搜索

​CV 图像分类常见的 36 个模型汇总！附完整论文和代码

小白学视觉

共 5061字，需浏览 11分钟

 ·

2020-11-01 20:11

点击上方“小白学视觉”，选择加"星标"或“置顶”

重磅干货，第一时间送达

转载 | 新机器视觉

今天给大家介绍自 2014 年以来，计算机视觉 CV 领域图像分类方向文献和代码的超全总结和列表！总共涉及 36 种 ConvNet 模型。该 GitHub 项目作者是 weiaicunzai，项目地址是：


https://github.com/weiaicunzai/awesome-image-classification


背景


我相信图像识别是深入到其它机器视觉领域一个很好的起点，特别是对于刚刚入门深度学习的人来说。当我初学 CV 时，犯了很多错。我当时非常希望有人能告诉我应该从哪一篇论文开始读起。到目前为止，似乎还没有一个像 deep-learning-object-detection 这样的 GitHub 项目。因此，我决定建立一个 GitHub 项目，列出深入学习中关于图像分类的论文和代码，以帮助其他人。


对于学习路线，我的个人建议是，对于那些刚入门深度学习的人，可以试着从 vgg 开始，然后是 googlenet、resnet，之后可以自由地继续阅读列出的其它论文或切换到其它领域。


性能表


基于简化的目的，我只从论文中列举出在 ImageNet 上准确率最高的 top1 和 top5。注意，这并不一定意味着准确率越高，一个网络就比另一个网络更好。因为有些网络专注于降低模型复杂性而不是提高准确性，或者有些论文只给出 ImageNet 上的 single crop results，而另一些则给出模型融合或 multicrop results。


关于性能表的标注：


  • ConvNet：卷积神经网络的名称

  • ImageNet top1 acc：论文中基于 ImageNet 数据集最好的 top1 准确率

  • ImageNet top5 acc：论文中基于 ImageNet 数据集最好的 top5 准确率

  • Published In：论文发表在哪个会议或期刊



论文&代码


1. VGG


Very Deep Convolutional Networks for Large-Scale Image Recognition. 

Karen Simonyan, Andrew Zisserman 


pdf: https://arxiv.org/abs/1409.1556 

code: torchvision : 

https://github.com/pytorch/vision/blob/master/torchvision/models/vgg.py 

code: keras-applications : 

https://github.com/keras-team/keras-applications/blob/master/keras_applications/vgg16.py 

code: keras-applications : 

https://github.com/keras-team/keras-applications/blob/master/keras_applications/vgg19.py


2. GoogleNet


Going Deeper with Convolutions 

Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke, Andrew Rabinovich 


pdf: https://arxiv.org/abs/1409.4842 

code: unofficial-tensorflow : 

https://github.com/conan7882/GoogLeNet-Inception 

code: unofficial-caffe : 

https://github.com/lim0606/caffe-googlenet-bn


3. PReLU-nets 


Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification 

Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun 


pdf: https://arxiv.org/abs/1502.01852 

code: unofficial-chainer : 

https://github.com/nutszebra/prelu_net


4. ResNet 


Deep Residual Learning for Image Recognition 

Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun 


pdf: https://arxiv.org/abs/1512.03385 

code: facebook-torch : 

https://github.com/facebook/fb.resnet.torch 

code: torchvision : 

https://github.com/pytorch/vision/blob/master/torchvision/models/resnet.py 

code: keras-applications : 

https://github.com/keras-team/keras-applications/blob/master/keras_applications/resnet.py 

code: unofficial-keras : 

https://github.com/raghakot/keras-resnet 

code: unofficial-tensorflow : 

https://github.com/ry/tensorflow-resnet


5. PreActResNet 


Identity Mappings in Deep Residual Networks 

Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun 


pdf: https://arxiv.org/abs/1603.05027 

code: facebook-torch : 

https://github.com/facebook/fb.resnet.torch/blob/master/models/preresnet.lua 

code: official : 

https://github.com/KaimingHe/resnet-1k-layers 

code: unoffical-pytorch : 

https://github.com/kuangliu/pytorch-cifar/blob/master/models/preact_resnet.py 

code: unoffical-mxnet : 

https://github.com/tornadomeet/ResNet


6. Inceptionv3

 

