基于Pytorch深度学习神经网络MNIST手写数字识别系统源码（带界面和手写画板）

第一步：准备数据

mnist开源数据集

第二步：搭建模型

我们这里搭建了一个LeNet5网络

参考代码如下：

import torch
from torch import nn


class Reshape(nn.Module):
    def forward(self, x):
        return x.view(-1, 1, 28, 28)


class LeNet5(nn.Module):
    def __init__(self):
        super(LeNet5, self).__init__()
        self.net = nn.Sequential(
            Reshape(),

            # CONV1, ReLU1, POOL1
            nn.Conv2d(in_channels=1, out_channels=6, kernel_size=5, padding=2),
            # nn.Conv2d(in_channels=1, out_channels=6, kernel_size=5),
            nn.ReLU(),
            nn.MaxPool2d(kernel_size=2, stride=2),

            # CONV2, ReLU2, POOL2
            nn.Conv2d(in_channels=6, out_channels=16, kernel_size=5),
            nn.ReLU(),
            nn.MaxPool2d(kernel_size=2, stride=2),
            nn.Flatten(),

            # FC1
            nn.Linear(in_features=16 * 5 * 5, out_features=120),
            nn.ReLU(),

            # FC2
            nn.Linear(in_features=120, out_features=84),
            nn.ReLU(),

            # FC3
            nn.Linear(in_features=84, out_features=10)
        )
        # 添加softmax层
        self.softmax = nn.Softmax()

    def forward(self, x):
        logits = self.net(x)
        # 将logits转为概率
        prob = self.softmax(logits)
        return prob


if __name__ == '__main__':
	model = LeNet5()
	X = torch.rand(size=(256, 1, 28, 28), dtype=torch.float32)
	for layer in model.net:
	    X = layer(X)
	    print(layer.__class__.__name__, '\toutput shape: \t', X.shape)
	X = torch.rand(size=(1, 1, 28, 28), dtype=torch.float32)
	print(model(X))

第三步：训练代码

import torch
from torch import nn
from torchvision import datasets
from torchvision.transforms import ToTensor
from torch.utils.data import DataLoader

from model import LeNet5


# DATASET
train_data = datasets.MNIST(
	root='./data',
	train=False,
	download=True,
	transform=ToTensor()
)

test_data = datasets.MNIST(
	root='./data',
	train=False,
	download=True,
	transform=ToTensor()
)


# PREPROCESS
batch_size = 256
train_dataloader = DataLoader(dataset=train_data, batch_size=batch_size)
test_dataloader = DataLoader(dataset=test_data, batch_size=batch_size)
for X, y in train_dataloader:
	print(X.shape)		# torch.Size([256, 1, 28, 28])
	print(y.shape)		# torch.Size([256])
	break


# MODEL
device = 'cuda' if torch.cuda.is_available() else 'cpu'
model = LeNet5().to(device)


# TRAIN MODEL
loss_func = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(params=model.parameters())

def train(dataloader, model, loss_func, optimizer, epoch):
	model.train()
	data_size = len(dataloader.dataset)
	for batch, (X, y) in enumerate(dataloader):
		X, y = X.to(device), y.to(device)

		y_hat = model(X)
		loss = loss_func(y_hat, y)

		optimizer.zero_grad()
		loss.backward()
		optimizer.step()

	loss, current = loss.item(), batch * len(X)
	print(f'EPOCH{epoch+1}\tloss: {loss:>7f}', end='\t')


# Test model
def test(dataloader, model, loss_fn):
	size = len(dataloader.dataset)
	num_batches = len(dataloader)
	model.eval()
	test_loss, correct = 0, 0
	with torch.no_grad():
		for X, y in dataloader:
			X, y = X.to(device), y.to(device)
			pred = model(X)
			test_loss += loss_fn(pred, y).item()
			correct += (pred.argmax(1) == y).type(torch.float).sum().item()
	test_loss /= num_batches
	correct /= size
	print(f'Test Error: Accuracy: {(100 * correct):>0.1f}%, Average loss: {test_loss:>8f}\n')


if __name__ == '__main__':
	epoches = 80
	for epoch in range(epoches):
		train(train_dataloader, model, loss_func, optimizer, epoch)
		test(test_dataloader, model, loss_func)

