【PyTorch】关于Transforms的简单使用

前提文章目录

【PyTorch】深度学习PyTorch环境配置及安装【详细清晰】
【PyTorch】深度学习PyTorch加载数据
【PyTorch】关于Tensorboard的简单使用

---

torchvision中的transforms

transforms主要是对图片进行变换

transforms该如何使用(python)

绝对路径：D:\PyCharm\learn_pytorch\dataset\train\ants_image\0013035.jpg
相对路径：dataset/train/ants_image/0013035.jpg
一般采用相对路径 因为Windows系统\会被当作转义符

from PIL import Image
from torchvision import transforms

img_path = "dataset/train/ants_image/0013035.jpg"
img = Image.open(img_path)
# print(img)  # <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=768x512 at 0x28E7F9E23B0>
tensor_trans = transforms.ToTensor()
tensor_img = tensor_trans(img)
print(tensor_img)

使用Tensor的类型

tensorboard 的SummaryWriter需要使用Tensor的类型的图片

from PIL import Image
from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms

img_path = "dataset/train/ants_image/0013035.jpg"
img = Image.open(img_path)
write = SummaryWriter("logs")

# 1、transforms该如何使用(python)
tensor_trans = transforms.ToTensor()
tensor_img = tensor_trans(img)
print(tensor_img)

#2、使用Tensor的类型
write.add_image("Tensor_img",tensor_img)
write.close()

常见的Transforms

Transforms的部分方法的使用

Python中的__call__用法

class Person:
    # 下划线表示内置函数
    def __call__(self, name):
        print("__call__" + "Hello" + name)

    def hello(self, name):
        print("hello" + name)


person = Person()
person("张三")  # 内置函数可以不用调用.方法名(参数)
person.hello("李四")  # 不是内置函数需要调用.方法名(参数)

ToTensor使用

from PIL import Image
from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms

writer = SummaryWriter("logs")

img = Image.open("images/pytorch.png")
# print(img)  # <PIL.PngImagePlugin.PngImageFile image mode=RGB size=1314x684 at 0x1644E187520>
trans_totensor = transforms.ToTensor()
img_tensor = trans_totensor(img)
writer.add_image("ToTensor",img_tensor)
writer.close()

Normalize归一化的使用

from PIL import Image
from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms

writer = SummaryWriter("logs")

img = Image.open("images/pytorch.png")
# print(img)  # <PIL.PngImagePlugin.PngImageFile image mode=RGB size=1314x684 at 0x1644E187520>
# ToTensor使用
trans_totensor = transforms.ToTensor()
img_tensor = trans_totensor(img)
writer.add_image("ToTensor", img_tensor)

# Normalize使用
print(img_tensor[0][0][0])
trans_norm = transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])  # 3个颜色通道
img_norm = trans_norm(img_tensor)
print(img_norm[0][0][0])
writer.add_image("Normalize",img_norm)
writer.close()

替换参数再次尝试：

trans_norm = transforms.Normalize([6, 3, 2], [9, 3, 5])  # 3个颜色通道

writer.add_image("Normalize", img_norm, 2)

Resize的使用

Resize输入两个参数时，输出图片的高和宽的像素点数量会按照你设定的值进行输出。而只输入一个参数时，代表你最短的那个边输出的像素点数量

# Resize使用
print(img.size)  # (3200, 1800)
trans_resize = transforms.Resize((512, 512)) # 注意这里一定要多加一个括号
img_resize = trans_resize(img)
print(img_resize)  # <PIL.Image.Image image mode=RGB size=512x512 at 0x15D57A081C0>
# img PIL ---->  Resize ----> img_resize PIL ----> ToTensor ----> img_resize tensor
# 要用tensorboard显示要将PIL的类型转换为tensor类型
img_resize = trans_totensor(img_resize)
writer.add_image("Resize", img_resize, 0)

writer.close()

图片被拉长了

Compose()的使用

trans_resize_2 = transforms.Resize(512)
# PIL  ---->   PIL  ---->tensor
trans_compose = transforms.Compose([trans_resize_2, trans_totensor])  # 注意一定要加中括号
img_resize_2 = trans_compose(img)
writer.add_image("Resize", img_resize_2, 1)

RandomCrop随机裁剪使用

# RandomCrop随机裁剪使用
trans_random = transforms.RandomCrop(512)
trans_compose_2 = transforms.Compose([trans_random, trans_totensor])
# 裁剪10个
for i in range(10):
    img_crop = trans_compose_2(img)
    writer.add_image("RandomCrop", img_crop, i)

加序列进行裁剪

trans_random = transforms.RandomCrop((500, 1000))  # 注意2个括号，裁剪500X1000的

writer.add_image("RandomCropHW", img_crop, i)

总结Transforms方法使用

以上测试全部代码：

from PIL import Image
from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms

writer = SummaryWriter("logs")

img = Image.open("images/pytorch.png")
# print(img)  # <PIL.PngImagePlugin.PngImageFile image mode=RGB size=1314x684 at 0x1644E187520>
# ToTensor使用
trans_totensor = transforms.ToTensor()
img_tensor = trans_totensor(img)
writer.add_image("ToTensor", img_tensor)

# Normalize使用
print(img_tensor[0][0][0])
# trans_norm = transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])  # 3个颜色通道
trans_norm = transforms.Normalize([6, 3, 2], [9, 3, 5])  # 3个颜色通道
img_norm = trans_norm(img_tensor)
print(img_norm[0][0][0])
writer.add_image("Normalize", img_norm, 2)

# Resize使用
print(img.size)  # (3200, 1800)
trans_resize = transforms.Resize((512, 512))  # 注意这里一定要多加一个括号
img_resize = trans_resize(img)
print(img_resize)  # <PIL.Image.Image image mode=RGB size=512x512 at 0x15D57A081C0>
# img PIL ---->  Resize ----> img_resize PIL ----> ToTensor ----> img_resize tensor
# 要用tensorboard显示要将PIL的类型转换为tensor类型
img_resize = trans_totensor(img_resize)
writer.add_image("Resize", img_resize, 0)
print(img_resize)

#  Compose - resize -2
trans_resize_2 = transforms.Resize(512)
# PIL  ---->   PIL  ---->tensor
trans_compose = transforms.Compose([trans_resize_2, trans_totensor])  # 注意一定要加中括号
img_resize_2 = trans_compose(img)
writer.add_image("Resize", img_resize_2, 1)

# RandomCrop随机裁剪使用
# trans_random = transforms.RandomCrop(512)
trans_random = transforms.RandomCrop((500, 1000))  # 注意2个括号，裁剪500X1000的
trans_compose_2 = transforms.Compose([trans_random, trans_totensor])
# 裁剪10个
for i in range(10):
    img_crop = trans_compose_2(img)
    writer.add_image("RandomCropHW", img_crop, i)

writer.close()