Rethinking the Inception Architecture for Computer Vision 

Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jonathon Shlens, Zbigniew Wojna 


pdf: https://arxiv.org/abs/1512.00567 

code: torchvision : 

https://github.com/pytorch/vision/blob/master/torchvision/models/inception.py 

code: keras-applications : 

https://github.com/keras-team/keras-applications/blob/master/keras_applications/inception_v3.py


7. Inceptionv4 && Inception-ResNetv2 


Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning 

Christian Szegedy, Sergey Ioffe, Vincent Vanhoucke, Alex Alemi 


pdf: https://arxiv.org/abs/1602.07261 

code: unofficial-keras : 

https://github.com/kentsommer/keras-inceptionV4 

code: unofficial-keras : 

https://github.com/titu1994/Inception-v4 

code: unofficial-keras : 

https://github.com/yuyang-huang/keras-inception-resnet-v2


8. RIR


Resnet in Resnet: Generalizing Residual Architectures 

Sasha Targ, Diogo Almeida, Kevin Lyman 


pdf: https://arxiv.org/abs/1603.08029 

code: unofficial-tensorflow : 

https://github.com/SunnerLi/RiR-Tensorflow 

code: unofficial-chainer : 

https://github.com/nutszebra/resnet_in_resnet


9. Stochastic Depth ResNet 


Deep Networks with Stochastic Depth 

Gao Huang, Yu Sun, Zhuang Liu, Daniel Sedra, Kilian Weinberger 


pdf: https://arxiv.org/abs/1603.09382 

code: unofficial-torch : 

https://github.com/yueatsprograms/Stochastic_Depth 

code: unofficial-chainer : 

https://github.com/yasunorikudo/chainer-ResDrop 

code: unofficial-keras : 

https://github.com/dblN/stochastic_depth_keras


10. WRN 


Wide Residual Networks 

Sergey Zagoruyko, Nikos Komodakis 


pdf: https://arxiv.org/abs/1605.07146 

code: official : 

https://github.com/szagoruyko/wide-residual-networks 

code: unofficial-pytorch : 

https://github.com/xternalz/WideResNet-pytorch 

code: unofficial-keras : 

https://github.com/asmith26/wide_resnets_keras 

code: unofficial-pytorch : 

https://github.com/meliketoy/wide-resnet.pytorch


11. squeezenet 


SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size 

Forrest N. Iandola, Song Han, Matthew W. Moskewicz, Khalid Ashraf, William J. Dally, Kurt Keutzer 


pdf: https://arxiv.org/abs/1602.07360 

code: torchvision : 

https://github.com/pytorch/vision/blob/master/torchvision/models/squeezenet.py 

code: unofficial-caffe : 

https://github.com/DeepScale/SqueezeNet 

code: unofficial-keras : 

https://github.com/rcmalli/keras-squeezenet 

code: unofficial-caffe : 

https://github.com/songhan/SqueezeNet-Residual


12. GeNet 


Genetic CNN 

Lingxi Xie, Alan Yuille 


pdf: https://arxiv.org/abs/1703.01513 

code: unofficial-tensorflow : 

https://github.com/aqibsaeed/Genetic-CNN


12. MetaQNN 


Designing Neural Network Architectures using Reinforcement Learning

Bowen Baker, Otkrist Gupta, Nikhil Naik, Ramesh Raskar 


pdf: https://arxiv.org/abs/1703.01513 

code: official : https://github.com/bowenbaker/metaqnn


13. PyramidNet 


Deep Pyramidal Residual Networks 

Dongyoon Han, Jiwhan Kim, Junmo Kim 


pdf: https://arxiv.org/abs/1610.02915 

code: official : 

https://github.com/jhkim89/PyramidNet 

code: unofficial-pytorch : 

https://github.com/dyhan0920/PyramidNet-PyTorch


14. DenseNet 


Densely Connected Convolutional Networks 

Gao Huang, Zhuang Liu, Laurens van der Maaten, Kilian Q. Weinberger 


pdf: https://arxiv.org/abs/1608.06993 

code: official : 

https://github.com/liuzhuang13/DenseNet 

code: unofficial-keras : 

https://github.com/titu1994/DenseNet 

code: unofficial-caffe : 

https://github.com/shicai/DenseNet-Caffe 

code: unofficial-tensorflow : 

