mlsanityData to 13 Oct 2026

Blink: Dynamic Visual Token Resolution for Enhanced Multimodal Understanding

arXiv 2512.10548 ↗First posted 11 Dec 2025cs.CVcited by 2

On the map: Multimodal LLMs / vision-language models → MLLM training and visual instruction tuning

Abstract (arXiv)

Multimodal large language models (MLLMs) have achieved remarkable progress on various vision-language tasks, yet their visual perception remains limited. Humans, in comparison, perceive complex scenes efficiently by dynamically scanning and focusing on salient regions in a sequential "blink-like" process. Motivated by this strategy, we first investigate whether MLLMs exhibit similar behavior. Our pilot analysis reveals that MLLMs naturally attend to different visual regions across layers and that selectively allocating more computation to salient tokens can enhance visual perception. Building on this insight, we propose Blink, a dynamic visual token resolution framework that emulates the human-inspired process within a single forward pass. Specifically, Blink includes two modules: saliency-guided scanning and dynamic token resolution. It first estimates the saliency of visual tokens in each layer based on the attention map, and extends important tokens through a plug-and-play token super-resolution (TokenSR) module. In the next layer, it drops the extended tokens when they lose focus. This dynamic mechanism balances broad exploration and fine-grained focus, thereby enhancing visual perception adaptively and efficiently. Extensive experiments validate Blink, demonstrating its effectiveness in enhancing visual perception and multimodal understanding.

Key facts. Cited by 2 papers in the map (data through 13 Oct 2026). Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 41 of its 51 references that are in the map. First posted on arXiv 11 Dec 2025.

Builds on

41 of 51 references.

Closest ancestors · 9

Same direction, newest first

  1. 2025Introducing Visual Perception Token into Multimodal Large Language Model24 Feb
  2. 2025MLLMs Know Where to Look: Training-free Perception of Small Visual Details with Multimodal LLMs24 Feb
  3. 2024LLaVA-OneVision: Easy Visual Task Transfer6 Aug
  4. 2024Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs11 Jan
  5. 2023InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks21 Dec
  6. 2023Improved Baselines with Visual Instruction Tuning5 Oct
  7. 2023MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilities4 Aug
  8. 2023MMBench: Is Your Multi-modal Model an All-around Player?12 Jul
  9. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models23 Jun

Other references · 32

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,588
  2. 2023GPT-4 Technical Reportcited by 18,278
  3. 2014Microsoft COCO: Common Objects in Contextcited by 17,550
  4. 2023Visual Instruction Tuningcited by 7,827
  5. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,103
  6. 2025Qwen2.5-VL Technical Reportcited by 5,499
  7. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,363
  8. 2021Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervisioncited by 3,855
  9. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,197
  10. 2016Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotationscited by 2,649
  11. 2023Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyondcited by 2,204
  12. 2022Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answeringcited by 1,896
20 more
  1. 2023Evaluating Object Hallucination in Large Vision-Language Modelscited by 1,709
  2. 2019Towards VQA Models That Can Readcited by 1,705
  3. 2014Image Super-Resolution Using Deep Convolutional Networkscited by 1,221
  4. 2015Accurate Image Super-Resolution Using Very Deep Convolutional Networkscited by 902
  5. 2024An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Modelscited by 629
  6. 2023V*: Guided Visual Search as a Core Mechanism in Multimodal LLMscited by 470
  7. 2024Hallucination of Multimodal Large Language Models: A Surveycited by 384
  8. 2024PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reductioncited by 284
  9. 2024Multi-Modal Hallucination Control by Visual Information Groundingcited by 200
  10. 2024Fit and Prune: Fast and Training-free Visual Token Pruning for Multi-modal Large Language Modelscited by 125
  11. 2023Hallucination Augmented Contrastive Learning for Multimodal Large Language Modelcited by 120
  12. 2025Grounded Chain-of-Thought for Multimodal Large Language Modelscited by 58
  13. 2025Inner Thinking Transformer: Leveraging Dynamic Depth Scaling to Foster Adaptive Internal Thinkingcited by 57
  14. 2025DyFo: A Training-Free Dynamic Focus Visual Search for Enhancing LMMs in Fine-Grained Visual Understandingcited by 48
  15. 2024NACL: A General and Effective KV Cache Eviction Framework for LLMs at Inference Timecited by 43
  16. 2024VL-Cache: Sparsity and Modality-Aware KV Cache Compression for Vision-Language Model Inference Accelerationcited by 41
  17. 2025Open Eyes, Then Reason: Fine-grained Visual Mathematical Understanding in MLLMscited by 16
  18. 2025GeoPQA: Bridging the Visual Perception Gap in MLLMs for Geometric Reasoningcited by 8
  19. 2025DIVE into MoE: Diversity-Enhanced Reconstruction of Large Language Models from Dense into Mixture-of-Expertscited by 7
  20. 2025BeamLoRA: Beam-Constraint Low-Rank Adaptationcited by 3

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Beyond Static Cropping: Layer-Adaptive Visual Localization and Decoding Enhancementcited by 1
  2. 2026Can LoRA Fusion Support Cross-Domain Tasks in Cloud-Edge Collaboration?cited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Learning to See What You Need: Gaze Attention for Multimodal Large Language Models
  2. 2025HERO: Rethinking Visual Token Early Dropping in High-Resolution Large Vision-Language Models
  3. 2025BLINK-Twice: You see, but do you observe? A Reasoning Benchmark on Visual Perception
  4. 2025Perceiving Beyond Language Priors: Enhancing Visual Comprehension and Attention in Multimodal Models
  5. 2024BLINK: Multimodal Large Language Models Can See but Not Perceive
  6. 2025Pyramid Token Pruning for High-Resolution Large Vision-Language Models via Region, Token, and Instruction-Guided Importance
  7. 2025VScan: Rethinking Visual Token Reduction for Efficient Large Vision-Language Models
  8. 2026How Many Visual Tokens Do Multimodal Language Models Need? Scaling Visual Token Pruning with F^3A