Aggregated Residual Transformations for Deep Neural Networks

最新推荐文章于 2025-02-05 13:05:42 发布
最新推荐文章于 2025-02-05 13:05:42 发布
阅读量287 收藏
点赞数
分类专栏： deep learning pytorch paper
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.youkuaiyun.com/seamanj/article/details/90677427
版权

在这里插入图片描述

/home/seamanj/Software/anaconda3/bin/python3.7 /home/seamanj/Workspace/AlexNet/main.py
Downloading: "https://download.pytorch.org/models/resnext50_32x4d-7cdf4587.pth" to /home/seamanj/.cache/torch/checkpoints/resnext50_32x4d-7cdf4587.pth
100%|██████████| 100441675/100441675 [00:03<00:00, 30273827.14it/s]
ResNet(
  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (relu): ReLU(inplace)
  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
  (layer1): Sequential(
    (0): Bottleneck(
      (conv1): Conv2d(64, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
      (downsample): Sequential(
        (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      )
    )
    (1): Bottleneck(
      (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
    (2): Bottleneck(
      (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
  )
  (layer2): Sequential(
    (0): Bottleneck(
      (conv1): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
      (downsample): Sequential(
        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      )
    )
    (1): Bottleneck(
      (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
    (2): Bottleneck(
      (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
    (3): Bottleneck(
      (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
  )
  (layer3): Sequential(
    (0): Bottleneck(
      (conv1): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(512, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
      (downsample): Sequential(
        (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
        (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      )
    )
    (1): Bottleneck(
      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(512, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
    (2): Bottleneck(
      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(512, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
    (3): Bottleneck(
      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(512, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
    (4): Bottleneck(
      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(512, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
    (5): Bottleneck(
      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(512, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
  )
  (layer4): Sequential(
    (0): Bottleneck(
      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
      (downsample): Sequential(
        (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
        (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      )
    )
    (1): Bottleneck(
      (conv1): Conv2d(2048, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
    (2): Bottleneck(
      (conv1): Conv2d(2048, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
    )
  )
  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))
  (fc): Linear(in_features=2048, out_features=1000, bias=True)
)