https://github.com/YixuanLi/densenet-tensorflow 

code: unofficial-pytorch : 

https://github.com/YixuanLi/densenet-tensorflow 

code: unofficial-pytorch : 

https://github.com/bamos/densenet.pytorch 

code: unofficial-keras : 

https://github.com/flyyufelix/DenseNet-Keras


15. FractalNet 


FractalNet: Ultra-Deep Neural Networks without Residuals 

Gustav Larsson, Michael Maire, Gregory Shakhnarovich 


pdf: https://arxiv.org/abs/1605.07648 

code: unofficial-caffe : 

https://github.com/gustavla/fractalnet 

code: unofficial-keras : 

https://github.com/snf/keras-fractalnet 

code: unofficial-tensorflow : 

https://github.com/tensorpro/FractalNet


16. ResNext 


Aggregated Residual Transformations for Deep Neural Networks 

Saining Xie, Ross Girshick, Piotr Dollár, Zhuowen Tu, Kaiming He 


pdf: https://arxiv.org/abs/1611.05431 

code: official : 

https://github.com/facebookresearch/ResNeXt 

code: keras-applications : 

https://github.com/keras-team/keras-applications/blob/master/keras_applications/resnext.py 

code: unofficial-pytorch : 

https://github.com/prlz77/ResNeXt.pytorch 

code: unofficial-keras : 

https://github.com/titu1994/Keras-ResNeXt 

code: unofficial-tensorflow : 

https://github.com/taki0112/ResNeXt-Tensorflow 

code: unofficial-tensorflow : 

https://github.com/wenxinxu/ResNeXt-in-tensorflow


17. IGCV1 


Interleaved Group Convolutions for Deep Neural Networks 

Ting Zhang, Guo-Jun Qi, Bin Xiao, Jingdong Wang 


pdf: https://arxiv.org/abs/1707.02725 

code official : 

https://github.com/hellozting/InterleavedGroupConvolutions


18. Residual Attention Network 


Residual Attention Network for Image Classification 

Fei Wang, Mengqing Jiang, Chen Qian, Shuo Yang, Cheng Li, Honggang Zhang, Xiaogang Wang, Xiaoou Tang 


pdf: https://arxiv.org/abs/1704.06904 

code: official : 

https://github.com/fwang91/residual-attention-network 

code: unofficial-pytorch : 

https://github.com/tengshaofeng/ResidualAttentionNetwork-pytorch 

code: unofficial-gluon : 

https://github.com/PistonY/ResidualAttentionNetwork 

code: unofficial-keras : 

https://github.com/koichiro11/residual-attention-network


19. Xception 


Xception: Deep Learning with Depthwise Separable Convolutions

François Chollet 


pdf: https://arxiv.org/abs/1610.02357 

code: unofficial-pytorch : 

https://github.com/jfzhang95/pytorch-deeplab-xception/blob/master/modeling/backbone/xception.py 

code: unofficial-tensorflow : 

https://github.com/kwotsin/TensorFlow-Xception 

code: unofficial-caffe : 

https://github.com/yihui-he/Xception-caffe 

code: unofficial-pytorch : 

https://github.com/tstandley/Xception-PyTorch 

code: keras-applications : 

https://github.com/keras-team/keras-applications/blob/master/keras_applications/xception.py


20. MobileNet 

MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications 

Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam 


pdf: https://arxiv.org/abs/1704.04861 

code: unofficial-tensorflow : 

https://github.com/Zehaos/MobileNet 

code: unofficial-caffe : 

https://github.com/shicai/MobileNet-Caffe 

code: unofficial-pytorch : 

https://github.com/marvis/pytorch-mobilenet 

code: keras-applications : 

https://github.com/keras-team/keras-applications/blob/master/keras_applications/mobilenet.py


