深度学习作业-基于pytorch框架python实现手写数字识别完整源码+代码注释+实验报告.zip

#!/usr/bin/env python # -*- coding: utf-8 -*- """ @Project ：深度学习 @File ：实验一手写体识别.py @Author ：郑家祥_202028019410032 @Date ：2021/5/17 21:57 @Description：利用MNIST数据集在Pytorch框架上进行手写数字体识别，测试误差达到98%及以上 """ import torch import torchvision from torch import nn import torch.optim as optim from torch.utils import data from torchvision import transforms import torch.utils.data as Data from tensorboardX import SummaryWriter import hiddenlayer as hl def createBatchData(batch_size, resize=None): ''' 使用DataLoader创建随机批量数据 ''' trans = [transforms.ToTensor()] if resize: trans.insert(0, transforms.Resize(resize)) trans = transforms.Compose(trans) mnist_train = torchvision.datasets.MNIST(root='data', train=True, transform=trans, download=False) minst_test = torchvision.datasets.MNIST(root='data', train=False, transform=trans, download=False) return (data.DataLoader(mnist_train, batch_size, shuffle = True, num_workers=4), data.DataLoader(minst_test, batch_size, shuffle = False, num_workers=4)) class myNet(nn.Module): ''' 构建卷积神经网络,与LeNet类似 ''' def __init__(self): super(myNet, self).__init__() self.CNN1 = nn.Sequential( nn.Conv2d(1, 6, kernel_size=3, stride=1, padding=1),#6*28*28 nn.ReLU(inplace=True), nn.AvgPool2d(kernel_size=2,stride=2)#6*14*14 ) self.CNN2 = nn.Sequential( nn.Conv2d(6, 16, kernel_size=5, stride=1), #16*10*10 nn.ReLU(inplace=True), nn.AvgPool2d(kernel_size=2,stride=2) #16*5*5 ) self.FC1 = nn.Sequential( nn.Flatten(), nn.Linear(400, 120), nn.ReLU(inplace=True), nn.Linear(120, 84), nn.ReLU(inplace=True), nn.Linear(84, 10) ) def reshape_(self, x): return x.reshape(-1, 1, 28, 28) def forward(self, x): x = self.reshape_(x) x = self.CNN1(x) x = self.CNN2(x) x = self.FC1(x) return x def sameNumber(y_hat, y): ''' 返回预测值与真实值相等的个数 ''' if len(y_hat.shape) > 1 and y_hat.shape[1] > 1: y_hat = y_hat.argmax(axis=1) cmp = y_hat.type(y.dtype) == y return float(cmp.type(y.dtype).sum()) class Accumulator(): ''' 构建n列变量，每列累加 ''' def __init__(self, n): self.data = [0.0] * n def add(self, *args): self.data = [a + float(b) for a, b in zip(self.data, args)] def reset(self): self.data = [0.0] * len(self.data) def __getitem__(self, index): return self.data[index] def testNet(net, testData): ''' 计算测试误差 ''' if isinstance(net, torch.nn.Module): net.eval() metric = Accumulator(2) for X, y in testData: X, y = X.to(device), y.to(device) y_hat = net(X) sameNum = sameNumber(y_hat, y) metric.add(sameNum, y.numel()) return metric[0] / metric[1] def trainNet(net, trainData, testData, num_epochs, lr, device): ''' 使用trainData训练网络 ''' def init(m): if type(m) == nn.Linear or type(m) == nn.Conv2d: nn.init.xavier_uniform_(m.weight) net.apply(init) net.to(device) optimizer = optim.SGD(net.parameters(), lr=lr) lossFun = nn.CrossEntropyLoss() writer = SummaryWriter(logdir='assets/visualize', comment="test1") for epoch in range(num_epochs): metric = Accumulator(3) net.train() for i, (X, y) in enumerate(trainData): X, y = X.to(device), y.to(device) y_hat = net(X) loss = lossFun(y_hat, y) optimizer.zero_grad() loss.sum().backward() optimizer.step() with torch.no_grad(): metric.add(loss * X.shape[0], sameNumber(y_hat, y), X.shape[0]) train_loss = metric[0] / metric[2] train_acc = metric[1] / metric[2] test_acc = testNet(net, testData) writer.add_scalar('TrainLoss', train_loss, epoch) writer.add_scalar('TrainAccuracy', train_acc, epoch) writer.add_scalar('TestAccuracy', test_acc, epoch) print(f'loss {train_loss:.3f}, train acc {train_acc:.3f}, 'f'test acc {test_acc:.3f}') if __name__ == '__main__': batch_size = 256 num_epochs = 100 lr = 0.01 device = torch.device('cuda' if torch.cuda.is_available else 'cpu') trainDataIter, testDataIter = createBatchData(batch_size); net = myNet() ''' modelVis = hl.build_graph(net, torch.zeros([1 ,1, 28, 28])) modelVis.theme = hl.graph.THEMES["blue"].copy() modelVis.save("assets/test2",format='png') ''' trainNet(net, trainDataIter, testDataIter, num_epochs=num_epochs, lr=lr, device=device);