原文：https://www.cnblogs.com/denny402/p/7520063.html
原文：https://www.jianshu.com/p/84f72791806f
原文：https://blog.csdn.net/lee813/article/details/89609691
1、下载fashion-mnist数据集
地址：https://github.com/zalandoresearch/fashion-mnist
下面这四个都要下载，下载完成后，解压到同一个目录，我是解压到“E:/fashion_mnist/”这个目录里面，好和下面的代码目录一致
2、在Geany中执行下面这段代码。
这段代码里面，需要先用pip安装skimage、torch、torchvision，前两篇文章有安装步骤。
这段代码的作用：将下载下来的 二进制文件 转换为 图片，会在目录中生成两个文件夹和两个文本。
文件夹里面全是图片，图片的内容是数字，N多数字。
文本的内容主要是图片和真实数字的一个关联。
importosfrom skimage importioimporttorchvision.datasets.mnist as mnist
root="E:/fashion_mnist/"train_set=(
mnist.read_image_file(os.path.join(root,'train-images-idx3-ubyte')),
mnist.read_label_file(os.path.join(root,'train-labels-idx1-ubyte'))
)
test_set=(
mnist.read_image_file(os.path.join(root,'t10k-images-idx3-ubyte')),
mnist.read_label_file(os.path.join(root,'t10k-labels-idx1-ubyte'))
)print("training set :",train_set[0].size())print("test set :",test_set[0].size())def convert_to_img(train=True):if(train):
f=open(root+'train.txt','w')
data_path=root+'/train/'
if(notos.path.exists(data_path)):
os.makedirs(data_path)for i, (img,label) in enumerate(zip(train_set[0],train_set[1])):
img_path=data_path+str(i)+'.jpg'io.imsave(img_path,img.numpy())
f.write(img_path+' '+str(label)+'\n')
f.close()else:
f= open(root + 'test.txt', 'w')
data_path= root + '/test/'
if (notos.path.exists(data_path)):
os.makedirs(data_path)for i, (img,label) in enumerate(zip(test_set[0],test_set[1])):
img_path= data_path+ str(i) + '.jpg'io.imsave(img_path, img.numpy())
f.write(img_path+ ' ' + str(label) + '\n')
f.close()
convert_to_img(True)
convert_to_img(False)
View Code
3、原文的这段代码编译会出错，主要是跟下载的数据有关，数据格式不一样，这里还在处理，原因是找到了的，就一个int的转换，下面贴出改过后的代码
出错的地方：
importtorchimportreimportnumpyfrom torch.autograd importVariablefrom torchvision importtransformsfrom torch.utils.data importDataset, DataLoaderfrom PIL importImage
root="E:/fashion_mnist/"
defdefault_loader(path):return Image.open(path).convert('RGB')classMyDataset(Dataset):def __init__(self, txt, transform=None, target_transform=None, loader=default_loader):
fh= open(txt, 'r')
imgs=[]for line infh:
line= line.strip('\n')
line=line.rstrip()
words=line.split()
p1= re.compile(r'[(](.*?)[)]', re.S)
arr= re.findall(p1, words[1])
word=arr[0]
imgs.append((words[0],int(word)))
self.imgs=imgs
self.transform=transform
self.target_transform=target_transform
self.loader=loaderdef __getitem__(self, index):
fn, label=self.imgs[index]
img=self.loader(fn)if self.transform is notNone:
img=self.transform(img)returnimg,labeldef __len__(self):returnlen(self.imgs)
train_data=MyDataset(txt=root+'train.txt', transform=transforms.ToTensor())
test_data=MyDataset(txt=root+'test.txt', transform=transforms.ToTensor())
train_loader= DataLoader(dataset=train_data, batch_size=64, shuffle=True)
test_loader= DataLoader(dataset=test_data, batch_size=64)
View Code
3、原文的代码，还有一部分也会报错，ERROR如下。
唉，感叹一下，下次还是看一下语法那些，能读懂了代码再改吧，本想怎个拿来主义的，结果拿来了还是不能运行
解决-原文地址：https://blog.csdn.net/weixin_43848267/article/details/88874584
解决：将 loss_return.data[0] 改为 loss_return.data
还有几个地方 也要将 .data[0] 改为 .data
4、可完整运行的代码
代码1：
importosfrom skimage importioimporttorchvision.datasets.mnist as mnist
root="E:/fashion_mnist/"train_set=(
