pytorch_transforms学习笔记（小土堆pytorch）

from PIL import Image
from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms
"""
观看了小土堆的pytorch，在transfors课时中学习常用的工具-> Totensor,Normalize,Resize
"""
writer = SummaryWriter("logs") 打开tensorboard可视化

img_path = "dataset/train/ants_image/6743948_2b8c096dda.jpg" #相对路径小土堆提供的数据集
img_PIL = Image.open(img_path) #->PIl 类型


# Totensor
tran_tensor = transforms.ToTensor()
img_tensor = tran_tensor(img_PIL) #->转化为tensor类型
writer.add_image("111",img_tensor)

#Normalize
trans_norm = transforms.Normalize([0.5,0.5,0.5],[0.5,0.5,0.5]) #mean = [0.5,0.5,0.5] 各通道均值 std=[0.5,0.5,0.5]
#各通道标准差 output = (input-mean)/std 实现归一化，归一化目的可使其为高斯分布，
img_norma = trans_norm(img_tensor)
print(img_norma)
writer.add_image("111",img_norma)

#resize

trans_resize = transforms.Resize((512,512))
img_resize = trans_resize(img_PIL)# 可以输入PIL类型或tensor张量 默认方法为双线性插值
print(img_resize)


#compose（）输入列表格式
trans_resize02 = transforms.Resize(200)
trans_compose = transforms.Compose([trans_resize02,tran_tensor])
img_compose = trans_compose(img_PIL)
print(img_compose.size)
writer.add_image("trans_compose",img_compose,0)


#Randomcorp 随机裁剪：在图片随机位置裁剪指定大小的图像块
trans_random = transforms.RandomCrop(122)
img_random = trans_random(img_tensor)#输入PIL或者tensor类型
writer.add_image("randomcorp",img_random)
writer.close()