11:00:05Tong Xiang, Noa Garcia, Yuta NakashimaAbstract
In-context learning (ICL) enables large vision-language models (LVLMs) to perform tasks by following patterns from in-context examples, yet its potential to amplify societal biases remains underexplored. We systematically investigate how ICL influences gender bias in LVLMs through VL-BICLE, an evaluation framework comprising six ICL settings, three tasks, and four datasets. Our experiments on six LVLMs reveal that gendered ICL demonstrations act as a directional force, shifting model bias toward the demonstrated gender through a cross-gender mechanism that disproportionately degrades performance on the opposite gender. This effect appears in image captioning and pronoun prediction but not in visual question answering, indicating that gendered ICL influences bias only when the task output involves gendered language. Similarity-based retrieval methods inherit the training pool's gender imbalance and offer no debiasing advantage, while standard quality metrics remain blind to these bias shifts. To mitigate this bias, we replace real in-context images with synthetic ones from stable diffusion models while keeping captions unchanged. This simple intervention reduces gender bias without degrading caption quality.
Abstract (translated)
URL
https://arxiv.org/abs/2609.27682
PDF
-
When Visual Quality Misleads: Intent Recognition under Rendered Avatar Distortions
2026-09-23 08:45:40Ning-Hsuan Chang, Kai-Siang Ma, Yu-Chih ChenAbstract
Avatar-streaming systems are commonly evaluated with image and video quality assessment (IQA/VQA) metrics, implicitly treating visual fidelity as a proxy for communicative success. We test this assumption through a controlled behavioral study of rendered 3D avatars across a pristine condition and fourteen geometric, photometric, temporal, and combined distortions. Fifty-nine participants contributed 2,688 judgments of perceived action, response confidence, and visual quality. We identify Misleading Quality in this dataset as distorted renderings that retain above-average perceived quality but yield below-average action-recognition accuracy. We also derive an Intent Quality Score (IQS) combining recognition correctness and confidence as the behavioral target for objective metrics. Among 126 distorted content--condition cells, 31 (24.6%) exhibited Misleading Quality; temporal and geometric distortions showed the highest rates, at 50.0% and 31.1%, respectively. The results reveal a quality--accuracy dissociation where distortion families affect appearance and communication differently. Across 24 direct-scoring IQA/VQA metrics and three supervised feature-regression baselines, alignment with IQS remained limited; at $\lambda=0.5$, the best leave-one-content-out baseline reached PLCC $=0.4435$. Under this controlled protocol, visual fidelity alone is insufficient for avatar communication, motivating intent-aware quality assessment and streaming objectives.
Abstract (translated)
URL
https://arxiv.org/abs/2609.27560
PDF
-
Invisible in Space, Visible in Time: Motion Vision CAPTCHA against GUI Agents
2026-09-23 07:27:23Zeyu Zhang, Dingyi Rong, Zijian Chen, Zicheng Zhang, Xiongkuo Min, Guangtao ZhaiAbstract
Most existing visual CAPTCHAs remain spatially solvable: the required information is exposed by static appearance, local structure, and interface state. This assumption is weakened by advances in multimodal large language models (MLLMs) and Graphical User Interface (GUI) agents, which exhibit strong visual perception, reasoning, and browser interaction capabilities. We propose Motion Vision CAPTCHA (MVCAP), a hierarchical motion-based CAPTCHA framework in which target semantics are instantiated as motion-defined foreground structures and become recoverable only through temporal segregation from a dynamically evolving background. Built on this shared principle, MVCAP is instantiated in three perceptually progressive levels: coherent motion, structural motion, and biological motion. To evaluate this framework, we introduce MVCAP-Bench, a browser-based benchmark with 600 live CAPTCHA instances, together with a matched foreground-only control benchmark, MVCAP-Bench-FG. We evaluate humans, Browser Use agents, native computer use agents, and a supplementary offline VQA setting derived from the same instances. Results reveal a substantial human--agent gap: on the full MVCAP-Bench, human accuracy reaches 99.6%, whereas the best GUI agent achieves only 16.8%, close to the six-way chance level. The foreground-only control further shows that the key difficulty comes from dynamic background camouflage rather than answer format or browser interaction alone. These findings identify a measurable human--agent perception gap and position MVCAP-Bench as a benchmark for studying motion-defined perception in current agents.