mnist.read_image_file(os.path.join(root,'train-images-idx3-ubyte')),
mnist.read_label_file(os.path.join(root,'train-labels-idx1-ubyte'))
)
test_set=(
mnist.read_image_file(os.path.join(root,'t10k-images-idx3-ubyte')),
mnist.read_label_file(os.path.join(root,'t10k-labels-idx1-ubyte'))
)print("training set :",train_set[0].size())print("test set :",test_set[0].size())def convert_to_img(train=True):if(train):
f=open(root+'train.txt','w')
data_path=root+'/train/'
if(notos.path.exists(data_path)):
os.makedirs(data_path)for i, (img,label) in enumerate(zip(train_set[0],train_set[1])):
img_path=data_path+str(i)+'.jpg'io.imsave(img_path,img.numpy())
f.write(img_path+' '+str(label)+'\n')
f.close()else:
f= open(root + 'test.txt', 'w')
data_path= root + '/test/'
if (notos.path.exists(data_path)):
os.makedirs(data_path)for i, (img,label) in enumerate(zip(test_set[0],test_set[1])):
img_path= data_path+ str(i) + '.jpg'io.imsave(img_path, img.numpy())
f.write(img_path+ ' ' + str(label) + '\n')
f.close()
convert_to_img(True)
convert_to_img(False)
View Code
代码2：
importreimportnumpyimporttorchfrom torch.autograd importVariablefrom torchvision importtransformsfrom torch.utils.data importDataset, DataLoaderfrom PIL importImage
root="E:/fashion_mnist/"
#-----------------ready the dataset--------------------------
defdefault_loader(path):return Image.open(path).convert('RGB')classMyDataset(Dataset):def __init__(self, txt, transform=None, target_transform=None, loader=default_loader):
fh= open(txt, 'r')
imgs=[]for line infh:
line= line.strip('\n')
line=line.rstrip()
words=line.split()
p1= re.compile(r'[(](.*?)[)]', re.S)
arr= re.findall(p1, words[1])
word=arr[0]
imgs.append((words[0],int(word)))
self.imgs=imgs
self.transform=transform
self.target_transform=target_transform
self.loader=loaderdef __getitem__(self, index):
fn, label=self.imgs[index]
img=self.loader(fn)if self.transform is notNone:
img=self.transform(img)returnimg,labeldef __len__(self):returnlen(self.imgs)
train_data=MyDataset(txt=root+'train.txt', transform=transforms.ToTensor())
test_data=MyDataset(txt=root+'test.txt', transform=transforms.ToTensor())
train_loader= DataLoader(dataset=train_data, batch_size=64, shuffle=True)
test_loader= DataLoader(dataset=test_data, batch_size=64)#-----------------create the Net and training------------------------
classNet(torch.nn.Module):def __init__(self):
super(Net, self).__init__()
self.conv1=torch.nn.Sequential(
torch.nn.Conv2d(3, 32, 3, 1, 1),
torch.nn.ReLU(),
torch.nn.MaxPool2d(2))
self.conv2=torch.nn.Sequential(
torch.nn.Conv2d(32, 64, 3, 1, 1),
torch.nn.ReLU(),
torch.nn.MaxPool2d(2)
)
self.conv3=torch.nn.Sequential(
torch.nn.Conv2d(64, 64, 3, 1, 1),
torch.nn.ReLU(),
torch.nn.MaxPool2d(2)
)
self.dense=torch.nn.Sequential(
torch.nn.Linear(64 * 3 * 3, 128),
torch.nn.ReLU(),
torch.nn.Linear(128, 10)
)defforward(self, x):
conv1_out=self.conv1(x)
conv2_out=self.conv2(conv1_out)
conv3_out=self.conv3(conv2_out)
res= conv3_out.view(conv3_out.size(0), -1)
out=self.dense(res)returnout
model=Net()print(model)
optimizer=torch.optim.Adam(model.parameters())
loss_func=torch.nn.CrossEntropyLoss()for epoch in range(10):print('epoch {}'.format(epoch + 1))#training-----------------------------
train_loss =0.
