arXiv 2403.075892024-03-16cs.CV

PeLK：基于周边卷积的参数高效大核卷积网络

PeLK: Parameter-efficient Large Kernel ConvNets with Peripheral Convolution

Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）；School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院）；Meituan（美团）；CAS Center for Excellence in Brain Science and Intelligence Technology（中国科学院脑科学与智能技术卓越创新中心）

Honghao Chen, Xiangxiang Chu, Yongjian Ren, Xin Zhao, Kaiqi Huang

备注：: CVPR 2024; Modification for Fig.1(b); Add Acknowledgements

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出类人周边卷积，将卷积复杂度从O(K²)降至O(logK)，减少90%以上参数，并基于此构建参数高效大核网络PeLK，首次将核尺寸扩展到101x101，在多个视觉任务上超越现有架构。

摘要：

近年来，一些大核卷积网络以出色的性能和效率强势回归。然而，鉴于卷积的平方复杂度，扩大核会带来大量的参数，且激增的参数会引发严重的优化问题。由于这些问题，当前的卷积神经网络以条带卷积的形式（即51x5+5x51）妥协地扩展到51x51，并随着核尺寸的持续增长开始饱和。在本文中，我们深入探讨解决这些关键问题，并探索我们能否继续扩大核以获得更多性能增益。受人类视觉启发，我们提出了一种类人的周边卷积，通过参数共享有效减少了密集网格卷积超过90%的参数数量，并成功将核尺寸扩展到极大。我们的周边卷积行为与人类高度相似，将卷积复杂度从O(K^2)降至O(logK)，且不会损害性能。基于此，我们提出了参数高效大核网络（PeLK）。我们的PeLK在各种视觉任务上优于现代视觉Transformer和ConvNet架构，如Swin、ConvNeXt、RepLKNet和SLaK，包括ImageNet分类、ADE20K上的语义分割和MS COCO上的目标检测。我们首次成功将CNN的核尺寸扩展到前所未有的101x101，并展示了持续的性能提升。

英文摘要：

Recently, some large kernel convnets strike back with appealing performance and efficiency. However, given the square complexity of convolution, scaling up kernels can bring about an enormous amount of parameters and the proliferated parameters can induce severe optimization problem. Due to these issues, current CNNs compromise to scale up to 51x51 in the form of stripe convolution (i.e., 51x5 + 5x51) and start to saturate as the kernel size continues growing. In this paper, we delve into addressing these vital issues and explore whether we can continue scaling up kernels for more performance gains. Inspired by human vision, we propose a human-like peripheral convolution that efficiently reduces over 90% parameter count of dense grid convolution through parameter sharing, and manage to scale up kernel size to extremely large. Our peripheral convolution behaves highly similar to human, reducing the complexity of convolution from O(K^2) to O(logK) without backfiring performance. Built on this, we propose Parameter-efficient Large Kernel Network (PeLK). Our PeLK outperforms modern vision Transformers and ConvNet architectures like Swin, ConvNeXt, RepLKNet and SLaK on various vision tasks including ImageNet classification, semantic segmentation on ADE20K and object detection on MS COCO. For the first time, we successfully scale up the kernel size of CNNs to an unprecedented 101x101 and demonstrate consistent improvements.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。