Abstract (translated)
URL
https://arxiv.org/abs/2609.27461
PDF
-
Beyond Balanced Accuracy: A Resolution and Parity-Controlled Benchmark for Vision-Language and Vision-Only Defect Assessment in UAV Power-Line Inspection
2026-09-23 07:23:58Linghao Zhang, Siyu Xiang, Junwei Kuang, Peiyu YiAbstract
Vision-language models (VLMs) are often reported to outperform task-specific vision backbones for unmanned aerial vehicle (UAV) power-line defect assessment. We test that claim on ElecVQA-Bench, a 56,972-item benchmark derived from the public InsPLAD dataset, across six evaluation choices: partition, evaluated item set, label space, replication, input resolution, and side information. On a matched partition, a Swin Transformer and the strongest adapted VLM differ by only 0.03 points at binary screening. At seven-way defect typing, increasing the vision backbones from 224 px to the measured pixel budget of the VLM preprocessor narrows the gap against InternVL3.5-8B from +20.53 to -0.57 points for ResNet-50 and from +23.67 to +4.70 points for Swin-T. A pixel-budget audit shifts Qwen3-VL-8B macro recall by 10.78 points, yet a source-pixel-matched InternVL control still leaves Qwen ahead by 7.43 to 13.61 points while using 56% fewer visual tokens, so neither source pixels nor token budget explains the difference between the two VLMs. A two-seed global replication changes Qwen binary accuracy and seven-way macro recall by 0.86 and 1.02 points. After split-specific retraining, Qwen does not lead at crop or image level, and a 14-tower, three-seed replication reverses the sign across seeds, giving mean common-six macro recall of 0.9085 for Qwen against 0.9509 for ResNet-50. No split regime yields a family-level advantage that survives multiple-comparison correction. The study supports a benchmark-audit contribution rather than a general claim of VLM superiority.
Abstract (translated)
URL
https://arxiv.org/abs/2609.27457
PDF
-
Reading Right, Answering Wrong: How Visual Configuration Changes Affect Evidence Use in VLMs
2026-09-22 07:04:15Dingyang Lin, Yingfeng Luo, Chenglong Wang, Chenwei Zhu, Anxiang Ma, Jingbo Zhu, Tong XiaoAbstract
Vision-language models (VLMs) have achieved strong performance on tasks such as visual question answering, yet small image resizes can turn correct answers into errors. We investigate whether changes in visual configuration, such as image tiling and token arrangement, contribute to this instability. Across seven checkpoints and four benchmarks, equally small resizes cause more correctness flips when they switch configurations. Surprisingly, in over half of these cases, models answer the question incorrectly but can still read the correct answer when told what to read. Furthermore, attention interventions in LLaVA-NeXT suggest that configuration changes can weaken the use of readable information during answering. We therefore guide models using field cues and their own transcriptions. With annotation assistance, these forms of guidance together correct 97.2% of errors with readable information. These findings show that configuration changes can affect how models use information they can still read.
Abstract (translated)
URL
https://arxiv.org/abs/2609.25770
PDF
-
SLICEChat: Progressive In-Encoder Token Pruning for Whole-Slide Pathology Language Models
2026-09-21 16:59:06Ali Kerem Bozkurt, Baris Cem Bakay, Ibrahim Kulac, Cigdem Gunduz-Demir, Erkut Erdem, Aykut ErdemAbstract
Whole-slide pathology images (WSIs) contain gigapixel-scale visual content, creating a major scalability challenge for slide-level multimodal large language models (MLLMs). Existing approaches process thousands of patch tokens and typically apply compression only after slide encoding, leaving multimodal attention computationally expensive. We introduce SLICEChat, a slide-level MLLM that integrates progressive token pruning within a hybrid Mamba--Transformer slide encoder. Mamba layers enable efficient long-range propagation, while Transformer layers preserve global interactions as the sequence is progressively shortened. Between stages, language-supervised, region-aware pruning removes spatially coherent low-utility regions under a controlled keep-rate schedule, producing compact slide representations before multimodal fusion. On SlideBench VQA, SLICEChat achieves 79.84% accuracy on TCGA and 59.09% on BCNB cohorts, outperforming prior slide-level pathology MLLMs, and achieves the highest overall WSI-Bench metrics. It also provides competitive memory usage and the inference latency among the evaluated models. These results demonstrate accurate and computationally efficient multimodal reasoning over gigapixel WSIs.
Abstract (translated)
URL
https://arxiv.org/abs/2609.24894
PDF