21. PolyNet 


PolyNet: A Pursuit of Structural Diversity in Very Deep Networks

Xingcheng Zhang, Zhizhong Li, Chen Change Loy, Dahua Lin 


pdf: https://arxiv.org/abs/1611.05725 


code: official : 

https://github.com/open-mmlab/polynet


22. DPN 


Dual Path Networks 

Yunpeng Chen, Jianan Li, Huaxin Xiao, Xiaojie Jin, Shuicheng Yan, Jiashi Feng 


pdf: https://arxiv.org/abs/1707.01629 

code: official : 

https://github.com/cypw/DPNs 

code: unoffical-keras : 

https://github.com/titu1994/Keras-DualPathNetworks 

code: unofficial-pytorch : 

https://github.com/oyam/pytorch-DPNs 

code: unofficial-pytorch : 

https://github.com/rwightman/pytorch-dpn-pretrained


23. Block-QNN 


Practical Block-wise Neural Network Architecture Generation 

Zhao Zhong, Junjie Yan, Wei Wu, Jing Shao, Cheng-Lin Liu 


pdf: https://arxiv.org/abs/1708.05552


24. CRU-Net 


Sharing Residual Units Through Collective Tensor Factorization in Deep Neural Networks 

Chen Yunpeng, Jin Xiaojie, Kang Bingyi, Feng Jiashi, Yan Shuicheng 


pdf: https://arxiv.org/abs/1703.02180 

code official : 

https://github.com/cypw/CRU-Net 

code unofficial-mxnet : 

https://github.com/bruinxiong/Modified-CRUNet-and-Residual-Attention-Network.mxnet


25. ShuffleNet 


ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices 

Xiangyu Zhang, Xinyu Zhou, Mengxiao Lin, Jian Sun 


pdf: https://arxiv.org/abs/1707.01083 

code: unofficial-tensorflow : 

https://github.com/MG2033/ShuffleNet 

code: unofficial-pytorch : 

https://github.com/jaxony/ShuffleNet 

code: unofficial-caffe : 

https://github.com/farmingyard/ShuffleNet 

code: unofficial-keras : 

https://github.com/scheckmedia/keras-shufflenet


26. CondenseNet 


CondenseNet: An Efficient DenseNet using Learned Group Convolutions

Gao Huang, Shichen Liu, Laurens van der Maaten, Kilian Q. Weinberger 


pdf: https://arxiv.org/abs/1711.09224 

code: official : 

https://github.com/ShichenLiu/CondenseNet 

code: unofficial-tensorflow : 

https://github.com/markdtw/condensenet-tensorflow


27. NasNet 


Learning Transferable Architectures for Scalable Image Recognition

Barret Zoph, Vijay Vasudevan, Jonathon Shlens, Quoc V. Le 


pdf: https://arxiv.org/abs/1707.07012 

code: unofficial-keras : 

https://github.com/titu1994/Keras-NASNet 

code: keras-applications : 

https://github.com/keras-team/keras-applications/blob/master/keras_applications/nasnet.py 

code: unofficial-pytorch : 

https://github.com/wandering007/nasnet-pytorch 

code: unofficial-tensorflow : 

https://github.com/yeephycho/nasnet-tensorflow


28. MobileNetV2 


MobileNetV2: Inverted Residuals and Linear Bottlenecks 

Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen 


pdf: https://arxiv.org/abs/1801.04381 

code: unofficial-keras : 

https://github.com/xiaochus/MobileNetV2 

code: unofficial-pytorch : 

https://github.com/Randl/MobileNetV2-pytorch 

code: unofficial-tensorflow : 

https://github.com/neuleaf/MobileNetV2


29. IGCV2 


IGCV2: Interleaved Structured Sparse Convolutional Neural Networks

Guotian Xie, Jingdong Wang, Ting Zhang, Jianhuang Lai, Richang Hong, Guo-Jun Qi 


pdf: https://arxiv.org/abs/1804.06202


30. hier 


Hierarchical Representations for Efficient Architecture Search 

Hanxiao Liu, Karen Simonyan, Oriol Vinyals, Chrisantha Fernando, Koray Kavukcuoglu 


pdf: https://arxiv.org/abs/1711.00436


31. PNasNet 


Progressive Neural Architecture Search 

Chenxi Liu, Barret Zoph, Maxim Neumann, Jonathon Shlens, Wei Hua, Li-Jia Li, Li Fei-Fei, Alan Yuille, Jonathan Huang, Kevin Murphy 


pdf: https://arxiv.org/abs/1712.00559 

code: tensorflow-slim : 

https://github.com/tensorflow/models/blob/master/research/slim/nets/nasnet/pnasnet.py 

