pytorch 加载mnist,cifar2,飞机车鸟

最新推荐文章于 2024-03-24 11:38:36 发布

小杨变老杨

最新推荐文章于 2024-03-24 11:38:36 发布

阅读量200

点赞数

文章标签： pytorch 深度学习人工智能

本文链接：https://blog.csdn.net/m0_67084346/article/details/128086623

版权

# torch加载cifar2训练集
# pytorch数据管道
import torch
from torch.utils.data import DataLoader
from torchvision import transforms,datasets
# transform= 对数据特征进行操作
# target_transform= 对数据标签进行操作
# （10000，32，32，3）
train_data = datasets.ImageFolder("../data/cifar2/train/",
                                transform=transforms.ToTensor(),
                                target_transform=lambda t:torch.tensor(t).float())

test_data = datasets.ImageFolder("../data/cifar2/test/",
                               transform=transforms.ToTensor(),
                               target_transform=lambda t:torch.tensor(t).float())
# 维度调整 （100, 3, 32, 32）* 100
train_data = DataLoader(train_data,batch_size=100,shuffle=True)
test_data = DataLoader(test_data,batch_size=100,shuffle=True)

for i in range(epoch)：
    for j, (imgs, labels) in enumerate(train_data):
        img = Variable(imgs)
        label = Variable(labels)
        optimizer.zero_grad()

# pytorch 加载内置数据集
import torch
from torch.utils.data import DataLoader
from torchvision import datasets, transforms
# MNIST dataset (60000,28,28)
# 其他常用内置数据集：datasets.FashionMNIST, datasets.CIFAR10, ...
mnist_train = datasets.MNIST(root='MNIST_data/',
                             train=True,
                             transform=transforms.ToTensor(),
                             target_transform=lambda t: torch.tensor(t).long(),
                             download=True)

mnist_test = datasets.MNIST(root='MNIST_data/',
                            train=False,
                            transform=transforms.ToTensor(),
                            target_transform=lambda t: torch.tensor(t).long(),
                            download=True)

# dataset loader (100, 1, 28, 28) * 600
train_data = DataLoader(dataset=mnist_train, batch_size=100, shuffle=True)
test_data = DataLoader(dataset=mnist_test, batch_size=100, shuffle=True)
# 应用演示
for epoch in range(epochs):
    for i, (batch_xs, batch_ys) in enumerate(train_data):
        X = Variable(batch_xs)  # （100，1，28，28）
        Y = Variable(batch_ys)  # （100，）

#飞机车鸟数据集
import os
import numpy as np
from sklearn.model_selection import train_test_split
import torch
import torch.nn as nn
from torch.autograd import Variable
import matplotlib.pyplot as plt
from torchvision import transforms
import torchvision.datasets as datasets
import random
from torch.optim.lr_scheduler import *
np.random.seed(0)
torch.manual_seed(0)
learning_rate=0.01
epochs=10
train_test=0.8
#data=r'E:\p1code\study\shendu2\0705   week+opencv/'
data=r'../data/data3'
imgdir = '../data/data3'
# 依次读取图像像素值并进行归一化处理
def loadimg(imgpath):
    # 读取图像像素值
    img = plt.imread(imgpath)
    img = img/255.0 # 归一化
    return img
imgs = [] # x
labels = [] # y
for path in os.listdir(imgdir):
    # imgdir + '/' + path = ../data/data3/'0_6.jpg'
    img = loadimg(imgdir + '/' + path)
    imgs.append(img)
    labels.append(int(path[0]))
# labels = utils.to_categorical(labels, 3)
# train_test_split只接受numpy.array类
imgs = np.array(imgs)
labels = np.array(labels)
train_x, test_x, train_y, test_y = train_test_split(imgs, labels, train_size=0.7, shuffle=True)

train_x=Variable(torch.Tensor(train_x.transpose([0,3,1,2])))
train_y=Variable(torch.Tensor(train_y))
test_x=Variable(torch.Tensor(test_x.transpose([0,3,1,2])))
test_y=Variable(torch.Tensor(test_y))