AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Scaling Up Your Kernels: Large Kernel Design in ConvNets Toward Universal Representations

Yiyuan Zhang, Xiaohan Ding, Xiangyu Yue

TPAMI 2025 · Architectures and efficiency › CNN design · approach: Large and dynamic kernel design · 38 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
remove identity shortcut when using 13x13 depth-wise kernels (ablation)MobileNet V2 · ImageNet-1Ktop-1 accuracy: 72.53 -> 53.98Hurt
replace 3x3 with 13x13 kernels when shortcuts exist (ablation)MobileNet V2 · ImageNet-1Ktop-1 accuracy: 71.76 -> 72.53Improved
add parallel 3x3 re-parameterization to 13x13 kernel (ablation)MobileNet V2 · CityscapesmIoU: 75.67 -> 76.60Improved
add parallel 3x3 re-parameterization to 13x13 kernel (ablation)MobileNet V2 · ImageNet-1Ktop-1 accuracy: 72.53 -> 73.24Improved
enlarge last-stage kernels from 3x3 to 13x13 (ablation)MobileNet V2 · CityscapesmIoU: 72.31 -> 74.62Improved
use Dilated Reparam Block instead of no re-parameterization (ablation)vanilla large-kernel ConvNet · ADE20KmIoU: 45.78 -> 46.37Improved
use Dilated Reparam instead of same-kernel-size parallel branches (ablation)vanilla large-kernel ConvNet · ADE20KmIoU: 46.07 -> 46.37Improved
add SE Block as extra depth structure (ablation)vanilla large-kernel ConvNet · ADE20KmIoU: 45.1 -> 46.5Improved
shrink kernels in stages 2-4 from 13 to 11 (ablation)vanilla large-kernel ConvNet · ADE20KmIoU: 46.5 -> 45.5Hurt
use 13x13 kernels in stage 1 instead of 3x3 (ablation)vanilla large-kernel ConvNet · ADE20KmIoU: 46.5 -> 44.9Hurt
add 3x3 DW conv to small-kernel blocks when scaling depth to 27 (ablation)vanilla large-kernel ConvNet · ADE20KmIoU: 47.8 -> 48.8Improved
use 27 large-kernel blocks instead of 9 large plus 18 small blocks (ablation)vanilla large-kernel ConvNet · ADE20KmIoU: 48.8 -> 49.0No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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