软件定制和开发_禅城南庄网站制作_有哪些网站可以免费推广_三亚百度推广公司电话

完整的神经网络模型训练过程

1.导入常用的库和函数

import osimport randomimport numpy as npimport torchimport torch.nn as nnfrom torch.utils.data import DataLoaderimport torchvision.transforms as transformsimport torch.optim as optimfrom matplotlib import pyplot as pltfrom model.lenet import LeNetfrom tools.my_dataset import RMBDataset

2.定义超参数（包括标签）

def set_seed(seed=1):random.seed(seed)np.random.seed(seed)torch.manual_seed(seed)torch.cuda.manual_seed(seed)set_seed()  # 设置随机种子rmb_label = {"1": 0, "100": 1}# 参数设置MAX_EPOCH = 10BATCH_SIZE = 16LR = 0.01log_interval = 10val_interval = 1

3.数据处理

3.1数据切分-找到或创建数据的位置

split_dir = os.path.join("..", "..", "data", "rmb_split")train_dir = os.path.join(split_dir, "train")valid_dir = os.path.join(split_dir, "valid")

3.2 定义数据的变换形式——transforms

norm_mean = [0.485, 0.456, 0.406]norm_std = [0.229, 0.224, 0.225]train_transform = transforms.Compose([transforms.Resize((32, 32)),transforms.RandomCrop(32, padding=4),transforms.ToTensor(),transforms.Normalize(norm_mean, norm_std),])valid_transform = transforms.Compose([transforms.Resize((32, 32)),transforms.ToTensor(),transforms.Normalize(norm_mean, norm_std),])

3.3 构建真实数据（利用X-Dataset类）

将数据路径和transforms变换形式作为参数传到X-Dataset类里，得到数据实例

train_data = RMBDataset(data_dir=train_dir, transform=train_transform)# 这里的RMBDataset是一个类，主要做两件事，一是根据数据路径便利获取数据信息，包括每张图片的路径和标签。@staticmethoddef get_img_info(data_dir):data_info = list()for root, dirs, _ in os.walk(data_dir):# 遍历类别for sub_dir in dirs:img_names = os.listdir(os.path.join(root, sub_dir))img_names = list(filter(lambda x: x.endswith('.jpg'), img_names))# 遍历图片for i in range(len(img_names)):img_name = img_names[i]path_img = os.path.join(root, sub_dir, img_name)label = rmb_label[sub_dir]data_info.append((path_img, int(label)))return data_info# 二是对获取的图片调用函数transforms进行变换处理def __getitem__(self, index):path_img, label = self.data_info[index]img = Image.open(path_img).convert('RGB')     # 0~255if self.transform is not None:img = self.transform(img)   # 在这里做transform，转为tensor等等

3.4 构建DataLoader（后续训练遍历图片）

将变换得到的真实数据作为参数传到DataLoader函数里，并设置超参数batch_size等。

train_loader = DataLoader(dataset=train_data, batch_size=BATCH_SIZE, shuffle=True)

4.构建模型

主要包括两步，一是实例化网络模型

net = LeNet(classes=2)

二是权重初始化

net.initialize_weights()

5.设置损失函数

criterion = nn.CrossEntropyLoss()         

6.设置优化器

optimizer = optim.SGD(net.parameters(), lr=LR, momentum=0.9)                        # 选择优化器scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=10, gamma=0.1)     # 设置学习率下降策略

7.训练

7.1 定义训练曲线存储list

train_curve = list()valid_curve = list()

7.2 训练过程

# 1.大循环，在epoch里for epoch in range(MAX_EPOCH):#2.定义收集训练结果的参数loss_mean = 0.correct = 0.total = 0.#3.开始训练net.train()#4.遍历数据，其中参数就是DataLoader的实例化参数for i, data in enumerate(train_loader):#5.前向传播，从数据中选择输入和标签，将输入传到net里# forwardinputs, labels = dataoutputs = net(inputs)#5.后向传播，优化器梯度置0，计算损失函数（输出与标签作为参数），然后损失函数后向传播计算梯度# backwardoptimizer.zero_grad()loss = criterion(outputs, labels)loss.backward()#6.更新梯度# update weightsoptimizer.step()#7.统计输出结果的准确性# 统计分类情况_, predicted = torch.max(outputs.data, 1)total += labels.size(0)correct += (predicted == labels).squeeze().sum().numpy()# 8.打印训练信息loss_mean += loss.item()train_curve.append(loss.item())if (i+1) % log_interval == 0:loss_mean = loss_mean / log_intervalprint("Training:Epoch[{:0>3}/{:0>3}] Iteration[{:0>3}/{:0>3}] Loss: {:.4f} Acc:{:.2%}".format(epoch, MAX_EPOCH, i+1, len(train_loader), loss_mean, correct / total))loss_mean = 0.#9.更新学习率scheduler.step()  # 更新学习率#10.验证模型，和获取训练结果一样的步骤，只是图片路径在valid里# validate the modelif (epoch+1) % val_interval == 0:correct_val = 0.total_val = 0.loss_val = 0.net.eval()with torch.no_grad():for j, data in enumerate(valid_loader):inputs, labels = dataoutputs = net(inputs)loss = criterion(outputs, labels)_, predicted = torch.max(outputs.data, 1)total_val += labels.size(0)correct_val += (predicted == labels).squeeze().sum().numpy()loss_val += loss.item()loss_val_epoch = loss_val / len(valid_loader)valid_curve.append(loss_val_epoch)# valid_curve.append(loss.item())    # 20191022改，记录整个epoch样本的loss，注意要取平均print("Valid:\t Epoch[{:0>3}/{:0>3}] Iteration[{:0>3}/{:0>3}] Loss: {:.4f} Acc:{:.2%}".format(epoch, MAX_EPOCH, j+1, len(valid_loader), loss_val_epoch, correct_val / total_val))

7.3 训练过程可视化

train_x = range(len(train_curve))train_y = train_curvetrain_iters = len(train_loader)valid_x = np.arange(1, len(valid_curve)+1) * train_iters*val_interval # 由于valid中记录的是epochloss，需要对记录点进行转换到iterationsvalid_y = valid_curveplt.plot(train_x, train_y, label='Train')plt.plot(valid_x, valid_y, label='Valid')plt.legend(loc='upper right')plt.ylabel('loss value')plt.xlabel('Iteration')plt.show()

8.推理（简单版）

直接利用前面训练的网络模型，将test图片输入得到输出结果即可。

BASE_DIR = os.path.dirname(os.path.abspath(__file__))test_dir = os.path.join(BASE_DIR, "test_data")test_data = RMBDataset(data_dir=test_dir, transform=valid_transform)valid_loader = DataLoader(dataset=test_data, batch_size=1)for i, data in enumerate(valid_loader):# forwardinputs, labels = dataoutputs = net(inputs)_, predicted = torch.max(outputs.data, 1)rmb = 1 if predicted.numpy()[0] == 0 else 100print("模型获得{}元".format(rmb))