AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

HIRI-ViT: Scaling Vision Transformer With High Resolution Inputs

Ting Yao, Yehao Li, Yingwei Pan, Tao Mei

TPAMI 2024 · Architectures and efficiency › Attention and transformers · approach: Hybrid and interpretable ViT designs · 58 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
use CFFN with depth-wise convolution instead of FFN (ablation)M-ViT · ImageNet-1K valtop-1: 82.7 -> 82.9Improved
remove MHA in first two stages (ablation)M-ViT · ImageNet-1K valtop-1: 82.9 -> 82.9No clear effect
replace LN with BN in first two stages (ablation)M-ViT · ImageNet-1K valtop-1: 82.9 -> 83.0Improved
replace ViT-stem with Conv-stem (ablation)M-ViT · ImageNet-1K valtop-1: 83.0 -> 83.4Improved
add inverted residual downsampling (ablation)M-ViT · ImageNet-1K valtop-1: 83.4 -> 83.6Improved
increase input resolution 224 to 448 in four stages (ablation)M-ViT · ImageNet-1K valtop-1 and GFLOPs: 83.6 -> 84.6; GFLOPs 4.7 -> 21.5Mixed
use five stages instead of four at 448 input (ablation)M-ViT · ImageNet-1K valtop-1 and GFLOPs: 84.6 -> 84.0; GFLOPs 21.5 -> 5.8Mixed
replace Conv-stem with HR-stem (ablation)HIRI-ViT-S · ImageNet-1K valGFLOPs: 5.8 -> 5.1Improved
replace CFFN with two HR blocks in first two stages (ablation)HIRI-ViT-S · ImageNet-1K valtop-1: 84.0 -> 84.1Improved
add EMA distillation (ablation)HIRI-ViT-S · ImageNet-1K valtop-1: 84.1 -> 84.3Improved
use three Transformer stages instead of one at the end (ablation)HIRI-ViT-S · ImageNet-1K valtop-1: 83.0 -> 84.3Improved
apply HIRI-ViT scaling to DaViT-Tiny at 448 (ablation)DaViT-Tiny · ImageNet-1K valtop-1: 82.8 -> 83.3Improved

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