code: unofficial-pytorch : 

https://github.com/chenxi116/PNASNet.pytorch 

code: unofficial-tensorflow : 

https://github.com/chenxi116/PNASNet.TF


32. AmoebaNet 


Regularized Evolution for Image Classifier Architecture Search 

Esteban Real, Alok Aggarwal, Yanping Huang, Quoc V Le 


pdf: https://arxiv.org/abs/1802.01548 

code: tensorflow-tpu : 

https://github.com/tensorflow/tpu/tree/master/models/official/amoeba_net


33. SENet 


Squeeze-and-Excitation Networks 

Jie Hu, Li Shen, Samuel Albanie, Gang Sun, Enhua Wu 


pdf: https://arxiv.org/abs/1709.01507 

code: official : 

https://github.com/hujie-frank/SENet 

code: unofficial-pytorch : 

https://github.com/moskomule/senet.pytorch 

code: unofficial-tensorflow : 

https://github.com/taki0112/SENet-Tensorflow 

code: unofficial-caffe : 

https://github.com/shicai/SENet-Caffe 

code: unofficial-mxnet : 

https://github.com/bruinxiong/SENet.mxnet


34. ShuffleNetV2 


ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design

Ningning Ma, Xiangyu Zhang, Hai-Tao Zheng, Jian Sun 


pdf: https://arxiv.org/abs/1807.11164 

code: unofficial-pytorch : 

https://github.com/Randl/ShuffleNetV2-pytorch 

code: unofficial-keras : 

https://github.com/opconty/keras-shufflenetV2 

code: unofficial-pytorch : 

https://github.com/Bugdragon/ShuffleNet_v2_PyTorch 

code: unofficial-caff2: 

https://github.com/wolegechu/ShuffleNetV2.Caffe2


35. IGCV3 


IGCV3: Interleaved Low-Rank Group Convolutions for Efficient Deep Neural Networks 

Ke Sun, Mingjie Li, Dong Liu, Jingdong Wang 


pdf: https://arxiv.org/abs/1806.00178 

code: official : 

https://github.com/homles11/IGCV3 

code: unofficial-pytorch : 

https://github.com/xxradon/IGCV3-pytorch 

code: unofficial-tensorflow : 

https://github.com/ZHANG-SHI-CHANG/IGCV3


36. MNasNet 


MnasNet: Platform-Aware Neural Architecture Search for Mobile

Mingxing Tan, Bo Chen, Ruoming Pang, Vijay Vasudevan, Quoc V. Le 


pdf: https://arxiv.org/abs/1807.11626 

code: unofficial-pytorch : 

https://github.com/AnjieZheng/MnasNet-PyTorch 

code: unofficial-caffe : 

https://github.com/LiJianfei06/MnasNet-caffe 

code: unofficial-MxNet : 

https://github.com/chinakook/Mnasnet.MXNet 

code: unofficial-keras : 

https://github.com/Shathe/MNasNet-Keras-Tensorflow


 End 


声明：部分内容来源于网络，仅供读者学术交流之目的。文章版权归原作者所有。如有不妥，请联系删除。


浏览 57
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报
评论
图片
表情
推荐
一个完整的pytorch预训练实现图像分类，模型融合
机器学习AI算法工程
0
（附论文&代码）​ICCV2021 | 首个CNN+Transformer的backbone 模型!
目标检测与深度学习
0
200+自动驾驶最新论文代码，超全分类汇总
2024年开始，将是高阶自动驾驶关键节点。大模型、端到端、OCC这些技术正逐步走向量产。今天就为大家盘点200+篇自动驾驶论文与代码，来自ICCV、NeurIPS等顶会，涉及大模型自动驾驶、端到端、BEV感知、检测、OCC等10余...
时序资料汇总：模型和常见库对比
Datawhale
0
CV领域Paper论文常见单词
程序员大白
0
图像分类：一个更鲁棒的场景分类模型
机器学习AI算法工程
0
深度：震惊世间的惊人代码（附完整代码）
C语言与CPP编程
0
（附论文&代码）大道至简！CVPR2021论文RepVGG！
目标检测与深度学习
0
提升分类模型acc(二)：图像分类技巧实战
GiantPandaCV
0
pandas 分类数据处理大全（附代码）
小数志
0
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报