paper-with-me

홈 › Papers

ShiftwiseConv: Small Convolutional Kernel with Large Kernel Effect

2025-01-01 · CVPR 2025 1 · Dachong Li, Li Li, Zhuangzhuang Chen, Jianqiang Li

Large kernels make standard convolutional neural networks (CNNs) great again over transformer architectures in various vision tasks. Nonetheless, recent studies meticulously designed around increasing kernel size have shown diminishing returns or stagnation in performance. Thus, the hidden factors of large kernel convolution that affect model performance remain unexplored. In this paper, we reveal that the key hidden factors of large kernels can be summarized as two separate components: extracting features at a certain granularity and fusing features by multiple pathways. To this end, we leverage the multi-path long-distance sparse dependency relationship to enhance feature utilization via the proposed Shiftwise (SW) convolution operator with a pure CNN architecture. In a wide range of vision tasks such as classification, segmentation, and detection, SW surpasses state-of-the-art transformers and CNN architectures, including SLaK and UniRepLKNet. More importantly, our experiments demonstrate that 3 x3 convolutions can replace large convolutions in existing large kernel CNNs to achieve comparable effects, which may inspire follow-up works. Code and all the models at https://github.com/lidc54/shift-wiseConv.

📄 PDF Abstract BibTeX

Code (1)

lidc54/shift-wiseconv 공식 구현 pytorch

Methods 이 논문이 사용한 방법론

Convolution A convolution is a type of matrix operation, consisting of a kernel, a small matrix of weights, that slides over input data performing element-wise multiplication with the…

Similar Papers 제목 키워드 기반

Shift-ConvNets: Small Convolutional Kernel with Large Kernel Effects

2024-01-23 · Dachong Li, Li Li, Zhuangzhuang Chen, Jianqiang Li

Recent studies reveal that the remarkable performance of Vision transformers (ViTs) benefits from large receptive fields. For this reason, the large convolutional kernel design becomes an ideal solution to make Convoluti…

Scaling Up Your Kernels to 31x31: Revisiting Large Kernel Design in CNNs

2022-03-13 · CVPR 2022 1 · Xiaohan Ding, Xiangyu Zhang, Yizhuang Zhou, Jungong Han 외

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…

Image Classification

Accelerating Large Kernel Convolutions with Nested Winograd Transformation.pdf

2021-02-26 · Jingbo Jiang, Xizi Chen, Chi-Ying Tsui

Recent literature has shown that convolutional neural networks (CNNs) with large kernels outperform vision transformers (ViTs) and CNNs with stacked small kernels in many computer vision tasks, such as object detection a…

Image Restorationobject-detectionObject DetectionSemantic Segmentation

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

2024-10-10 · Yiyuan Zhang, Xiaohan Ding, Xiangyu Yue

This paper proposes the paradigm of large convolutional kernels in designing modern Convolutional Neural Networks (ConvNets). We establish that employing a few large kernels, instead of stacking multiple smaller ones, ca…

Time Series ForecastingVideo Recognition

Are Large Kernels Better Teachers than Transformers for ConvNets?

2023-05-30 · Tianjin Huang, Lu Yin, Zhenyu Zhang, Li Shen 외

This paper reveals a new appeal of the recently emerged large-kernel Convolutional Neural Networks (ConvNets): as the teacher in Knowledge Distillation (KD) for small-kernel ConvNets. While Transformers have led state-of…

Knowledge Distillation