Cover Image

Seeing Beyond Redundancy: Task Complexity's Role in Vision Token Specialization in VLLMs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Hannan, Darryl, Cooper, John, White, Dylan, Watkins, Yijing
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Foundation Models for Remote Sensing: An Analysis of MLLMs for Object Localization
by: Hannan, Darryl, et al.
Published: (2025)

FMG-Det: Foundation Model Guided Robust Object Detection
by: Hannan, Darryl, et al.
Published: (2025)

Event-to-Video Conversion for Overhead Object Detection
by: Hannan, Darryl, et al.
Published: (2024)

When Token Pruning is Worse than Random: Understanding Visual Token Information in VLLMs
by: Wang, Yahong, et al.
Published: (2025)

Visual and Textual Prompts in VLLMs for Enhancing Emotion Recognition
by: Wang, Zhifeng, et al.
Published: (2025)

Line of Sight: On Linear Representations in VLLMs
by: Rajaram, Achyuta, et al.
Published: (2025)

Seeing Beyond Views: Multi-View Driving Scene Video Generation with Holistic Attention
by: Lu, Hannan, et al.
Published: (2024)

CAD-Assistant: Tool-Augmented VLLMs as Generic CAD Task Solvers
by: Mallis, Dimitrios, et al.
Published: (2024)

Thought Graph Traversal for Test-time Scaling in Chest X-ray VLLMs
by: Yao, Yue, et al.
Published: (2025)

Prune Redundancy, Preserve Essence: Vision Token Compression in VLMs via Synergistic Importance-Diversity
by: Fang, Zhengyao, et al.
Published: (2026)

Visual-Word Tokenizer: Beyond Fixed Sets of Tokens in Vision Transformers
by: Gee, Leonidas, et al.
Published: (2024)

RedundancyLens: Revealing and Exploiting Visual Token Processing Redundancy for Efficient Decoder-Only MLLMs
by: Li, Hongliang, et al.
Published: (2025)

Leveraging Gaze and Set-of-Mark in VLLMs for Human-Object Interaction Anticipation from Egocentric Videos
by: Materia, Daniele, et al.
Published: (2026)

Seeing Beyond the Scene: Enhancing Vision-Language Models with Interactional Reasoning
by: Liang, Dayong, et al.
Published: (2025)

Semantic-aware DropSplat: Adaptive Pruning of Redundant Gaussians for 3D Aerial-View Segmentation
by: Tang, Xu, et al.
Published: (2025)

ERASE: Eliminating Redundant Visual Tokens via Adaptive Two-Stage Token Pruning
by: Lee, Yuna, et al.
Published: (2026)

Waste-Bench: A Comprehensive Benchmark for Evaluating VLLMs in Cluttered Environments
by: Ali, Muhammad, et al.
Published: (2025)

A Closed-Form Solution for Debiasing Vision-Language Models with Utility Guarantees Across Modalities and Tasks
by: Lian, Tangzheng, et al.
Published: (2026)

VLLMs Provide Better Context for Emotion Understanding Through Common Sense Reasoning
by: Xenos, Alexandros, et al.
Published: (2024)

Cached Adaptive Token Merging: Dynamic Token Reduction and Redundant Computation Elimination in Diffusion Model
by: Saghatchian, Omid, et al.
Published: (2025)

Delta-LLaVA: Base-then-Specialize Alignment for Token-Efficient Vision-Language Models
by: Zamini, Mohamad, et al.
Published: (2025)

TinyChemVL: Advancing Chemical Vision-Language Models via Efficient Visual Token Reduction and Complex Reaction Tasks
by: Zhao, Xuanle, et al.
Published: (2025)

VLTP: Vision-Language Guided Token Pruning for Task-Oriented Segmentation
by: Chen, Hanning, et al.
Published: (2024)

Saddle-Free Guidance: Improved On-Manifold Sampling without Labels or Additional Training
by: Yeats, Eric, et al.
Published: (2025)

Efficient Video Sampling: Pruning Temporally Redundant Tokens for Faster VLM Inference
by: Bagrov, Natan, et al.
Published: (2025)

TimeChat-Online: 80% Visual Tokens are Naturally Redundant in Streaming Videos
by: Yao, Linli, et al.
Published: (2025)

Beyond Intermediate States: Explaining Visual Redundancy through Language
by: Yang, Dingchen, et al.
Published: (2025)

See the Text: From Tokenization to Visual Reading
by: Xing, Ling, et al.
Published: (2025)

Beyond Surrogate Gradients: Fully Differentiable Token Pruning for Vision-Language Models
by: He, Landi, et al.
Published: (2026)

Decoupled Similarity for Task-Aware Token Pruning in Large Vision-Language Models
by: Ma, Kexin, et al.
Published: (2026)

Patch-as-Decodable-Token: Towards Unified Multi-Modal Vision Tasks in MLLMs
by: Su, Yongyi, et al.
Published: (2025)

Faster Parameter-Efficient Tuning with Token Redundancy Reduction
by: Kim, Kwonyoung, et al.
Published: (2025)

Leveraging Vision Language Models for Specialized Agricultural Tasks
by: Arshad, Muhammad Arbab, et al.
Published: (2024)

Redundancy-Aware Pretraining of Vision-Language Foundation Models in Remote Sensing
by: Adler, Mathis Jürgen, et al.
Published: (2025)

Beyond Degradation Redundancy: Contrastive Prompt Learning for All-in-One Image Restoration
by: Wu, Gang, et al.
Published: (2025)

Vision-Language Models Can't See the Obvious
by: Dahou, Yasser, et al.
Published: (2025)

ReVisionLLM: Recursive Vision-Language Model for Temporal Grounding in Hour-Long Videos
by: Hannan, Tanveer, et al.
Published: (2024)

Adaptive-VoCo: Complexity-Aware Visual Token Compression for Vision-Language Models
by: Guo, Xiaoyang, et al.
Published: (2025)

Seeing Beyond Haze: Generative Nighttime Image Dehazing
by: Lin, Beibei, et al.
Published: (2025)

How Do Large Vision-Language Models See Text in Image? Unveiling the Distinctive Role of OCR Heads
by: Baek, Ingeol, et al.
Published: (2025)