定义InceptionV3模型用于caltech101物体分类

最新推荐文章于 2022-06-10 19:05:34 发布

青山青

最新推荐文章于 2022-06-10 19:05:34 发布

阅读量552

点赞数

分类专栏：深度视觉

原文链接：https://edu.csdn.net/course/play/29546/423555

版权

深度视觉专栏收录该内容

20 篇文章 1 订阅

订阅专栏

导入函数库

import tensorflow as tf
import numpy as np
import tensorflow_datasets as tfds # 这个是指Tensorflow Datasets
import matplotlib.pyplot as plt

定义网络结构

一些参数设置

layers = tf.keras.layers
models = tf.keras.models

定义InceptionV3模型用于caltech101物体分类

def conv2d_bn(x,
              filters,
              num_row,
              num_col,
              padding='same',
              strides=(1, 1),
              name=None):
    if name is not None:
        bn_name = name + '_bn'
        conv_name = name + '_conv'
    else:
        bn_name = None
        conv_name = None
    bn_axis = 3
    x = layers.Conv2D(
        filters, (num_row, num_col),
        strides=strides,
        padding=padding,
        use_bias=False,
        name=conv_name)(x)
    x = layers.BatchNormalization(axis=bn_axis, scale=False, name=bn_name)(x)
    x = layers.Activation('relu', name=name)(x)
    return x

def InceptionV3(input_shape=(224, 224, 3)): 
    img_input = layers.Input(shape=input_shape)

    channel_axis = 3

    x = conv2d_bn(img_input, 32, 3, 3, strides=(2, 2), padding='valid')
    x = conv2d_bn(x, 32, 3, 3, padding='valid')
    x = conv2d_bn(x, 64, 3, 3)
    x = layers.MaxPooling2D((3, 3), strides=(2, 2))(x)

    x = conv2d_bn(x, 80, 1, 1, padding='valid')
    x = conv2d_bn(x, 192, 3, 3, padding='valid')
    x = layers.MaxPooling2D((3, 3), strides=(2, 2))(x)

    # mixed 0: (None, 25, 25, 256)
    branch1x1 = conv2d_bn(x, 64, 1, 1)

    branch5x5 = conv2d_bn(x, 48, 1, 1)
    branch5x5 = conv2d_bn(branch5x5, 64, 5, 5)

    branch3x3dbl = conv2d_bn(x, 64, 1, 1)
    branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
    branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)

    branch_pool = layers.AveragePooling2D((3, 3),
                                          strides=(1, 1),
                                          padding='same')(x)
    branch_pool = conv2d_bn(branch_pool, 32, 1, 1)
    x = layers.concatenate(
        [branch1x1, branch5x5, branch3x3dbl, branch_pool],
        axis=channel_axis,
        name='mixed0')

    # mixed 1: (None, 25, 25, 288)
    branch1x1 = conv2d_bn(x, 64, 1, 1)

    branch5x5 = conv2d_bn(x, 48, 1, 1)
    branch5x5 = conv2d_bn(branch5x5, 64, 5, 5)

    branch3x3dbl = conv2d_bn(x, 64, 1, 1)
    branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
    branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)

    branch_pool = layers.AveragePooling2D((3, 3),
                                          strides=(1, 1),
                                          padding='same')(x)
    branch_pool = conv2d_bn(branch_pool, 64, 1, 1)
    x = layers.concatenate(
        [branch1x1, branch5x5, branch3x3dbl, branch_pool],
        axis=channel_axis,
        name='mixed1')

    # mixed 2: (None, 25, 25, 288)
    branch1x1 = conv2d_bn(x, 64, 1, 1)

    branch5x5 = conv2d_bn(x, 48, 1, 1)
    branch5x5 = conv2d_bn(branch5x5, 64, 5, 5)

    branch3x3dbl = conv2d_bn(x, 64, 1, 1)
    branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
    branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)

    branch_pool = layers.AveragePooling2D((3, 3),
                                          strides=(1, 1),
                                          padding='same')(x)
    branch_pool = conv2d_bn(branch_pool, 64, 1, 1)
    x = layers.concatenate(
        [branch1x1, branch5x5, branch3x3dbl, branch_pool],
        axis=channel_axis,
        name='mixed2')

    # mixed 3: (None, 12, 12, 768)
    branch3x3 = conv2d_bn(x, 384, 3, 3, strides=(2, 2), padding='valid')

    branch3x3dbl = conv2d_bn(x, 64, 1, 1)
    branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
    branch3x3dbl = conv2d_bn(
        branch3x3dbl, 96, 3, 3, strides=(2, 2), padding='valid')

    branch_pool = layers.MaxPooling2D((3, 3), strides=(2, 2))(x)
    x = layers.concatenate(
        [branch3x3, branch3x3dbl, branch_pool],
        axis=channel_axis,
        name='mixed3')

