first commit

exp2/datasets/.gitignore

@@ -0,0 +1,2 @@
+# Ignore datasets file
+mnist.npz

exp2/modified/train.py

@@ -0,0 +1,139 @@
+from pathlib import Path
+import sys
+import numpy
+import torch
+from torch.utils.data import DataLoader, Dataset
+import matplotlib.pyplot as plt
+import torch.nn.functional as F
+
+sys.path.append(str(Path(__file__).resolve().parent.parent.parent))
+import gpu_utils
+
+
+class CNN(torch.nn.Module):
+    """卷积神经网络模型"""
+
+    def __init__(self):
+        super(CNN, self).__init__()
+
+        # 使用Ceil模式设置MaxPooling，因为tensorflow默认是这个模式。
+        self.conv1 = torch.nn.Conv2d(1, 32, kernel_size=(3, 3))
+        self.pool1 = torch.nn.MaxPool2d(kernel_size=(2, 2), ceil_mode=True)
+        self.conv2 = torch.nn.Conv2d(32, 64, kernel_size=(3, 3))
+        self.pool2 = torch.nn.MaxPool2d(kernel_size=(2, 2), ceil_mode=True)
+        self.conv3 = torch.nn.Conv2d(64, 64, kernel_size=(3, 3))
+        self.flatten = torch.nn.Flatten()
+        # 28x28过第一轮卷积后变为26x26，过第一轮池化后变为13x13
+        # 过第二轮卷积后变为11x11，过第二轮池化后变为5x5
+        # 过第三轮卷积后变为3x3。
+        # 最后一轮卷积核个数为64。
+        self.fc1 = torch.nn.Linear(64 * 3 * 3, 64)
+        self.fc2 = torch.nn.Linear(64, 10)
+
+    def forward(self, x):
+        x = F.relu(self.conv1(x))
+        x = self.pool1(x)
+        x = F.relu(self.conv2(x))
+        x = self.pool2(x)
+        x = F.relu(self.conv3(x))
+        x = self.flatten(x)
+        x = F.relu(self.fc1(x))
+        x = self.fc2(x)
+        return F.softmax(x, dim=1)
+
+
+class MnistDataset(Dataset):
+    """用于加载Mnist的自定义数据集"""
+
+    shape: int
+    x_data: torch.Tensor
+    y_data: torch.Tensor
+
+    def __init__(self, x_data: torch.Tensor, y_data: torch.Tensor):
+        x_len = x_data.shape[0]
+        y_len = y_data.shape[0]
+        assert (x_len == y_len)
+        self.shape = x_len
+
+        self.x_data = x_data
+        self.y_data = y_data
+
+    def __getitem__(self, index):
+        return self.x_data[index], self.y_data[index]
+
+    def __len__(self):
+        return self.shape
+
+
+class DataSource:
+    """用于读取MNIST数据的数据读取器"""
+
+    train_data: DataLoader
+    test_data: DataLoader
+
+    def __init__(self, batch_size: int):
+        datasets_path = Path(
+            __file__).resolve().parent.parent / 'datasets' / 'mnist.npz'
+        datasets = numpy.load(datasets_path)
+
+        # 6万张训练图片：60000x28x28。标签只有第一维。
+        train_images = torch.from_numpy(datasets['x_train'])
+        train_label = torch.from_numpy(datasets['y_train'])
+        # 1万张测试图片：10000x28x28。标签只有第一维。
+        test_images = torch.from_numpy(datasets['x_test'])
+        test_label = torch.from_numpy(datasets['y_test'])
+
+        # 为了符合后面图像的输入颜色通道条件，要在最后挤出一个新的维度
+        train_images.unsqueeze(-1)
+        test_images.unsqueeze(-1)
+        # 像素值归一化
+        train_images /= 255.0
+        test_images /= 255.0
+
+        # 创建数据集
+        train_dataset = MnistDataset(train_images, train_label)
+        test_dataset = MnistDataset(test_images, test_label)
+
+        # 赋值到自身
+        self.train_data = DataLoader(dataset=train_dataset,
+                                     batch_size=batch_size,
+                                     shuffle=True)
+        self.test_data = DataLoader(dataset=test_dataset,
+                                    batch_size=batch_size,
+                                    shuffle=False)
+
+
+def main():
+    n_epoch = 5
+    n_batch_size = 25
+
+    device = gpu_utils.get_gpu_device()
+    data_source = DataSource(n_batch_size)
+    cnn = CNN().to(device)
+
+    optimizer = torch.optim.Adam(cnn.parameters())
+    loss_func = torch.nn.CrossEntropyLoss()
+
+    for epoch in range(n_epoch):
+        cnn.train()
+
+        batch_images: torch.Tensor
+        batch_labels: torch.Tensor
+        for batch_index, (batch_images, batch_labels) in enumerate(data_source.train_data):
+            gpu_images = batch_images.to(device)
+            gpu_labels = batch_labels.to(device)
+
+            optimizer.zero_grad()
+            prediction: torch.Tensor = cnn(gpu_images)
+            loss: torch.Tensor = loss_func(prediction, gpu_labels)
+            loss.backward()
+            optimizer.step()
+
+            loss_showcase = loss.item()
+            print(f'Epoch: {epoch+1}, Batch: {batch_index}, Loss: {loss.item():.4f}')
+
+
+
+if __name__ == "__main__":
+    gpu_utils.print_gpu_availability()
+    main()

exp2/source/predict.py

@@ -0,0 +1,45 @@
+import tensorflow as tf
+from PIL import Image
+import numpy as np
+import matplotlib.pyplot as plt
+
+from train import CNN
+
+'''
+python 3.9 
+tensorflow 2.0.0b0
+pillow(PIL) 4.3.0
+''' 
+
+
+class Predict(object):
+    def __init__(self):
+        latest = tf.train.latest_checkpoint('./ckpt')
+        self.cnn = CNN()
+        # 恢复网络权重
+        self.cnn.model.load_weights(latest)
+
+    def predict(self, image_path):
+        # 以黑白方式读取图片
+        img = Image.open(image_path).convert('L')
+        img = np.reshape(img, (28, 28, 1)) / 255.
+        x = np.array([1 - img])
+        y = self.cnn.model.predict(x)
+
+        # 因为x只传入了一张图片，取y[0]即可
+        # np.argmax()取得最大值的下标，即代表的数字
+        print(image_path)
+#        print(y[0])
+        print('        -> Predict digit', np.argmax(y[0]))
+        plt.figure("Image")  # 图像窗口名称
+        plt.imshow(img)
+        plt.axis('on')  # 关掉坐标轴为 off
+        plt.title(np.argmax(y[0]))  # 图像题目 # 必须有这个，要不然无法显示
+        plt.show()
+
+
+if __name__ == "__main__":
+    app = Predict()
+    app.predict('./test_images/0.png') 
+    app.predict('./test_images/1.png')
+    app.predict('./test_images/4.png')

exp2/source/train.py

@@ -0,0 +1,61 @@
+import os
+import tensorflow as tf
+from tensorflow.keras import datasets, layers, models
+
+'''
+python 3.9
+tensorflow 2.0.0b0
+''' 
+class CNN(object):
+    def __init__(self):
+        model = models.Sequential()
+        # 第1层卷积，卷积核大小为3*3，32个，28*28为待训练图片的大小
+        model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1)))
+        model.add(layers.MaxPooling2D((2, 2)))
+        # 第2层卷积，卷积核大小为3*3，64个
+        model.add(layers.Conv2D(64, (3, 3), activation='relu'))
+        model.add(layers.MaxPooling2D((2, 2)))
+        # 第三层卷积，卷积核大小为3*3，64个
+        model.add(layers.Conv2D(64, (3, 3), activation='relu'))
+        model.add(layers.Flatten())
+        model.add(layers.Dense(64, activation='relu'))
+        model.add(layers.Dense(10, activation='softmax'))
+        model.summary()
+        self.model = model
+
+class DataSource(object):
+    def __init__(self):
+        # mnist数据集存储的位置，如何不存在将自动下载
+        data_path = os.path.abspath(os.path.dirname(
+            __file__)) + '.'
+        (train_images, train_labels), (test_images,
+                                       test_labels) = datasets.mnist.load_data(path=data_path)
+        # 6万张训练图片，1万张测试图片
+        train_images = train_images.reshape((60000, 28, 28, 1))
+        test_images = test_images.reshape((10000, 28, 28, 1))
+        # 像素值映射到 0 - 1 之间
+        train_images, test_images = train_images / 255.0, test_images / 255.0
+        self.train_images, self.train_labels = train_images, train_labels
+        self.test_images, self.test_labels = test_images, test_labels
+
+class Train:
+    def __init__(self):
+        self.cnn = CNN()
+        self.data = DataSource()
+    def train(self):
+        check_path = './ckpt/cp-{epoch:04d}.ckpt'
+        # period 每隔5epoch保存一次
+        save_model_cb = tf.keras.callbacks.ModelCheckpoint(
+            check_path, save_weights_only=True, verbose=1, period=5)
+        self.cnn.model.compile(optimizer='adam',
+                               loss='sparse_categorical_crossentropy',
+                               metrics=['accuracy'])
+        self.cnn.model.fit(self.data.train_images, self.data.train_labels,
+                           epochs=5, callbacks=[save_model_cb])
+        test_loss, test_acc = self.cnn.model.evaluate(
+            self.data.test_images, self.data.test_labels)
+        print("准确率: %.4f，共测试了%d张图片 " % (test_acc, len(self.data.test_labels)))
+
+if __name__ == "__main__":
+    app = Train()
+    app.train()