AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

FALCON: Resolving Visual Redundancy and Fragmentation in High-resolution Multimodal Large Language Models via Visual Registers

Renshan Zhang, Rui Shao, Gongwei Chen, Miao Zhang, Kaiwen Zhou, Weili Guan, Liqiang Nie

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Training-free token reduction for MLLMs · 13 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add ReAtten (ablation)FALCON · V*V*_Avg: 51.3 -> 61.3Improved
add ReAtten (ablation)FALCON · MME-RealWorld PerceptionAvg-C: 38.2 -> 42.1Improved
add ReAtten (ablation)FALCON · MME-RealWorld ReasoningAvg-C: 35.6 -> 39.0Improved
add ReAtten (ablation)FALCON · POPEaccuracy: 85.7 -> 87.3Improved
use ReAtten instead of W-Atten shifted window attention (ablation)FALCON · V*V*_Avg: 60.2 -> 61.3Improved
use Complementary Image Pyramid (CIP) instead of no continuity method (ablation)FALCON baseline · V*V*_Avg: 51.3 -> 50.3Hurt
use CIP instead of no continuity method (ablation)FALCON baseline · MME-RealWorld ReasoningAvg-C: 35.6 -> 37.2Improved
train FALCON on LLaVA-1.5 data instead of LLaVA-v1.5-7BFALCON · TextVQAaccuracy: 58.2 -> 61.3Improved
train FALCON on LLaVA-1.5 data instead of LLaVA-v1.5-7BFALCON · MM-Vetscore: 31.1 -> 31.4No clear effect
increase visual registers from 36 to 64 to 144 (ablation)FALCON · MME-RealWorld / V* (registers)accuracy: 1.15 and 2.41 times more training timeMixed
use ReCompact instead of pooling, Pixel Shuffle or Abstractor (ablation)FALCON · MME-RealWorld Avg-C sumAvg-CImproved
Abstractor compression instead of pooling (ablation)FALCON · MME-RealWorld Avg-C sumAvg-CHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy