PIL、cv2、numpy，和pytorch(torch)之间的转换

时间：2025-04-28 12:46:25

在图像处理和深度学习中，经常需要在PIL（Python Imaging Library）、OpenCV（cv2）、NumPy和PyTorch之间进行图像数据的转换。下面是这些库之间常见的转换方法。

1. PIL和numpy之间的转换

import numpy as np
from PIL import Image


image_path = "./"
pil_image = (image_path)
numpy_image = (pil_image)
#从PIL转换到numpy
pil_image = (numpy_image)
#从numpy转换到PIL

2. cv2和numpy之间的转换

import cv2
import numpy as np


image_path = "./"
opencv_image = (image_path)
numpy_image = (opencv_image)
#从cv2到numpy的转换
opencv_image = (numpy_image, cv2.COLOR_RGB2BGR)  
# 从numpy到cv2的转换。
#如果是RGB图像，要进行颜色空间转换

和pytorch之间的转换

import torch
import numpy as np

numpy_image = ()  
# NumPy数组
torch_image = torch.from_numpy(numpy_image)
#从numpy到torch的转换
numpy_image = torch_image.numpy()
#从torch到numpy的转换

和torch之间的转换

import torch
from PIL import Image
import  as transforms


image_path = "./"
pil_image = (image_path)

# 定义转换操作
transform = ()

# 应用转换操作
torch_image = transform(pil_image)
#从PIL到torch的转换
pil_image = ()(torch_image)
#从torch到PIL的转换

5.cv2和pytorch之间的转换

import torch
import cv2
import numpy as np
import  as transforms

image_path = "./"
cv2_image = (image_path)
cv2_image = (cv2_image, cv2.COLOR_BGR2RGB)  
# 转换为RGB格式

# 定义转换操作
transform = ([
    (),
])

torch_image = transform(cv2_image)
# 将NumPy数组转换为PyTorch张量
cv2_image = (torch_image, cv2.COLOR_RGB2BGR)
# 将torch转换为cv2

和cv2之间的转换

import cv2
from PIL import Image

image_path ="./"
pil_image = (image_path)

# PIL图像转换为cv2图像
cv2_image = ((pil_image), cv2.COLOR_RGB2BGR)
# cv2图像转换为PIL图像
pil_image = ((cv2_image, cv2.COLOR_BGR2RGB))

相关文章