	# Save models
	torch.save(model.state_dict(), 'model.pth')
	print('Saved PyTorch LeNet5 State to model.pth')

第四步：统计训练过程

EPOCH1	loss: 1.908403	Test Error: Accuracy: 58.3%, Average loss: 1.943602

EPOCH2	loss: 1.776060	Test Error: Accuracy: 72.2%, Average loss: 1.750917

EPOCH3	loss: 1.717706	Test Error: Accuracy: 73.6%, Average loss: 1.730332

EPOCH4	loss: 1.719344	Test Error: Accuracy: 76.0%, Average loss: 1.703456

EPOCH5	loss: 1.659312	Test Error: Accuracy: 76.6%, Average loss: 1.694500

EPOCH6	loss: 1.647946	Test Error: Accuracy: 76.9%, Average loss: 1.691286

EPOCH7	loss: 1.653712	Test Error: Accuracy: 77.0%, Average loss: 1.690819

EPOCH8	loss: 1.653270	Test Error: Accuracy: 76.8%, Average loss: 1.692459

EPOCH9	loss: 1.649021	Test Error: Accuracy: 77.5%, Average loss: 1.686158

EPOCH10	loss: 1.648204	Test Error: Accuracy: 78.3%, Average loss: 1.678802

EPOCH11	loss: 1.647159	Test Error: Accuracy: 78.4%, Average loss: 1.676133

EPOCH12	loss: 1.647390	Test Error: Accuracy: 78.6%, Average loss: 1.674455

EPOCH13	loss: 1.646807	Test Error: Accuracy: 78.4%, Average loss: 1.675752

EPOCH14	loss: 1.630824	Test Error: Accuracy: 79.1%, Average loss: 1.668470

EPOCH15	loss: 1.524222	Test Error: Accuracy: 86.3%, Average loss: 1.599240

EPOCH16	loss: 1.524022	Test Error: Accuracy: 86.7%, Average loss: 1.594947

EPOCH17	loss: 1.524296	Test Error: Accuracy: 87.1%, Average loss: 1.588946

EPOCH18	loss: 1.523599	Test Error: Accuracy: 87.3%, Average loss: 1.588275

EPOCH19	loss: 1.523655	Test Error: Accuracy: 87.5%, Average loss: 1.586576

EPOCH20	loss: 1.523659	Test Error: Accuracy: 88.2%, Average loss: 1.579286

EPOCH21	loss: 1.523733	Test Error: Accuracy: 87.9%, Average loss: 1.582472

EPOCH22	loss: 1.523748	Test Error: Accuracy: 88.2%, Average loss: 1.578699

EPOCH23	loss: 1.523788	Test Error: Accuracy: 88.0%, Average loss: 1.579700

EPOCH24	loss: 1.523708	Test Error: Accuracy: 88.1%, Average loss: 1.579758

EPOCH25	loss: 1.523683	Test Error: Accuracy: 88.4%, Average loss: 1.575913

EPOCH26	loss: 1.523646	Test Error: Accuracy: 88.7%, Average loss: 1.572831

EPOCH27	loss: 1.523654	Test Error: Accuracy: 88.9%, Average loss: 1.570528

EPOCH28	loss: 1.523642	Test Error: Accuracy: 89.0%, Average loss: 1.570223

EPOCH29	loss: 1.523663	Test Error: Accuracy: 89.0%, Average loss: 1.570385

EPOCH30	loss: 1.523658	Test Error: Accuracy: 88.9%, Average loss: 1.571195

EPOCH31	loss: 1.523653	Test Error: Accuracy: 88.4%, Average loss: 1.575981

EPOCH32	loss: 1.523653	Test Error: Accuracy: 89.0%, Average loss: 1.570087

EPOCH33	loss: 1.523642	Test Error: Accuracy: 88.9%, Average loss: 1.571018

EPOCH34	loss: 1.523649	Test Error: Accuracy: 89.0%, Average loss: 1.570439

EPOCH35	loss: 1.523629	Test Error: Accuracy: 90.4%, Average loss: 1.555473

EPOCH36	loss: 1.461187	Test Error: Accuracy: 97.1%, Average loss: 1.491042

EPOCH37	loss: 1.461230	Test Error: Accuracy: 97.7%, Average loss: 1.485049

EPOCH38	loss: 1.461184	Test Error: Accuracy: 97.7%, Average loss: 1.485653

EPOCH39	loss: 1.461156	Test Error: Accuracy: 98.2%, Average loss: 1.479966