train_acc=0.for batch_x, batch_y intrain_loader:
batch_x, batch_y=Variable(batch_x), Variable(batch_y)
out=model(batch_x)
loss=loss_func(out, batch_y)
train_loss+=loss.item()
pred= torch.max(out, 1)[1]
train_correct= (pred ==batch_y).sum()
train_acc+=train_correct.item()
optimizer.zero_grad()
loss.backward()
optimizer.step()print('Train Loss: {:.6f}, Acc: {:.6f}'.format(train_loss /(len(
train_data)), train_acc/(len(train_data))))#evaluation--------------------------------
model.eval()
eval_loss=0.
eval_acc=0.for batch_x, batch_y intest_loader:
batch_x, batch_y=Variable(batch_x), Variable(batch_y)
out=model(batch_x)
loss=loss_func(out, batch_y)
eval_loss+=loss.item()
pred= torch.max(out, 1)[1]
num_correct= (pred ==batch_y).sum()
eval_acc+=num_correct.item()print('Test Loss: {:.6f}, Acc: {:.6f}'.format(eval_loss /(len(
test_data)), eval_acc/ (len(test_data))))
View Code
5、总结
提示：训练模型有点耗时，这里注意一下
图片如果过小，标签页里面单独打开图片会大些，排版搞得屁理解一下，一来没时间写文章，二来排版还没学，以后空了就会学。还是先把文章的质量提高了来
出现的问题主要是因为 torch的版本不同造成的，所以一会我把 我这里的环境贴出来，避免发生同样的错误。
6、环境
系统：win7 64位
Python 3.7.3
各个包的版本号，其它的好像就没啥了
可测试代码-版本2
代码1：
#coding=utf-8
importosfrom skimage importioimporttorchvision.datasets.mnist as mnist
root="E:/fashion_mnist/"train_set=(
mnist.read_image_file(os.path.join(root,'train-images-idx3-ubyte')),
mnist.read_label_file(os.path.join(root,'train-labels-idx1-ubyte'))
)
test_set=(
mnist.read_image_file(os.path.join(root,'t10k-images-idx3-ubyte')),
mnist.read_label_file(os.path.join(root,'t10k-labels-idx1-ubyte'))
)print("training set :",train_set[0].size())print("test set :",test_set[0].size())def convert_to_img(train=True):if(train):
f=open(root+'train.txt','w')
data_path=root+'/train/'
if(notos.path.exists(data_path)):
os.makedirs(data_path)for i, (img,label) in enumerate(zip(train_set[0],train_set[1])):
img_path=data_path+str(i)+'.jpg'io.imsave(img_path,img.numpy())
f.write(img_path+' '+str(label.numpy())+'\n') #label改为label.numpy()
f.close()else:
f= open(root + 'test.txt', 'w')
data_path= root + '/test/'
if (notos.path.exists(data_path)):
os.makedirs(data_path)for i, (img,label) in enumerate(zip(test_set[0],test_set[1])):
img_path= data_path+ str(i) + '.jpg'io.imsave(img_path, img.numpy())
f.write(img_path+ ' ' + str(label.numpy()) + '\n')
f.close()
convert_to_img(True)
convert_to_img(False)
View Code
代码2：
importtorchfrom torch.autograd importVariablefrom torchvision importtransformsfrom torch.utils.data importDataset, DataLoaderfrom PIL importImage
root="E:/fashion_mnist/"
defdefault_loader(path):return Image.open(path).convert('RGB')classMyDataset(Dataset):def __init__(self, txt, transform=None, target_transform=None, loader=default_loader):
fh= open(txt, 'r')
imgs=[]for line infh:
line= line.strip('\n')
line=line.rstrip()
words=line.split()
imgs.append((words[0],int(words[1])))
self.imgs=imgs
self.transform=transform
self.target_transform=target_transform
