【深度學(xué)習(xí)】使用transformer進(jìn)行圖像分類

import osimport mathimport numpy as npimport pickle as pimport tensorflow as tffrom tensorflow import kerasimport matplotlib.pyplot as pltfrom tensorflow.keras import layersimport tensorflow_addons as tfa%matplotlib inline

def load_CIFAR_data(data_dir):    """load CIFAR data"""     images_train=[]    labels_train=[]    for i in range(5):        f=os.path.join(data_dir,'data_batch_%d' % (i+1))        print('loading ',f)        # 調(diào)用 load_CIFAR_batch( )獲得批量的圖像及其對(duì)應(yīng)的標(biāo)簽        image_batch,label_batch=load_CIFAR_batch(f)        images_train.append(image_batch)        labels_train.append(label_batch)        Xtrain=np.concatenate(images_train)        Ytrain=np.concatenate(labels_train)        del image_batch ,label_batch        Xtest,Ytest=load_CIFAR_batch(os.path.join(data_dir,'test_batch'))    print('finished loadding CIFAR-10 data')        # 返回訓(xùn)練集的圖像和標(biāo)簽，測(cè)試集的圖像和標(biāo)簽return (Xtrain,Ytrain),(Xtest,Ytest)

def load_CIFAR_batch(filename):    """ load single batch of cifar """      with open(filename, 'rb')as f:        # 一個(gè)樣本由標(biāo)簽和圖像數(shù)據(jù)組成        #  (3072=32x32x3)        # ...        #         data_dict = p.load(f, encoding='bytes')        images= data_dict[b'data']        labels = data_dict[b'labels']                        # 把原始數(shù)據(jù)結(jié)構(gòu)調(diào)整為: BCWH        images = images.reshape(10000, 3, 32, 32)        # tensorflow處理圖像數(shù)據(jù)的結(jié)構(gòu)：BWHC        # 把通道數(shù)據(jù)C移動(dòng)到最后一個(gè)維度        images = images.transpose (0,2,3,1)             labels = np.array(labels)                return images, labels

data_dir = r'C:\Users\wumg\jupyter-ipynb\data\cifar-10-batches-py'(x_train,y_train),(x_test,y_test) = load_CIFAR_data(data_dir)

train_dataset = tf.data.Dataset.from_tensor_slices((x_train, y_train))test_dataset = tf.data.Dataset.from_tensor_slices((x_test, y_test))

num_classes = 10input_shape = (32, 32, 3)  learning_rate = 0.001weight_decay = 0.0001batch_size = 256num_epochs = 10image_size = 72  # We'll resize input images to this sizepatch_size = 6  # Size of the patches to be extract from the input imagesnum_patches = (image_size // patch_size) ** 2projection_dim = 64num_heads = 4transformer_units = [    projection_dim * 2,    projection_dim,]  # Size of the transformer layerstransformer_layers = 8mlp_head_units = [2048, 1024]  # Size of the dense layers of the final classifier

data_augmentation = keras.Sequential(    [        layers.experimental.preprocessing.Normalization(),        layers.experimental.preprocessing.Resizing(image_size, image_size),        layers.experimental.preprocessing.RandomFlip("horizontal"),        layers.experimental.preprocessing.RandomRotation(factor=0.02),        layers.experimental.preprocessing.RandomZoom(            height_factor=0.2, width_factor=0.2        ),    ],    name="data_augmentation",)# 使預(yù)處理層的狀態(tài)與正在傳遞的數(shù)據(jù)相匹配#Compute the mean and the variance of the training data for normalization.data_augmentation.layers[0].adapt(x_train)

def mlp(x, hidden_units, dropout_rate):    for units in hidden_units:        x = layers.Dense(units, activation=tf.nn.gelu)(x)        x = layers.Dropout(dropout_rate)(x)    return x

class Patches(layers.Layer):    def __init__(self, patch_size):        super(Patches, self).__init__()        self.patch_size = patch_size     def call(self, images):        batch_size = tf.shape(images)[0]        patches = tf.image.extract_patches(            images=images,            sizes=[1, self.patch_size, self.patch_size, 1],            strides=[1, self.patch_size, self.patch_size, 1],            rates=[1, 1, 1, 1],            padding="VALID",        )        patch_dims = patches.shape[-1]        patches = tf.reshape(patches, [batch_size, -1, patch_dims])        return patches

import matplotlib.pyplot as plt plt.figure(figsize=(4, 4))image = x_train[np.random.choice(range(x_train.shape[0]))]plt.imshow(image.astype("uint8"))plt.axis("off") resized_image = tf.image.resize(    tf.convert_to_tensor([image]), size=(image_size, image_size))patches = Patches(patch_size)(resized_image)print(f"Image size: {image_size} X {image_size}")print(f"Patch size: {patch_size} X {patch_size}")print(f"Patches per image: {patches.shape[1]}")print(f"Elements per patch: {patches.shape[-1]}") n = int(np.sqrt(patches.shape[1]))plt.figure(figsize=(4, 4))for i, patch in enumerate(patches[0]):    ax = plt.subplot(n, n, i + 1)    patch_img = tf.reshape(patch, (patch_size, patch_size, 3))    plt.imshow(patch_img.numpy().astype("uint8"))    plt.axis("off")

