目录
1.1 Decomposition and Combination of Convolution
1.2 Sparse Dependencies of Large Convolution Kernels
1.3 Intermodule Feature Manipulation
2.3 创建添加优化点模块的yolov5x-C3-Shift-ConvNets.yaml
1 Shift-ConvNets模块原理
Shift-ConvNets: Small Convolutional Kernelwith Large Kernel Effectshttps://arxiv.org/pdf/2401.12736最近的研究表明,Vision transformers (ViTs)的卓越性能得益于大的感受野。因此,大型卷积核设计成为使卷积神经网络(CNNs)再次变得出色的理想解决方案。然而&#x