EPOCH40	loss: 1.461335	Test Error: Accuracy: 98.2%, Average loss: 1.479197

EPOCH41	loss: 1.461152	Test Error: Accuracy: 98.7%, Average loss: 1.475477

EPOCH42	loss: 1.461153	Test Error: Accuracy: 98.7%, Average loss: 1.475124

EPOCH43	loss: 1.461153	Test Error: Accuracy: 98.9%, Average loss: 1.472885

EPOCH44	loss: 1.461151	Test Error: Accuracy: 99.1%, Average loss: 1.470957

EPOCH45	loss: 1.461156	Test Error: Accuracy: 99.1%, Average loss: 1.471141

EPOCH46	loss: 1.461152	Test Error: Accuracy: 99.1%, Average loss: 1.470793

EPOCH47	loss: 1.461151	Test Error: Accuracy: 98.8%, Average loss: 1.474548

EPOCH48	loss: 1.461151	Test Error: Accuracy: 99.1%, Average loss: 1.470666

EPOCH49	loss: 1.461151	Test Error: Accuracy: 99.1%, Average loss: 1.471546

EPOCH50	loss: 1.461151	Test Error: Accuracy: 99.0%, Average loss: 1.471407

EPOCH51	loss: 1.461151	Test Error: Accuracy: 98.8%, Average loss: 1.473795

EPOCH52	loss: 1.461164	Test Error: Accuracy: 98.2%, Average loss: 1.480009

EPOCH53	loss: 1.461151	Test Error: Accuracy: 99.2%, Average loss: 1.469931

EPOCH54	loss: 1.461152	Test Error: Accuracy: 99.2%, Average loss: 1.469916

EPOCH55	loss: 1.461151	Test Error: Accuracy: 98.9%, Average loss: 1.472574

EPOCH56	loss: 1.461151	Test Error: Accuracy: 98.6%, Average loss: 1.476035

EPOCH57	loss: 1.461151	Test Error: Accuracy: 98.2%, Average loss: 1.478933

EPOCH58	loss: 1.461150	Test Error: Accuracy: 99.4%, Average loss: 1.468186

EPOCH59	loss: 1.461151	Test Error: Accuracy: 99.4%, Average loss: 1.467602

EPOCH60	loss: 1.461151	Test Error: Accuracy: 99.1%, Average loss: 1.471206

EPOCH61	loss: 1.461151	Test Error: Accuracy: 98.8%, Average loss: 1.473356

EPOCH62	loss: 1.461151	Test Error: Accuracy: 99.2%, Average loss: 1.470242

EPOCH63	loss: 1.461150	Test Error: Accuracy: 99.1%, Average loss: 1.470826

EPOCH64	loss: 1.461151	Test Error: Accuracy: 98.7%, Average loss: 1.474476

EPOCH65	loss: 1.461150	Test Error: Accuracy: 99.3%, Average loss: 1.469116

EPOCH66	loss: 1.461150	Test Error: Accuracy: 99.4%, Average loss: 1.467823

EPOCH67	loss: 1.461150	Test Error: Accuracy: 99.5%, Average loss: 1.466486

EPOCH68	loss: 1.461152	Test Error: Accuracy: 99.3%, Average loss: 1.468688

EPOCH69	loss: 1.461150	Test Error: Accuracy: 99.5%, Average loss: 1.466256

EPOCH70	loss: 1.461150	Test Error: Accuracy: 99.5%, Average loss: 1.466588

EPOCH71	loss: 1.461150	Test Error: Accuracy: 99.6%, Average loss: 1.465280

EPOCH72	loss: 1.461150	Test Error: Accuracy: 99.4%, Average loss: 1.467110

EPOCH73	loss: 1.461151	Test Error: Accuracy: 99.6%, Average loss: 1.465245

EPOCH74	loss: 1.461150	Test Error: Accuracy: 99.5%, Average loss: 1.466551

EPOCH75	loss: 1.461150	Test Error: Accuracy: 99.5%, Average loss: 1.466001

EPOCH76	loss: 1.461150	Test Error: Accuracy: 99.3%, Average loss: 1.468074

EPOCH77	loss: 1.461151	Test Error: Accuracy: 99.6%, Average loss: 1.465709

EPOCH78	loss: 1.461150	Test Error: Accuracy: 99.5%, Average loss: 1.466567

EPOCH79	loss: 1.461150	Test Error: Accuracy: 99.6%, Average loss: 1.464922

EPOCH80	loss: 1.461150	Test Error: Accuracy: 99.6%, Average loss: 1.465109

第五步：搭建GUI界面