AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

CvT: Introducing Convolutions to Vision Transformers

Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang

ICCV 2021 · Architectures and efficiency › Attention and transformers · approach: Hybrid and interpretable ViT designs · 2,818 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 position embedding (ablation)DeiT-S · ImageNet-1k valtop-1 accuracy (%): 78.0 -> 79.8Improved
add position embedding (every stage) (ablation)CvT-13 · ImageNet-1k valtop-1 accuracy (%): 81.6 -> 81.5No clear effect
add convolutional token embedding (no position embedding) (ablation)CvT-13 · ImageNet-1k valtop-1 accuracy (%): 80.7 -> 81.6Improved
add convolutional token embedding (with position embedding) (ablation)CvT-13 · ImageNet-1k valtop-1 accuracy (%): 81.1 -> 81.4Improved
add convolutional projection in all three stages (ablation)CvT-13 · ImageNet-1k valtop-1 accuracy (%): 80.6 -> 81.6Improved
add convolutional projection in stage 1 only (ablation)CvT-13 · ImageNet-1k valtop-1 accuracy (%): 80.6 -> 80.8Improved
add convolutional projection in stages 1 and 2 (ablation)CvT-13 · ImageNet-1k valtop-1 accuracy (%): 80.6 -> 81.0Improved
use stride 2 instead of 1 for key/value projection (ablation)CvT-13 · ImageNet-1k valtop-1 accuracy (%): 82.3 -> 81.6Mixed
neural architecture search over stride and MLP ratioCvT-13 · ImageNet-1k valtop-1 accuracy (%): 81.6 -> 82.2Improved
scale from CvT-21 to wide CvT-W24 with ImageNet-22k pretrainingCvT · ImageNet-1k valtop-1 accuracy (%): 84.9 -> 87.7Improved
raise input resolution from 224 to 384CvT-21 · ImageNet-1k valtop-1 accuracy (%): 82.5 -> 83.3Improved
ImageNet-22k pretrainingCvT-13 at 384 · ImageNet-1k valtop-1 accuracy (%): 83.0 -> 83.3No clear effect

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