手撕代码：ResNet实战

关于resnet原理的教程有很多，这里推荐一个~
https://www.bilibili.com/video/BV1T7411T7wa?from=search&%3Bseid=1879396105190151950

然后本次学习手写的代码是resnet18，数据集为CIFAR100，网络结构即为下图红框中的结构

图片截自开头提到的教程

代码

网络结构部分：

import  tensorflow as tf
from    tensorflow import keras
from    tensorflow.keras import layers, Sequential


# 构造含有一个短接层的residual结构
class BasicBlock(layers.Layer):    # 继承layers.Layer

    def __init__(self, filter_num, stride=1):  # filer_num即channel
        super(BasicBlock, self).__init__()     # 对继承自父类的属性进行初始化。而且是用父类的初始化方法来初始化继承的属性。
        # 子类新定义属性初始化
        self.conv1 = layers.Conv2D(filter_num, (3, 3), strides=stride, padding='same')
        self.bn1 = layers.BatchNormalization()
        self.relu = layers.Activation('relu')

        self.conv2 = layers.Conv2D(filter_num, (3, 3), strides=1, padding='same')
        self.bn2 = layers.BatchNormalization()

        if stride != 1:
            self.downsample = Sequential()
            self.downsample.add(layers.Conv2D(filter_num, (1, 1), strides=stride))
        else:
            self.downsample = lambda x:x  # 若stride=1，不需要下采样


    # 前向传播
    def call(self, inputs, training=None):

        # [b, h, w, c]
        out = self.conv1(inputs)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)
        # residual结构
        identity = self.downsample(inputs)
        # 短接层与卷积层的相加
        output = layers.add([out, identity])
        output = tf.nn.relu(output)

        return output

# 将多个BasicBlock连接形成Res Block
class ResNet(keras.Model):
         # 第一个参数为层数组合，Eg：[2, 2, 2, 2]；第二个参数为分类数量(全连接层)
    def __init__(self, layer_dims, num_classes=100):
        super(ResNet, self).__init__()

        # 预处理层
        self.stem = Sequential([layers.Conv2D(64, (3, 3), strides=(1, 1)),
                                layers.BatchNormalization(),
                                layers.Activation('relu'),
                                layers.MaxPool2D(pool_size=(2, 2), strides=(1, 1), padding='same')
                                ])
        # resnet主干部分
        self.layer1 = self.build_resblock(64,  layer_dims[0])        # Eg：【2，2，2，2】,第一个resblock就含有两个basblock
        self.layer2 = self.build_resblock(128, layer_dims[1], stride=2)
        self.layer3 = self.build_resblock(256, layer_dims[2], stride=2)
        self.layer4 = self.build_resblock(512, layer_dims[3], stride=2)

        # 全连接层    output: [b, 512, h, w],
        self.avgpool = layers.GlobalAveragePooling2D()  # 自适应确定输出，以输入全连接层
        self.fc = layers.Dense(num_classes)

    # 前向运算
    def call(self, inputs, training=None):
        
        x = self.stem(inputs)

        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)

        # [b, c]
        x = self.avgpool(x)
        # [b, 100]
        x = self.fc(x)

        return x


    def build_resblock(self, filter_num, blocks, stride=1):

        res_blocks = Sequential()
        # may down sample
        res_blocks.add(BasicBlock(filter_num, stride))
        # 后续的BasicBlock没有下采样，保持shape不变
        for _ in range(1, blocks):
            res_blocks.add(BasicBlock(filter_num, stride=1))

        return res_blocks

# 四个Res Block，其中每个包含两个Basicblock
def resnet18():
    return ResNet([2, 2, 2, 2])


def resnet34():
    return ResNet([3, 4, 6, 3])

训练部分

可和VGG的训练部分对比，学习代码逻辑结构

import  tensorflow as tf
from    tensorflow.keras import layers, optimizers, datasets, Sequential
import  os
from    resnet import resnet18   # 导入resnet18网络

os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
tf.random.set_seed(2345)


def preprocess(x, y):
    # [-1~1]
    x = tf.cast(x, dtype=tf.float32) / 255. - 0.5    # 调整这里可以改善梯度弥散现象
    y = tf.cast(y, dtype=tf.int32)
    return x,y


(x,y), (x_test, y_test) = datasets.cifar100.load_data()
y = tf.squeeze(y, axis=1)
y_test = tf.squeeze(y_test, axis=1)
print(x.shape, y.shape, x_test.shape, y_test.shape)


train_db = tf.data.Dataset.from_tensor_slices((x,y))
train_db = train_db.shuffle(1000).map(preprocess).batch(512)

test_db = tf.data.Dataset.from_tensor_slices((x_test,y_test))
test_db = test_db.map(preprocess).batch(512)

sample = next(iter(train_db))
print('sample:', sample[0].shape, sample[1].shape,
      tf.reduce_min(sample[0]), tf.reduce_max(sample[0]))


def main():

    # [b, 32, 32, 3] => [b, 1, 1, 512]
    model = resnet18()
    model.build(input_shape=(None, 32, 32, 3))
    model.summary()
    optimizer = optimizers.Adam(lr=1e-3)

    for epoch in range(500):

        for step, (x,y) in enumerate(train_db):

            with tf.GradientTape() as tape:
                # [b, 32, 32, 3] => [b, 100]
                logits = model(x)
                # [b] => [b, 100]
                y_onehot = tf.one_hot(y, depth=100)
                # compute loss
                loss = tf.losses.categorical_crossentropy(y_onehot, logits, from_logits=True)
                loss = tf.reduce_mean(loss)

            grads = tape.gradient(loss, model.trainable_variables)
            optimizer.apply_gradients(zip(grads, model.trainable_variables))

            if step %50 == 0:
                print(epoch, step, 'loss:', float(loss))



        total_num = 0
        total_correct = 0
        for x,y in test_db:

            logits = model(x)
            prob = tf.nn.softmax(logits, axis=1)
            pred = tf.argmax(prob, axis=1)
            pred = tf.cast(pred, dtype=tf.int32)

            correct = tf.cast(tf.equal(pred, y), dtype=tf.int32)
            correct = tf.reduce_sum(correct)

            total_num += x.shape[0]
            total_correct += int(correct)

        acc = total_correct / total_num
        print(epoch, 'acc:', acc)



if __name__ == '__main__':
    main()