arXiv AI

FocusVTC: Efficient and High-Performance Visual Text Compression with Adaptive Resolution

arXiv AI
Oct 28

V-LUMEN: Visual Lookup Memory for Embedding Scaling in Vision-Language Models

V-LUMEN introduces a visual lookup memory for vision‑language models by discretizing visual features with a frozen codebook, creating spatially aware memory keys via hashing, and modulating them with text‑conditioned relevance signals. The retrieved embeddings are fused into a frozen Qwen3‑VL‑2B backbone through a gated residual pathway, yielding substantial improvements on five reasoning‑intensive benchmarks and zero‑shot transfer to three disjoint image datasets. At a matched parameter budget, reallocating capacity from LoRA or MoE adapters to V‑LUMEN preserves accuracy on two backbones, demonstrating that external visual memory offers an independent capacity axis. whyItMatters":"The paper shows that adding a non‑parametric visual lookup memory can significantly boost vision‑language model performance without increasing per‑token compute, offering a new way to scale such models."

By Miso Choi, Daekeun Kim, Eunji Kim, Jungbeom Lee
arXiv Computer Vision
Sep 22

Pay More Attention To Text In High-Resolution MLLMs

The paper introduces EviSpec, a training‑free compiler that generates complementary evidence specifications to improve high‑resolution multimodal large language models (MLLMs). By explicitly guiding visual search with structured evidence specifications, EviSpec achieves significant relative gains—up to 14.8% over random evidence—across five MLLMs and three benchmarks, and also sets new state‑of‑the‑art results on VQA and hallucination‑focused tasks.

By Zhongkuan Mao, Wenzhuo Zhao, Xianjie Liu, Yidong Wang, Zhao Gao, Ronghao Xian, Yao Jiang, Yi Zhang, Liangjian Wen, Keren Fu
arXiv Computer Vision
Oct 9

Look Back, Think Ahead: Visual Memory on Demand for Efficient Multimodal Reasoning

The paper introduces ViMoD, a lightweight framework that dynamically manages visual memory for multimodal large language models. ViMoD uses Deformable Aggregation of Region-wise Tokens (DART) to create content‑adaptive coarse representations linked to recoverable fine tokens, and Temporal Routing for Adaptive Contextual Evidence (TRACE) to anticipate and adjust evidence needs during reasoning. On the Qwen3‑VL‑4B model, ViMoD achieves superior performance across eight reasoning benchmarks with only a 20% visual token budget and minimal additional parameters.

By Yicheng Xue, Han Wu, Jufeng Yang, Minjing Dong, Xinghao Chen, Hanting Chen, Jianyuan Guo
arXiv AI
Oct 16

Loud and Clear: Making Visual Evidence Count in Multimodal Document RAG

The paper introduces Loud and Clear (LoC), a method that enhances multimodal document retrieval-augmented generation by fully utilizing visual token capacity and refining retrieved pages to highlight evidence. LoC combines higher-resolution pages with OCR of query-relevant regions, achieving significant accuracy gains over eight prior methods on multiple benchmarks. The study demonstrates that the way retrieved pages are presented to the generator is a critical bottleneck beyond retrieval itself.

By Eunchan Jo, Seounghyun Jang, Sangwoo Mo
arXiv AI
6d ago

Seeing Isn't Enough: Unlocking Visual Reasoning by Bridging the Reasoning-Access Gap with TIGER

The paper introduces TIGER, a training framework that addresses the modality reasoning‑access gap in multimodal large language models (MLLMs). By generating synthetic counterfactuals where models succeed on text but fail on rendered images, TIGER fine‑tunes MLLMs using Group Relative Policy Optimization, SFT, and DPO, improving performance on real‑world visual benchmarks such as MathVerse, EMMA, MMMU, ChartQAPro, and RoboSpatial‑Home. The approach demonstrates that visual reasoning can be enhanced without multimodal annotations, and it complements existing RLVR methods.

By Gregory Kang Ruey Lau, Minh Huynh Nguyen, Bryan Kian Hsiang Low
arXiv AI
Aug 26

VisCache: Visual KV Cache Pruning for Efficient Vision Large Language Model Inference

VisCache introduces a two-stage, plug‑and‑play framework for pruning visual key‑value caches in Vision Large Language Models without retraining. The first stage filters out temporally redundant keyframes, while the second stage, PruneKV, applies a parabolic layer‑wise budget and asymmetric update to selectively prune keys and fuse values, preserving essential context. Experiments show up to 2.35× speedup and significant memory savings with only 19–28% of the original cache retained, outperforming existing baselines.

By Lyuke Wang, Zhuo Li, Guangxu Zhu
arXiv Computer Vision
Sep 7

FAVE: Foveated Adaptive Visual Encoding for Efficient Fine-Grained Visual Understanding

FAVE (Foveated Adaptive Visual Encoding) is a lightweight, variable‑resolution Vision Transformer that encodes user‑selected image regions at high acuity while maintaining the image’s native geometry. In controlled experiments on small‑object ImageNet crops, FAVE outperforms a fixed‑resolution ViT by 9.4 top‑1 points while using 12.7× fewer FLOPs. When added as a local branch to FastVLM, FAVE improves TextVQA by 1.60 points and GQA attribute accuracy by 1.31 points, achieving a 3.3× speedup over SmolVLM2-2.2B with only 16 extra local tokens.

By Amitangshu Mukherjee, Kaushik Roy
arXiv AI
Oct 7

VisionWeave: Weaving Elastic Visual Representations as a Native Capability of MLLMs

VisionWeave introduces elastic visual representation weaving, a native capability for multimodal large language models that learns where and at what granularity to encode visual information. The method combines a gated spatial pooler for coarse representations with a granularity router that allocates content‑adaptive token usage, trained end‑to‑end on large‑scale data. Experiments on Qwen3.5‑4B and Qwen3.8‑27B show that VisionWeave can save 43.0% of tokens while preserving 98.9% of performance across eight benchmarks, and delivers significant throughput gains and latency reductions when deployed on the SGLang serving engine.

By Yuan Feng, Qize Yang, Ruizhe Chen, Sibo Song, Haolin He, Muzhi Zhu, Zihan Liu, Yunfei Chu, Xize Cheng, Yuxuan Wang, Jin Xu, Xike Xie