阶段2-练习1

介绍：本次练习为图像分类（image classification）
要求：
- 适配CIFAR100数据集，任选3种不同的网络结构（参考下方Implemented network），使用默认超参数训练、测试模型：
  - 关于CIFAR100数据集，直接使用torchvision.datasets.CIFAR100加载数据，用法与torchvision.datasets.CIFAR10一样，具体参考代码；
  - 结果1：将测试结果添加到下方的Results表格中；
  - 结果2：将你实验中效果最好的模型在CIFAR100测试集上跑一遍，将预测结果保存为一个文本文件，命名为results_xxx.txt（xxx，为网络结构名称，如vgg16），每一行保存一张图片的预测类别，即0-99的整数，样本顺序与测试集中样本的原始顺序一致；
  - 结果3：通读本项目代码，给每一行添加注释（不含models目录）。
提交结果： results_xxx.txt放在本项目根目录下，将本项目打包为zip文件提交。注意，把data目录删掉。

Pytorch-cifar10

practice on cifar10 using pytorch

Requirements

Experiment environment

python3.6
pytorch1.6.0
cuda10 (optional)

Usage (example)

1. train the model

# use gpu to train vgg16
$ python train.py -net vgg16

sometimes, you might want to use warmup training by set -warm to 1 or 2, to prevent network diverge during early training phase.

The supported net args are:

squeezenet
mobilenet
mobilenetv2
shufflenet
shufflenetv2
vgg11
vgg13
vgg16
vgg19
densenet121
densenet161
densenet201
googlenet
inceptionv3
inceptionv4
inceptionresnetv2
xception
resnet18
resnet34
resnet50
resnet101
resnet152
preactresnet18
preactresnet34
preactresnet50
preactresnet101
preactresnet152
resnext50
resnext101
resnext152
attention56
attention92
seresnet18
seresnet34
seresnet50
seresnet101
seresnet152
nasnet

Normally, the weights file with the best accuracy would be written to the disk with name suffix 'best'(default in checkpoint folder).

2. test the model

Test the model using test.py

$ python test.py -net vgg16 -weights path_to_vgg16_weights_file

Implementated NetWork

vgg Very Deep Convolutional Networks for Large-Scale Image Recognition
googlenet Going Deeper with Convolutions
inceptionv3 Rethinking the Inception Architecture for Computer Vision
inceptionv4, inception_resnet_v2 Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
xception Xception: Deep Learning with Depthwise Separable Convolutions
resnet Deep Residual Learning for Image Recognition
resnext Aggregated Residual Transformations for Deep Neural Networks
resnet in resnet Resnet in Resnet: Generalizing Residual Architectures
densenet Densely Connected Convolutional Networks
shufflenet ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices
shufflenetv2 ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design
mobilenet MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
mobilenetv2 MobileNetV2: Inverted Residuals and Linear Bottlenecks
residual attention network Residual Attention Network for Image Classification
senet Squeeze-and-Excitation Networks
squeezenet SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size
nasnet Learning Transferable Architectures for Scalable Image Recognition

Results

dataset	network	params	top1 err	top5 err	memory
cifar100	vgg16_bn	34.0M	27.07	8.84	2.03GB
cifar100	vgg13_bn	27.37M	33.12	10.94	109.49MB
cifar100	vgg11_bn	27.18M	36.14	12.92	108.79MB
cifar100	mobilenet	3.16M	36.81	11.97	12.64MB

njuhuxw/B.2-image-classification-master