class PatchEncoder(layers.Layer):    def __init__(self, num_patches, projection_dim):        super(PatchEncoder, self).__init__()        self.num_patches = num_patches        #一個(gè)全連接層，其輸出維度為projection_dim，沒(méi)有指明激活函數(shù)        self.projection = layers.Dense(units=projection_dim)        #定義一個(gè)嵌入層，這是一個(gè)可學(xué)習(xí)的層        #輸入維度為num_patches，輸出維度為projection_dim        self.position_embedding = layers.Embedding(            input_dim=num_patches, output_dim=projection_dim        )     def call(self, patch):        positions = tf.range(start=0, limit=self.num_patches, delta=1)        encoded = self.projection(patch) + self.position_embedding(positions)        return encoded

def create_vit_classifier():    inputs = layers.Input(shape=input_shape)    # Augment data.    augmented = data_augmentation(inputs)    #augmented = augmented_train_batches(inputs)        # Create patches.    patches = Patches(patch_size)(augmented)    # Encode patches.    encoded_patches = PatchEncoder(num_patches, projection_dim)(patches)     # Create multiple layers of the Transformer block.    for _ in range(transformer_layers):        # Layer normalization 1.        x1 = layers.LayerNormalization(epsilon=1e-6)(encoded_patches)        # Create a multi-head attention layer.        attention_output = layers.MultiHeadAttention(            num_heads=num_heads, key_dim=projection_dim, dropout=0.1        )(x1, x1)        # Skip connection 1.        x2 = layers.Add()([attention_output, encoded_patches])        # Layer normalization 2.        x3 = layers.LayerNormalization(epsilon=1e-6)(x2)        # MLP.        x3 = mlp(x3, hidden_units=transformer_units, dropout_rate=0.1)        # Skip connection 2.        encoded_patches = layers.Add()([x3, x2])     # Create a [batch_size, projection_dim] tensor.    representation = layers.LayerNormalization(epsilon=1e-6)(encoded_patches)    representation = layers.Flatten()(representation)    representation = layers.Dropout(0.5)(representation)    # Add MLP.    features = mlp(representation, hidden_units=mlp_head_units, dropout_rate=0.5)    # Classify outputs.    logits = layers.Dense(num_classes)(features)    # Create the Keras model.    model = keras.Model(inputs=inputs, outputs=logits)return model

def run_experiment(model):    optimizer = tfa.optimizers.AdamW(        learning_rate=learning_rate, weight_decay=weight_decay    )     model.compile(        optimizer=optimizer,        loss=keras.losses.SparseCategoricalCrossentropy(from_logits=True),        metrics=[            keras.metrics.SparseCategoricalAccuracy(name="accuracy"),            keras.metrics.SparseTopKCategoricalAccuracy(5, name="top-5-accuracy"),        ],    )     #checkpoint_filepath = r".\tmp\checkpoint"    checkpoint_filepath ="model_bak.hdf5"    checkpoint_callback = keras.callbacks.ModelCheckpoint(        checkpoint_filepath,        monitor="val_accuracy",        save_best_only=True,        save_weights_only=True,    )     history = model.fit(        x=x_train,        y=y_train,        batch_size=batch_size,        epochs=num_epochs,        validation_split=0.1,        callbacks=[checkpoint_callback],    )     model.load_weights(checkpoint_filepath)    _, accuracy, top_5_accuracy = model.evaluate(x_test, y_test)    print(f"Test accuracy: {round(accuracy * 100, 2)}%")    print(f"Test top 5 accuracy: {round(top_5_accuracy * 100, 2)}%")     return history

vit_classifier = create_vit_classifier()history = run_experiment(vit_classifier)

acc = history.history['accuracy']val_acc = history.history['val_accuracy'] loss = history.history['loss']val_loss =history.history['val_loss'] plt.figure(figsize=(8, 8))plt.subplot(2, 1, 1)plt.plot(acc, label='Training Accuracy')plt.plot(val_acc, label='Validation Accuracy')plt.legend(loc='lower right')plt.ylabel('Accuracy')plt.ylim([min(plt.ylim()),1.1])plt.title('Training and Validation Accuracy') plt.subplot(2, 1, 2)plt.plot(loss, label='Training Loss')plt.plot(val_loss, label='Validation Loss')plt.legend(loc='upper right')plt.ylabel('Cross Entropy')plt.ylim([-0.1,4.0])plt.title('Training and Validation Loss')plt.xlabel('epoch')plt.show()