self.loader=loaderdef __getitem__(self, index):
fn, label=self.imgs[index]
img=self.loader(fn)if self.transform is notNone:
img=self.transform(img)returnimg,labeldef __len__(self):returnlen(self.imgs)
train_data=MyDataset(txt=root+'train.txt', transform=transforms.ToTensor())
test_data=MyDataset(txt=root+'test.txt', transform=transforms.ToTensor())
train_loader= DataLoader(dataset=train_data, batch_size=64, shuffle=True)
test_loader= DataLoader(dataset=test_data, batch_size=64)#-----------------create the Net and training------------------------
classNet(torch.nn.Module):def __init__(self):
super(Net, self).__init__()
self.conv1=torch.nn.Sequential(
torch.nn.Conv2d(3, 32, 3, 1, 1),
torch.nn.ReLU(),
torch.nn.MaxPool2d(2))
self.conv2=torch.nn.Sequential(
torch.nn.Conv2d(32, 64, 3, 1, 1),
torch.nn.ReLU(),
torch.nn.MaxPool2d(2)
)
self.conv3=torch.nn.Sequential(
torch.nn.Conv2d(64, 64, 3, 1, 1),
torch.nn.ReLU(),
torch.nn.MaxPool2d(2)
)
self.dense=torch.nn.Sequential(
torch.nn.Linear(64 * 3 * 3, 128),
torch.nn.ReLU(),
torch.nn.Linear(128, 10)
)defforward(self, x):
conv1_out=self.conv1(x)
conv2_out=self.conv2(conv1_out)
conv3_out=self.conv3(conv2_out)
res= conv3_out.view(conv3_out.size(0), -1)
out=self.dense(res)returnout
model=Net()print(model)
optimizer=torch.optim.Adam(model.parameters())
loss_func=torch.nn.CrossEntropyLoss()for epoch in range(10):print('epoch {}'.format(epoch + 1))#training-----------------------------
train_loss =0.
train_acc=0.for batch_x, batch_y intrain_loader:
batch_x, batch_y=Variable(batch_x), Variable(batch_y)
out=model(batch_x)
loss=loss_func(out, batch_y)
train_loss+=loss.data
pred= torch.max(out, 1)[1]
train_correct= (pred ==batch_y).sum()
train_acc+=train_correct.data
optimizer.zero_grad()
loss.backward()
optimizer.step()print('Train Loss: {:.6f}, Acc: {:.6f}'.format(train_loss /(len(
train_data)), train_acc/(len(train_data))))#evaluation--------------------------------
model.eval()
eval_loss=0.
eval_acc=0.for batch_x, batch_y intest_loader:
batch_x, batch_y= Variable(batch_x, volatile=True), Variable(batch_y, volatile=True)
out=model(batch_x)
loss=loss_func(out, batch_y)
eval_loss+=loss.data
pred= torch.max(out, 1)[1]
num_correct= (pred ==batch_y).sum()
eval_acc+=num_correct.dataprint('Test Loss: {:.6f}, Acc: {:.6f}'.format(eval_loss /(len(
test_data)), eval_acc/ (len(test_data))))
View Code
版本2修改的地方
原文：https://blog.csdn.net/shang_jia/article/details/82936074
原文：https://www.liangzl.com/get-article-detail-8524.html
注意：下面的代码不管，下面是第一次测试的时候，下载错了数据集
问题：这里的数据集是数字，不是这个数据集，代码里面是用的fashion-mnist这个数据集
1、下载mnist数据集
地址：http://yann.lecun.com/exdb/mnist/
下面这四个都要下载，下载完成后，解压到同一个目录，我是解压到“E:/fashion_mnist/”这个目录里面，好和下面的代码目录一致
解压完成后，需要修改一下文件名，如（修改原因：保持和下面代码一样，避免出现其它问题）：
修改前：t10k-images.idx3-ubyte
修改后：t10k-images-idx3-ubyte
我是第一次弄这玩意，所以尽量弄得白痴些，走弯路很烦，有时候一点点小问题就弄半天，其实就是别人有那么一点没讲清楚，然后就会搞很久
1118
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
