AI Paper Atlas

Research map / Detection and segmentation / CNN features

ResFormer: Scaling ViTs With Multi-Resolution Training

Rui Tian, Zuxuan Wu, Qi Dai, Han Hu, Yu Qiao, Yu-Gang Jiang

CVPR 2023 · Detection and segmentation › CNN features · approach: Plain and high-resolution ViTs · 30 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
Train on multiple resolutions with global-local positional embeddingsResFormer-T · ImageNet-1K, test 96Top-1 accuracy: 8.06 -> 61.40Improved
Train on multiple resolutions with global-local positional embeddingsResFormer-S · ImageNet-1K, test 224Top-1 accuracy: 79.83 -> 82.16Improved
Train on multiple resolutions with global-local positional embeddingsResFormer-B · ImageNet-1K, test 640Top-1 accuracy: 74.23 -> 81.72Improved
Use multi-resolution pretrainingResFormer-S · ADE20KmIoU: 45.47 -> 46.31Improved
Use multi-resolution pretrainingResFormer-B · COCO2017 mini-valbox AP: +1.3 box APImproved
Add cross-scale distillation (ablation)ResFormer-S-MR · ImageNet-1K, test 224Top-1 accuracy: 81.77 -> 82.16Improved
Use one randomly selected resolution per iteration instead of joint multi-resolution batches (ablation)ResFormer-S-MR · ImageNet-1K, test 224Top-1 accuracy: 80.32 -> 82.16Hurt
Use one resolution per epoch instead of joint multi-resolution batches (ablation)ResFormer-S-MR · ImageNet-1K, test 128Top-1 accuracy: 75.18 -> 78.24Hurt
Combine global and local positional embeddings (ablation)ResFormer-S-MR · ImageNet-1K, test 224Top-1 accuracy: 81.42 -> 82.16Improved
Use class-token distillation with smooth L1 rather than logit distillation with KL (ablation)ResFormer-S-MR · ImageNet-1K, test 224Top-1 accuracy: 78.82 -> 79.68Improved
Use class-token distillation with smooth L1 rather than class-token L2 (ablation)ResFormer-S-MR · ImageNet-1K, test 224Top-1 accuracy: 79.33 -> 79.68Improved
Apply separate random augmentation at each resolution (ablation)ResFormer-S-MR · ImageNet-1K, test 128Top-1 accuracy: 77.50 -> 78.24Improved

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