外部数据集的接入
相关模块：torchvision具体操作自定义数据集的基础方法：使用 Torchvision Transforms 结合 Pandas 使用 __getitem__()使用 Dataloader 读取自定义数据集Stanford Dogs 数据集自定义实例FaceLandmarks实例相关模块：torchvision
torchvision 是独立于pytorch 之外的图像操作库
具体介绍详见:DrHW的文章
torchvision主要包括一下几个包：1
torchvision.datasets:几个常用视觉数据集，可以下载和加载，这里主要的高级用法就是可以看源码如何自己写自己的Dataset的子类
这部分就是本文要介绍的重点torchvision.models: 流行的模型，例如 AlexNet, VGG, ResNet 和 Densenet 以及 与训练好的参数。torchvision.transforms: 常用的图像操作，例如：随机切割，旋转，数据类型转换，图像到tensor ,numpy 数组到tensor , tensor 到 图像等。torchvision.utils: 用于把形似(3 x H x W)的张量保存到硬盘中，给一个mini-batch的图像可以产生一个图像格网。
shape = (channel, height, weight)
具体操作
自定义数据集的基础方法：
引文2
"""inout pipline for custom dataset"""from torch.utils.data.dataset import Datasetclass CustomDataset(Dataset):def __init__(self):"""一些初始化过程写在这里"""# TODO# 1. Initialize file paths or a list of file names. passdef __getitem__(self, index):"""返回数据和标签，可以这样显示调用：img, label = MyCustomDataset.__getitem__(99)"""# TODO# 1. Read one data from file (e.g. using numpy.fromfile, PIL.Image.open).# 2. Preprocess the data (e.g. torchvision.Transform).# 3. Return a data pair (e.g. image and label).passdef __len__(self):"""返回所有数据的数量"""# You should change 9 to the total size of your dataset.return 9 # e.g. 9 is size of dataset
使用 Torchvision Transforms
方法一：from torch.utils.data.dataset import Datasetfrom torchvision import transformsclass MyCustomDataset(Dataset):def __init__(self, ..., transforms=None):# stuff...self.transforms = transformsdef __getitem__(self, index):# stuff...data = # 一些读取的数据if self.transforms is not None:data = self.transforms(data)# 如果 transform 不为 None，则进行 transform 操作return (img, label)def __len__(self):return count if __name__ == \'__main__\':# 定义我们的 transforms (1)transformations = pose([transforms.CenterCrop(100), transforms.ToTensor()])# 创建 datasetcustom_dataset = MyCustomDataset(..., transformations)
方法二：
有些人不喜欢将transform写在Dataset外， 即在Dataset内定义transform
from torch.utils.data.dataset import Datasetfrom torchvision import transformsclass MyCustomDataset(Dataset):def __init__(self, ...):# stuff...# (2) 一种方法是单独定义 transformself.center_crop = transforms.CenterCrop(100)self.to_tensor = transforms.ToTensor()# (3) 或者写成下面这样 self.transformations = \pose([transforms.CenterCrop(100),transforms.ToTensor()])def __getitem__(self, index):# stuff...data = #一些读取的数据# 当第二次调用 transform 时，调用的是 __call__()data = self.center_crop(data) # (2)data = self.to_tensor(data) # (2)# 或者写成下面这样data = self.trasnformations(data) # (3)# 注意 (2) 和 (3) 中只需要实现一种return (img, label)def __len__(self):return countif __name__ == \'__main__\':custom_dataset = MyCustomDataset(...)
