Scaling Up Your Kernels to 31×31: Revisiting Large Kernel Design in CNNs

Xiaohan Ding, Xiangyu Zhang, Jungong Han, Guiguang Ding

Research output: Other contribution


We revisit large kernel design in modern convolutional neural networks (CNNs). Inspired by recent advances in vision transformers (ViTs), in this paper, we demonstrate that using a few large convolutional kernels instead of a stack of small kernels could be a more powerful paradigm. We suggested five guidelines, e.g., applying re-parameterized large depthwise convolutions, to design efficient high-performance large-kernel CNNs. Following the guidelines, we propose RepLKNet, a pure CNN architecture whose kernel size is as large as 31×31, in contrast to commonly used 3×3. RepLKNet greatly closes the performance gap between CNNs and ViTs, e.g., achieving comparable or superior results than Swin Transformer on ImageNet and a few typical downstream tasks, with lower latency. RepLKNet also shows nice scalability to big data and large models, obtaining 87.8% top-1 accuracy on ImageNet and 56.0% mIoU on ADE20K, which is very competitive among the state-of-the-arts with similar model sizes. Our study further reveals that, in contrast to small-kernel CNNs, large-kernel CNNs have much larger effective receptive fields and higher shape bias rather than texture bias. Code & models at

Original languageEnglish
PublisherIEEE Press
Number of pages13
ISBN (Print)1063-6919, 978-1-6654-6946-3
ISBN (Electronic)9781665469463
Publication statusPublished - 2022

Publication series

NameProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
ISSN (Print)1063-6919


  • Deep learning architectures and techniques


Dive into the research topics of 'Scaling Up Your Kernels to 31×31: Revisiting Large Kernel Design in CNNs'. Together they form a unique fingerprint.

Cite this