pytorch之调用resnet、alexnet、vgg、squeezenet、densenet、inception预训练模型

pytorch调用resnet、alexnet、vgg、squeezenet、densenet、inception预训练模型

from __future__ import print_function 
from __future__ import division

import torch.nn as nn
from torchvision import datasets, models



# Top level data directory. Here we assume the format of the directory conforms 
#   to the ImageFolder structure




# Number of classes in the dataset
num_classes = 2

# Batch size for training (change depending on how much memory you have)
BATCH_SIZE = 128      #批处理尺寸(batch_size)


# Flag for feature extracting. When False, we finetune the whole model, 
#   when True we only update the reshaped layer params
feature_extract = True

是否改变卷及层的参数

def set_parameter_requires_grad(model, feature_extracting):
    if feature_extracting:
        for param in model.parameters():
            param.requires_grad = False

初始化模型

def initialize_model(mo