(19.04) Res2Net: A New Multi-scale Backbone Architecture
代码地址：https://arxiv.org/pdf/1904.01169.pdf
代码地址：https://mmcheng.net/res2net
(19.04) OctaveConv: Drop an Octave: Reducing Spatial Redundancy in Convolutional Neural Networks with Octave Convolution
论文地址: https://arxiv.org/abs/1904.05049
代码地址: https://github.com/terrychenism/OctaveConv
(19.04) ScaleNet:Data-Driven Neuron Allocation for Scale Aggregation Networks
论文地址: https://arxiv.org/abs/1904.09460
代码地址: https://github.com/Eli-YiLi/ScaleNet
(19.04) Data-Driven Neuron Allocation for Scale Aggregation Networks
论文地址：https://arxiv.org/pdf/1904.09460.pdf
代码地址：https://github.com/Eli-YiLi/ScaleNet
(19.07) MixConv: Mixed Depthwise Convolutional Kernels
论文地址：https://arxiv.org/pdf/1907.09595.pdf
代码地址：https://github.com/tensorflow/tpu/tree/master/models/official/mnasnet/mixnet
(19.08) Drop an Octave: Reducing Spatial Redundancy in Convolutional Neural Networks with Octave Convolution
论文地址：https://arxiv.org/pdf/1904.05049.pdf
(19.09) ACNet: Strengthening the Kernel Skeletons for Powerful CNN via Asymmetric Convolution Blocks
论文地址：https://arxiv.org/pdf/1908.03930.pdf
(19.10) CondConv: Conditionally Parameterized Convolutions for Efficient Inference
论文地址：https://arxiv.org/pdf/1904.04971.pdf
代码地址：https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet/condconv
(20.02) XSepConv: Extremely Separated Convolution
论文地址：https://arxiv.org/pdf/2002.12046.pdf
(20.03) Rethinking Depthwise Separable Convolutions: How Intra-Kernel Correlations Lead to Improved MobileNets
论文地址：https://arxiv.org/pdf/2003.13549.pdf
(20.03) Dynamic Convolution: Attention over Convolution Kernels
论文地址：https://arxiv.org/pdf/1912.03458.pdf
(20.04) MUXConv: Information Multiplexing in Convolutional Neural Networks
论文地址：https://arxiv.org/pdf/2003.13880.pdf
代码地址：https://github.com/ human-analysis/MUXConv
(20.04) Exploring self-attention for image recognition
论文地址：https://hszhao.github.io/papers/cvpr20_san.pdf
代码地址：https://github.com/hszhao/SAN
(20.04) DYNET: DYNAMIC CONVOLUTION FOR ACCELERATING CONVOLUTIONAL NEURAL NETWORKS
论文地址：https://arxiv.org/pdf/2004.10694.pdf
(20.05) Improving Convolutional Networks with Self-calibrated Convolutions
论文地址：http://mftp.mmcheng.net/Papers/20cvprSCNet.pdf
代码地址：https://github.com/MCG-NKU/SCNet
(20.06)Pyramidal Convolution:Rethinking Convolutional Neural Networks for Visual Recognition
论文地址：https://arxiv.org/pdf/2006.11538.pdf
代码地址：https://github.com/iduta/pyconv
(20.06) DO-Conv: Depthwise Over-parameterized Convolutional Layer
论文地址：https://arxiv.org/pdf/2006.12030.pdf
代码地址：https://github.com/yangyanli/DO-Conv