Process finished with exit code 0
炼丹-深度学习-《Aggregated Residual Transformations for Deep Neural Networks》
qq_24375467的博客
12-18 1021
Aggregated Residual Transformations for Deep Neural Networks 原文地址：Aggregated Residual Transformations for Deep Neural Networks 文章目录Aggregated Residual Transformations for Deep Neural Networks生词：Abstra...
Aggregated Residual Transformations for Deep Neural Networks【ResNeXt翻译】
qq_32244493的博客
11-26 418
论文链接 代码链接（Pytorch） 介绍 图1图1图1 本文提出一种简单的体系结构，这个结构采用VGG和ResNet使用的层堆叠策略，同时以一种简单、可扩展的方式利用拆分-转换-合并的策略。我们网络中的一个模块执行一组转换，每个转换都在低维嵌入中进行，其输出通过求和进行聚合。我们追求该想法的简单实现——要聚合的转换都是相同的拓扑（如图1右边所示）。这种设计允许我们扩展到任何大量的转换，而不需要专门去设计。 在这种简化的情况下，我们的模型还有另外两种等效形式(图3)。图3(b)中的重构看起来类似于In.
ICLR 2025 高分论文！谢赛宁团队提出3DGS分布式训练新方案：Grendel
最新发布
阿木寺的博客
02-05 287
点击下方卡片，关注“CVer”公众号AI/CV重磅干货，第一时间送达点击进入—>【3DGS和NeRF】交流群添加微信：CVer2233，助手会拉你进群！扫描下方二维码，加入CVer学术星球！可获得最新顶会/顶刊上的论文idea和CV从入门到精通资料，及最前沿应用！发论文/搞科研/涨薪必备！克雷西 发自 凹非寺转载自：量子位（QbitAI）高斯泼溅模型训练的内存瓶颈，终于被谢赛宁团队和NYU系...
ResNeXt：Aggregated Residual Transformations for Deep Neural Networks
女王の专属领地
03-22 631
论文链接：https://arxiv.org/abs/1611.05431 PyTorch代码：https://github.com/miraclewkf/ResNeXt-PyTorch 这是一篇发表在2017CVPR上的论文，介绍了ResNet网络的升级版：ResNeXt。 作者提出 ResNeXt 的主要原因在于： 传统的要提高模型的准确率，都是加深或加宽网络，但是随着超参数数量的增加...
【翻译】Aggregated Residual Transformations for Deep Neural Networks
jokerwu192的博客
07-04 595
Aggregated Residual Transformations for Deep Neural Networks 作者：Saining Xie1 Ross Girshick2 Piotr Dollar2 Zhuowen Tu1 Kaiming He2 1、UC San Diego 2、Facebook AI Research
Aggregated Residual Transformations for Deep Neural Networks.pdf
05-09
论文标题为"Aggregated Residual Transformations for Deep Neural Networks"，简而言之，它描述了一种新颖的网络架构设计方法，专门用于图像分类任务。这篇论文通过研究深度神经网络中的残差变换，提出了一种通过...
Aggregated Residual Transformations for Deep Neural Networks（论文翻译）
MJ17709005513的博客
03-18 2124
摘要 我们提出了一种用于图像分类的简单、高度模块化的网络架构。我们的网络是通过重复一个构建块来构建的，该构建块聚合了一组具有相同拓扑的转换。我们简单的设计产生了一个同质的多分支架构，只需设置几个超参数。这个策略暴露了一个新的维度，我们称之为“基数”（转换集的大小），作为深度和宽度维度之外的一个重要因素。在 ImageNet-1K 数据集上，我们凭经验表明，即使在保持复杂性的限制条件下，增加基数也能够提高分类精度。此外，当我们增加容量时，增加基数比更深或更宽更有效。我们的模型名为 ResNeXt，是我们进入
Aggregated Residual Transformations for Deep Neural Networks论文笔记
xuefengxiaoyang的博客
07-04 347
摘要： 这篇论文是ResNet的改进版，通常要增加网络的性能，是增加网络的参数，比如加深或加宽网络都可以，但是这么做会增加网络的复杂度。这篇论文，作者提出一个新感念‘cardinality’，指的是转换层的数量，叫做基数。并且证明了增加基数的大小也能有效提高网络的性能，并且比增加网络的宽度和深度更有效，并且并不会使网络更复杂。结构： 增加网络的深度只是增加卷积层的堆叠数量加深网络一般来说，层数...
[论文阅读]ResNetXt|Aggregated Residual Transformations for Deep Neural Networks
一碗白开水一
08-12 1045
题目： Aggregated Residual Transformations for Deep Neural Networks 作者： Saining Xie； Ross Girshick ； Piotr Doll´ar；Zhuowen Tu； Kaiming He 看到Kaiming He就不陌生了吧 Motivation: 作者还是为了找到一种更优的方法或者卷积方式来做目标分类。于是乎提出ResNetXt。 Method: 作者想出这样一种方法： a)b)只是concatenate的方法不一样，c
【ResNext】《Aggregated Residual Transformations for Deep Neural Networks》
bryant_meng
01-24 1898
CVPR-2017 Torch 版代码：https://github.com/facebookresearch/ResNeXt Caffe 版代码：https://github.com/soeaver/caffe-model/tree/master/cls/resnext Caffe 代码可视化工具：http://ethereon.github.io/netscope/#/editor 文章...
resnet101.zip
03-14
来自download.pytorch.org/models，resnet101(pretrained=True)
[ResNet系] 003 ResNeXt
weixin_33894992的博客
09-18 234
ResNeXt Aggregated Residual Transformations for Deep Neural NetworksSaining Xie, Ross Girshick, Piotr Dollár, Zhuowen Tu, Kaiming He Caffe实现：https://github.com/binLearnin...
READING NOTE: Aggregated Residual Transformations for Deep Neural Networks
joshua_1988
11-17 1353
TITLE: Aggregated Residual Transformations for Deep Neural Networks
论文阅读：Aggregated Residual Transformations for Deep Neural Networks（ResNeXt）
j879159541的博客
06-27 517
文章目录1、论文总述2、vgg/resnet 与 Inception系列网络设计机制的不同点3、ResNeXt并不是ensembling4、两个设计准则5、split-transform-merge的本质6、BN和Relu的位置 1、论文总述 这篇论文提出的网络是resnet的升级版，设计思想结合了vgg/resnet 的stacking building blocks 以及 Inception系列的 split - transform - merge，ResNeXt中的next是指作者在这篇论文中提出了另
ResNeXt : Aggregated Residual Transformations for Deep Neural Networks
小菜鸟的AI之路
03-05 1371
论文链接：https://arxiv.org/abs/1611.05431 创新点：ResNeXt在ResNet的基础上，结合ResNet的block stack策略以及Inception结构分组卷积的思想，设计aggregrated transformations策略，在不增加模型复杂度的情况下，提高了模型识别的准确率，虽然没有提出特别新奇的网络结构，但是ResNeXt利用更简单的拓扑结构在不...
《2017-Aggregated Residual Transformations for Deep Neural Networks》论文阅读
u010067397的专栏
11-07 295
如何评价谷歌的xception网络？ 动机 传统的要提高模型的准确率，都是加深或加宽网络，但是随着超参数数量的增加（比如channels数，filter size等等），网络设计的难度和计算开销也会增加。 本文提出的 ResNeXt 结构可以在不增加参数复杂度的前提下提高准确率，同时还减少了超参数的数量； 贡献 作者在这篇论文中提出网络 ResNeXt，同时采用 VGG 堆叠的思想和 I...
Resnet 官方代码 pytorch
苏凯祺的博客
10-26 6997
import torch import torch.nn as nn #from .utils import load_state_dict_from_url #其中torch.nn 为其提供基础函数，model_zoo提供权重数据的下载。 __all__ = ['ResNet', 'resnet18', 'resnet34', 'resnet50', 'resnet101', ...
基于残差网络的田区智能检测系统的参考文献
04-30
以下是一些可能有用的参考文献： 1. He, Kaiming, et al. "Deep residual learning for image recognition." Proceedings of the IEEE conference on computer vision and pattern recognition. 2016. 2. Wang, Xiaoyu, et al. "A deep learning approach for sugarcane rust detection using residual networks." Computers and Electronics in Agriculture 157 (2019): 459-466. 3. Xie, Lingxi, et al. "Aggregated residual transformations for deep neural networks." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017. 4. Zhang, Zhiyuan, et al. "Crop row detection in maize fields with morphological residual networks." Computers and Electronics in Agriculture 144 (2017): 99-107. 5. Wu, Yanan, et al. "A novel deep learning approach for rice leaf diseases recognition using residual networks." Computers and Electronics in Agriculture 160 (2019): 31-38. 6. Li, Hui, et al. "A deep residual network approach for automatic detection of maize tassels in field conditions." Computers and Electronics in Agriculture 163 (2019): 104850. 希望这些文献对你有所帮助！
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值