AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Edge-RecViT: Efficient Vision Transformer via Semantic-Refined Dynamic Recursion

YiZhou Li, Jinyi Xu, Mingyu Yin, Xianyi Zhao

CVPR 2026 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers

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 the edge-aware ranker (EARR) to the fixed-depth shared model (ablation)Edge-RecViT Base · ImageNet-1KFLOPs (G): 35.2 -> 24.4 FLOPs, Top-1 82.1 -> 82.0Improved
share only middle layers instead of using fully independent layers (ablation)Edge-RecViT Base · ImageNet-1KTop-1 accuracy (%): 81.4 -> 82.0 (params 86M -> 23M)Improved
share a single layer across the whole transformer instead of only the middle (ablation)Edge-RecViT Base · ImageNet-1KTop-1 accuracy (%): 82.0 -> 68.9Hurt
add STB-loss to ranker regularization (ablation)Edge-RecViT Base · ImageNet-1KTop-1 accuracy (%): 69.2 -> 82.0Improved
add DU-loss to ranker regularization (ablation)Edge-RecViT Base · ImageNet-1KTop-1 accuracy (%): 61.1 -> 82.0 (FLOPs 9.5 -> 24.39)Improved
add both STB and DU regularization (ablation)Edge-RecViT Base · ImageNet-1KTop-1 accuracy (%): 55.7 -> 82.0Improved
Edge-RecViT versus DeiT on ImageNet-1K at BaseDeiT-B · ImageNet-1KTop-1 accuracy (%): 81.8 -> 82.0 (86.0M -> 23.2M params, 35.2 -> 24.39 GFLOPs)Improved
Edge-RecViT versus DeiT on ImageNet-1K at TinyDeiT-Ti · ImageNet-1KTop-1 accuracy (%): 72.2 -> 72.4 (5.8M -> 1.6M params)Improved
Edge-RecViT versus DeiT-Ti on CIFAR-100DeiT-Ti · CIFAR-100Top-1 accuracy (%): 86.61 -> 90.11Improved
Edge-RecViT versus DeiT-B on CIFAR-100DeiT-B · CIFAR-100Top-1 accuracy (%): 90.8 -> 90.78No 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