    # mixed 4: (None, 12, 12, 768)
    branch1x1 = conv2d_bn(x, 192, 1, 1)

    branch7x7 = conv2d_bn(x, 128, 1, 1)
    branch7x7 = conv2d_bn(branch7x7, 128, 1, 7)
    branch7x7 = conv2d_bn(branch7x7, 192, 7, 1)

    branch7x7dbl = conv2d_bn(x, 128, 1, 1)
    branch7x7dbl = conv2d_bn(branch7x7dbl, 128, 7, 1)
    branch7x7dbl = conv2d_bn(branch7x7dbl, 128, 1, 7)
    branch7x7dbl = conv2d_bn(branch7x7dbl, 128, 7, 1)
    branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 1, 7)

    branch_pool = layers.AveragePooling2D((3, 3),
                                          strides=(1, 1),
                                          padding='same')(x)
    branch_pool = conv2d_bn(branch_pool, 192, 1, 1)
    x = layers.concatenate(
        [branch1x1, branch7x7, branch7x7dbl, branch_pool],
        axis=channel_axis,
        name='mixed4')

    # mixed 5, 6: (None, 12, 12, 768)
    for i in range(2):
        branch1x1 = conv2d_bn(x, 192, 1, 1)

        branch7x7 = conv2d_bn(x, 160, 1, 1)
        branch7x7 = conv2d_bn(branch7x7, 160, 1, 7)
        branch7x7 = conv2d_bn(branch7x7, 192, 7, 1)

        branch7x7dbl = conv2d_bn(x, 160, 1, 1)
        branch7x7dbl = conv2d_bn(branch7x7dbl, 160, 7, 1)
        branch7x7dbl = conv2d_bn(branch7x7dbl, 160, 1, 7)
        branch7x7dbl = conv2d_bn(branch7x7dbl, 160, 7, 1)
        branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 1, 7)

        branch_pool = layers.AveragePooling2D(
            (3, 3), strides=(1, 1), padding='same')(x)
        branch_pool = conv2d_bn(branch_pool, 192, 1, 1)
        x = layers.concatenate(
            [branch1x1, branch7x7, branch7x7dbl, branch_pool],
            axis=channel_axis,
            name='mixed' + str(5 + i))

    # mixed 7: (None, 12, 12, 768)
    branch1x1 = conv2d_bn(x, 192, 1, 1)

    branch7x7 = conv2d_bn(x, 192, 1, 1)
    branch7x7 = conv2d_bn(branch7x7, 192, 1, 7)
    branch7x7 = conv2d_bn(branch7x7, 192, 7, 1)

    branch7x7dbl = conv2d_bn(x, 192, 1, 1)
    branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 7, 1)
    branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 1, 7)
    branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 7, 1)
    branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 1, 7)

    branch_pool = layers.AveragePooling2D((3, 3),
                                          strides=(1, 1),
                                          padding='same')(x)
    branch_pool = conv2d_bn(branch_pool, 192, 1, 1)
    x = layers.concatenate(
        [branch1x1, branch7x7, branch7x7dbl, branch_pool],
        axis=channel_axis,
        name='mixed7')

    # mixed 8: (None, 5, 5, 1280)
    branch3x3 = conv2d_bn(x, 192, 1, 1)
    branch3x3 = conv2d_bn(branch3x3, 320, 3, 3,
                          strides=(2, 2), padding='valid')

    branch7x7x3 = conv2d_bn(x, 192, 1, 1)
    branch7x7x3 = conv2d_bn(branch7x7x3, 192, 1, 7)
    branch7x7x3 = conv2d_bn(branch7x7x3, 192, 7, 1)
    branch7x7x3 = conv2d_bn(
        branch7x7x3, 192, 3, 3, strides=(2, 2), padding='valid')

    branch_pool = layers.MaxPooling2D((3, 3), strides=(2, 2))(x)
    x = layers.concatenate(
        [branch3x3, branch7x7x3, branch_pool],
        axis=channel_axis,
        name='mixed8')

    # mixed 9: (None, 5, 5, 2048)
    for i in range(2):
        branch1x1 = conv2d_bn(x, 320, 1, 1)

        branch3x3 = conv2d_bn(x, 384, 1, 1)
        branch3x3_1 = conv2d_bn(branch3x3, 384, 1, 3)
        branch3x3_2 = conv2d_bn(branch3x3, 384, 3, 1)
        branch3x3 = layers.concatenate(
            [branch3x3_1, branch3x3_2],
            axis=channel_axis,
            name='mixed9_' + str(i))

        branch3x3dbl = conv2d_bn(x, 448, 1, 1)
        branch3x3dbl = conv2d_bn(branch3x3dbl, 384, 3, 3)
        branch3x3dbl_1 = conv2d_bn(branch3x3dbl, 384, 1, 3)
        branch3x3dbl_2 = conv2d_bn(branch3x3dbl, 384, 3, 1)
        branch3x3dbl = layers.concatenate(
            [branch3x3dbl_1, branch3x3dbl_2], axis=channel_axis)

