AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

DaViT: Dual Attention Vision Transformers

Mingyu Ding, Bin Xiao, Noel Codella, Ping Luo, Jingdong Wang, Lu Yuan

ECCV 2022 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 471 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 dual attention block in stage 1 (ablation)window-attention transformer · ImageNet-1K (28.3M model)Top-1: 81.1 -> 81.7Improved
add dual attention block in stage 2 (ablation)window-attention transformer · ImageNet-1K (28.3M model)Top-1: 81.1 -> 82.2Improved
add dual attention block in stage 3 (ablation)window-attention transformer · ImageNet-1K (28.3M model)Top-1: 81.1 -> 82.1Improved
add dual attention block in stage 4 (ablation)window-attention transformer · ImageNet-1K (28.3M model)Top-1: 81.1 -> 81.9Improved
add dual attention in all four stages (ablation)window-attention transformer · ImageNet-1K (28.3M model)Top-1: 81.1 -> 82.8Improved
put channel attention before window attention instead of after (ablation)DaViT-Tiny · ImageNet-1K (28.3M model)Top-1: 82.8 -> 82.6Hurt
arrange window and channel attention in parallel instead of sequentially (ablation)DaViT-Tiny · ImageNet-1K (28.3M model)Top-1: 82.8 -> 82.6Hurt
replace half of DeiT attention blocks with channel attention (ablation)DeiT-Tiny · ImageNet-1KTop-1: 72.2 -> 74.9Improved
replace half of DeiT attention blocks with channel attention (ablation)DeiT-Base · ImageNet-1KTop-1: 81.8 -> 82.3Improved
use dual attention instead of pure window attention without FFN (ablation)Tiny w/o FFN · ImageNet-1KTop-1: 79.1 -> 80.8Improved
use dual attention instead of pure channel attention without FFN (ablation)Tiny w/o FFN · ImageNet-1KTop-1: 79.3 -> 80.8Improved
use DaViT instead of Swin at 2x fewer third-stage layers (ablation)Tiny-scale, depths · ImageNet-1KTop-1: 81.2 -> 82.8Improved

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