README.md

### Exp1-关键代码
> 主要学习如何导入数据
```Python
class HandWritingNumberRecognize_Dataset(Dataset):
def __init__(self,opt,dataset,transform=None):
self.path = opt.file_path #文件路径
self.dataset = dataset #train or val or test
self.transform = transform # 变换（这里是totensor）
self.label = [] # 标签定义
self.fname = [] # 图片的文件路径
folder_path = os.path.join(self.path,self.dataset,"images") # 获取图片文件夹
path_list = os.listdir(folder_path) # 得到文件夹下所有问价
path_list.sort(key=lambda x:int((x.split('.')[0]).split("_")[1])) # 排序

for fname in path_list:
self.fname.append(os.path.join(folder_path,fname)) # 导入所有文件路径
if self.dataset != "test":
label_file = os.path.join(self.path,self.dataset,"labels_"+self.dataset+".txt")
with open(label_file) as f:
self.label = f.readlines() # 读入每行
for i in range(0, len(self.label)):
self.label[i] = eval(self.label[i].rstrip('\n')) #去掉换行符，转成数字
f.close() #关闭文件
def __getitem__(self, index):
if self.dataset != "test": #如果是测试集，它没有label所以需要特别处理
image_name,image_label = self.fname[index],self.label[index]
image = Image.open(image_name)
image = self.transform(image) #转成tensor
return image,image_label
else:
image_name = self.fname[index]
image = Image.open(image_name)
image = self.transform(image)
return image
def __len__(self):
return len(self.fname) #返回数据长度
```