Pytorch学习之transforms（三）

ToTensor()

Convert a PIL Image or numpy.ndarray to tensor. This transform does not support torchscript.
目前学到的一般的图片转换类型有如下三种

1. PIL ImageImage.open('路径')
2. numpy.ndarraycv.imread('路径')或np.array(PIL Image)
3. torch.TensorToTensor(PIL Image or numpy.ndarray)

下面代码主要实现功能为：
1、将PIL Image转化为torch.Tensor
2、将tensor类型图片传入writer
3、使用tensorboard绘制图片tensorboard --logdir=设置目录 --port=设置端口号

from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms
from PIL import Image
'''
transforms就是一个文件
类似一个工具箱，有很多工具
如ToTensor，Compose，resize等
输入是一个指定格式的图片
经过处理
输出是一个想要的结果
'''
# tensor数据类型
# 通过transforms.ToTensor去解决两个问题
# 1）transforms该如何使用
# 2）为什么需要ToTensor
def PIL_to_Tensor():
    # 1）获取PIL.image类型图片
    img_path = '../datasets/data/train/ants_image/196757565_326437f5fe.jpg'
    img_PIL = Image.open(img_path)

    # 2）实例化一个Tensor转换器
    ToTensor = transforms.ToTensor()

    # 3）转化为tensor类型
    img_Tensor = ToTensor(img_PIL)

    # 4）实例化writer
    writer = SummaryWriter('../logs')

    # 5）传入tensor类型图片进行绘制
    writer.add_image('ants', img_Tensor)
    writer.close()

if __name__ == '__main__':
    PIL_to_Tensor()

Normalize()

Normalize a tensor image with mean and standard deviation.
用均值和标准差对张量图像进行归一化。
Given mean: (mean[1],...,mean[n]) and std: (std[1],..,std[n]) for n
channels, this transform will normalize each channel of the input
torch.*Tensor i.e.,
output[channel] = (input[channel] - mean[channel]) / std[channel]

from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms
from PIL import Image
'''
transforms就是一个文件
类似一个工具箱，有很多工具
如ToTensor，Compose，resize等
输入是一个指定格式的图片
经过处理
输出是一个想要的结果
'''
# tensor数据类型
# 通过transforms.ToTensor去解决两个问题
# 1）transforms该如何使用
# 2）为什么需要ToTensor


def normalize():
    '''
    图片归一化
    :return:
    '''
    # 1）获取PIL.image类型图片
    img_path = '../datasets/data/train/ants_image/7759525_1363d24e88.jpg'
    img_PIL = Image.open(img_path)

    # 2）实例化一个Tensor转换器
    ToTensor = transforms.ToTensor()

    # 3）转化为tensor类型
    img_Tensor = ToTensor(img_PIL)

    # # 4）实例化Normalize实例
    # Normalize = transforms.Normalize([1, 4, 9], [2, 0.4, 8])
    #
    # # 5）归一化图片
    # img_Tensor_normalized = Normalize(img_Tensor)

    # 6）实例化SummaryWriter
    writer = SummaryWriter('../logs')

    # 5）传入归一化图片进行绘制
    for i in range(10):
        Normalizer = transforms.Normalize([i+1, i+2, i+3], [i-1.5, i+2.5, i-3.5])
        img_Tensor_normalized = Normalizer(img_Tensor)
        writer.add_image('ants_normalized', img_Tensor_normalized, i)
    writer.close()


if __name__ == '__main__':
    PIL_to_Tensor()
    normalize()

Resize()

Resize the input image to the given size.
If the image is torch Tensor, it is expected
to have […, H, W] shape, where … means an arbitrary number of leading dimensions

def resize():
    """
    调整图片大小
    :return:
    """
    img_path = '../datasets/data/train/bees_image/132511197_0b86ad0fff.jpg'
    img_PIL = Image.open(img_path)
    Totensor = transforms.ToTensor()
    img_tensor = Totensor(img_PIL)
    resizer = transforms.Resize((512, 512))
    img_resized = resizer(img_tensor)
    writer = SummaryWriter('../logs')
    writer.add_image('bees', img_tensor)
    writer.add_image('bees_resized', img_resized)
    writer.close()