AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

LLaVA-UHD: an LMM Perceiving any Aspect Ratio and High-Resolution Images

Zonghao Guo, Ruyi Xu, Yuan Yao, Junbo Cui, Zanlin Ni, Chunjiang Ge, Tat-Seng Chua, Zhiyuan Liu, Gao Huang

ECCV 2024 · Vision-language and multimodal › Multimodal LLMs · approach: Training-free token reduction for MLLMs · 172 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
replace padding with adaptive aspect-ratio encoding (no padding) (ablation)LLaVA-1.5 · TextVQAaccuracy: 61.3 -> 61.8Improved
use perceiver resampler instead of 2-layer MLP projector (ablation)LLaVA-UHD · 9 benchmarks (TFLOPs)TFLOPs: 113.65 -> 14.63 TFLOPs, TextVQA 66.9 -> 67.7Improved
use LLaVA-UHD adaptive partition instead of fixed 2x2 slices (ablation)LLaVA-UHD with MLP · TextVQAaccuracy: 66.1 -> 66.9Improved
add spatial schema (separators for slice positions) (ablation)LLaVA-UHD · VQAv2accuracy: 79.3 -> 81.7Improved
add spatial schema (separators for slice positions) (ablation)LLaVA-UHD · TextVQAaccuracy: 65.4 -> 67.7Improved
update ViT during pretraining as well as instruction tuning (ablation)LLaVA-UHD · TextVQAaccuracy: 67.7 -> 62.6Hurt
update ViT during instruction tuning rather than keeping frozen (ablation)LLaVA-UHD · TextVQAaccuracy: 65.7 -> 67.7Improved
native-resolution slicing instead of fixed 336x336 inputLLaVA-1.5 · TextVQAaccuracy: 61.3 -> 67.7Improved
extreme aspect ratio images with LLaVA-UHD vs LLaVA-1.5LLaVA-UHD · VQAv2 (extreme aspect ratio)accuracy: LLaVA-1.5 74.6 (-5.4) vs LLaVA-UHD 81.4 (-0.3)Improved
replace MLP with resampler on ScienceQA (ablation)LLaVA-UHD · GQAaccuracy: 65.4 (MLP) vs 65.2Mixed

Later papers that compared with this method

1 later accepted paper reports LLaVA-UHD-v2* as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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