结合 Pandas 使用getitem()
另一种情况是 csv 文件中保存了我们需要的图像文件的像素值（比如有些 MNIST 教程就是这样的）。我们需要改动一下getitem() 函数。
class CustomDatasetFromCSV(Dataset):def __init__(self, csv_path, height, width, transforms=None):"""Args:csv_path (string): csv 文件路径height (int): 图像高度width (int): 图像宽度transform: transform 操作"""self.data = pd.read_csv(csv_path)self.labels = np.asarray(self.data.iloc[:, 0])self.height = heightself.width = widthself.transforms = transformdef __getitem__(self, index):single_image_label = self.labels[index]# 读取所有像素值，并将 1D array ([784]) reshape 成为 2D array ([28,28]) img_as_np = np.asarray(self.data.iloc[index][1:]).reshape(28,28).astype(\'uint8\')# 把 numpy array 格式的图像转换成灰度 PIL imageimg_as_img = Image.fromarray(img_as_np)img_as_img = img_as_img.convert(\'L\')# 将图像转换成 tensorif self.transforms is not None:img_as_tensor = self.transforms(img_as_img)# 返回图像及其 labelreturn (img_as_tensor, single_image_label)def __len__(self):return len(self.data.index)if __name__ == "__main__":transformations = pose([transforms.ToTensor()])custom_mnist_from_csv = \CustomDatasetFromCSV(\'../data/mnist_in_csv.csv\', 28, 28, transformations)
使用 Dataloader 读取自定义数据集
PyTorch 中的 Dataloader 只是调用getitem() 方法并组合成 batch，我们可以这样调用：
...if __name__ == "__main__":# 定义 transformstransformations = pose([transforms.ToTensor()])# 自定义数据集custom_mnist_from_csv = \CustomDatasetFromCSV(\'../data/mnist_in_csv.csv\',28, 28,transformations)# 定义 data loadermn_dataset_loader = torch.utils.data.DataLoader(dataset=custom_mnist_from_csv,batch_size=10,shuffle=False)for images, labels in mn_dataset_loader:# 将数据传给网络模型
需要注意的是使用多卡训练时，PyTorch dataloader 会将每个 batch 平均分配到各个 GPU。所以如果 batch size 过小，可能发挥不了多卡的效果。
Stanford Dogs 数据集自定义实例
from torch.utils.data.dataset import Datasetfrom torchvision import transformsclass MyDateset(Dataset):def __init__(self, file_folder, is_test=False, transform=None):self.img_folder_path = '../input/images/Images/'self.annotation_folder_path = '../input/annotations/Annotation/'self.file_folder = file_folderself.transform = transform#self.transform = poseself.is_test = is_testdef __getitem__(self, idx):file = self.file_folder[idx]img_path = self.img_folder_path + fileimg = Image.open(img_path).convert('RGB')if not self.is_test:annotation_path = self.annotation_folder_path + file.split('.')[0]with open(annotation_path) as f:annotation = f.read()xy = self.get_xy(annotation)box = torch.FloatTensor(list(xy))new_box = self.box_resize(box, img)if self.transform is not None:img = self.transform(img)return img, new_boxelse:if self.transform is not None:img = self.transform(img)return imgdef __len__(self):return len(self.file_folder)def get_xy(self, annotation):xmin = int(re.findall('(?<=<xmin>)[0-9]+?(?=</xmin>)', annotation)[0])xmax = int(re.findall('(?<=<xmax>)[0-9]+?(?=</xmax>)', annotation)[0])ymin = int(re.findall('(?<=<ymin>)[0-9]+?(?=</ymin>)', annotation)[0])ymax = int(re.findall('(?<=<ymax>)[0-9]+?(?=</ymax>)', annotation)[0])return xmin, ymin, xmax, ymaxdef show_box(self):file = random.choice(self.file_folder)annotation_path = self.annotation_folder_path + file.split('.')[0]img_box = Image.open(self.img_folder_path + file)with open(annotation_path) as f:annotation = f.read()draw = ImageDraw.Draw(img_box)xy = self.get_xy(annotation)print('bbox:', xy)draw.rectangle(xy=[xy[:2], xy[2:]])return img_boxdef box_resize(self, box, img, dims=(332, 332)):old_dims = torch.FloatTensor([img.width, img.height, img.width, img.height]).unsqueeze(0)new_box = box / old_dimsnew_dims = torch.FloatTensor([dims[1], dims[0], dims[1], dims[0]]).unsqueeze(0)new_box = new_box * new_dimsreturn new_box
FaceLandmarks实例
class FaceLandmarksDataset(Dataset):"""Face Landmarks dataset."""def __init__(self, csv_file, root_dir, transform=None):"""Args:csv_file (string): Path to the csv file with annotations.root_dir (string): Directory with all the images.transform (callable, optional): Optional transform to be appliedon a sample."""self.landmarks_frame = pd.read_csv(csv_file)self.root_dir = root_dirself.transform = transformdef __len__(self):return len(self.landmarks_frame)def __getitem__(self, idx):img_name = os.path.join(self.root_dir,self.landmarks_frame.iloc[idx, 0])image = io.imread(img_name)landmarks = self.landmarks_frame.iloc[idx, 1:].as_matrix()landmarks = landmarks.astype('float').reshape(-1, 2)sample = {'image': image, 'landmarks': landmarks}if self.transform:sample = self.transform(sample)return sample
参考文献：
yunjey的github代码pytorch官方教程数据集：Stanford Dogs Datasetpytorch中文网:PyTorch 中自定义数据集的读取方法小结
/yjphhw/p/9773333.html ↩︎
/yunjey/pytorch-tutorial/ ↩︎