        branch_pool = layers.AveragePooling2D(
            (3, 3), strides=(1, 1), padding='same')(x)
        branch_pool = conv2d_bn(branch_pool, 192, 1, 1)
        x = layers.concatenate(
            [branch1x1, branch3x3, branch3x3dbl, branch_pool],
            axis=channel_axis,
            name='mixed' + str(9 + i))
        
    inputs = img_input
    outputs = x
    # Create model.
    model = models.Model(inputs, outputs, name='inception_v3')

    return model

pre_trained_model = InceptionV3()

x = layers.GlobalAveragePooling2D()(pre_trained_model.output)
x = layers.Dense(512, activation='relu', name='fc1')(x)
x = layers.Dense(102, activation='softmax', name='predictions')(x)

model = models.Model(pre_trained_model.input, x)
# Load weights.
weights_path = './models/inceptionv3.h5'
model.load_weights(weights_path)

模型训练日志记录

log_dir = './logs'
tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir=log_dir, histogram_freq=1)

进行数据增强

def convert(image, label):
    image = tf.image.convert_image_dtype(image, tf.float32) # Cast and normalize the image to [0,1]
    image = tf.image.resize_with_crop_or_pad(image, 256, 256)
    image = tf.image.random_crop(image, size=[224, 224, 3]) # Random crop back to 224x224
    return image, label

def augment(image, label):
    image, label = convert(image, label)
    image = tf.image.random_flip_left_right(image)
    image = tf.image.random_flip_up_down(image)
    image = tf.image.random_brightness(image, max_delta=0.3) # Random brightness
    return image, label

数据读取并预处理

# 此处使用tfds的方式构建data pipeline
(raw_test, raw_train), metadata = tfds.load(
    'caltech101', # 数据集名称，这个是caltech101分类数据集，共102个类别(包含background类别)
    split=['test', 'train'], # 这里的raw_test和split的'test'对应，raw_train和split的'train'对应
    with_info=True, # 这个参数和metadata对应
    as_supervised=True, # 这个参数的作用是返回tuple形式的(input, label),举个例子，raw_test=tuple(input, label)
    data_dir='./tensorflow_datasets'
)

BATCH_SIZE = 4
SHUFFLE_BUFFER_SIZE = 3060 # 原始train_num=3060，把整个数据集加载进内存进行shuffle，效果更好

# 可以体验下这里是否加prefetch(tf.data.experimental.AUTOTUNE)和cache()的区别，对训练速度，以及CPU负载有影响
train_batches = raw_train.shuffle(SHUFFLE_BUFFER_SIZE).map(augment).batch(BATCH_SIZE).prefetch(tf.data.experimental.AUTOTUNE)
test_batches = raw_test.map(convert).batch(BATCH_SIZE)

进行模型训练

model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=1e-5),
             loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
             metrics=['accuracy'])

model.fit(
    train_batches,
    epochs=5,
    callbacks=[tensorboard_callback]
)

Epoch 1/5
765/765 [==============================] - 31s 41ms/step - loss: 3.8264 - accuracy: 0.8196
Epoch 2/5
765/765 [==============================] - 33s 44ms/step - loss: 3.8226 - accuracy: 0.8232
Epoch 3/5
765/765 [==============================] - 31s 41ms/step - loss: 3.8229 - accuracy: 0.8225
Epoch 4/5
765/765 [==============================] - 32s 42ms/step - loss: 3.8221 - accuracy: 0.8225
Epoch 5/5
765/765 [==============================] - 33s 43ms/step - loss: 3.8205 - accuracy: 0.8242





<tensorflow.python.keras.callbacks.History at 0x7fe924095080>

模型训练后预测展示

get_label_name = metadata.features['label'].int2str

for image, label in raw_test.take(5):
    image, label = convert(image, label)
    predict = np.argmax(model.predict(np.expand_dims(image, axis=0)))
    print(get_label_name(predict))
    plt.figure()
    plt.imshow(image)
    plt.title(get_label_name(label))

buddha
camera
motorbikes
leopards
menorah

在这里插入图片描述

Baseline的test acc，并保存模型

_, baseline_model_accuracy = model.evaluate(test_batches, verbose=1)
print('Baseline test accuracy: ', baseline_model_accuracy)

keras_file = './test.h5'
tf.keras.models.save_model(model, keras_file, include_optimizer=False)
print('Saved baseline model to: ', keras_file)

1521/1521 [==============================] - 21s 14ms/step - loss: 3.8558 - accuracy: 0.7914
Baseline test accuracy:  0.7914201021194458
Saved baseline model to:  ./test.h5

青山青

关注

0
点赞
踩
2

收藏

觉得还不错? 一键收藏
2
评论
定义InceptionV3模型用于caltech101物体分类

导入函数库import tensorflow as tfimport numpy as npimport tensorflow_datasets as tfds # 这个是指Tensorflow Datasetsimport matplotlib.pyplot as plt定义网络结构一些参数设置layers = tf.keras.layersmodels = tf.keras.models定义InceptionV3模型用于caltech101物体分类def conv2d_bn(x,
复制链接

扫一扫