六、项目实战---识别猫和狗

kagglecatsanddogs网上一搜一大堆，这里我就不上传了，需要的话可以私信

导包

import os
import zipfile
import random
import shutil
import tensorflow as tf
from tensorflow.keras.optimizers import RMSprop
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from shutil import copyfile

猫和狗的照片各12500张

print(len(os.listdir('./temp/cats/')))
print(len(os.listdir('./temp/dogs/')))
"""
12500
12500
"""

生成训练数据文件夹和测试数据文件夹

import os
import zipfile
import random
import shutil
import tensorflow as tf
from tensorflow.keras.optimizers import RMSprop
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from shutil import copyfile


def create_dir(file_dir):
    if os.path.exists(file_dir):
        print("True")
        shutil.rmtree(file_dir)#删除再创建
        os.makedirs(file_dir)
    else:
        os.makedirs(file_dir)


cat_source_dir = "./temp/cats/"
train_cats_dir = "./temp/train/cats/"
test_cats_dir = "./temp/test/cats/"

dot_source_dir = "./temp/dogs/"
train_dogs_dir = "./temp/train/dogs/"
test_dogs_dir = "./temp/test/dogs/"


create_dir(train_cats_dir)#创建猫的训练集文件夹
create_dir(test_cats_dir)#创建猫的测试集文件夹
create_dir(train_dogs_dir)#创建狗的训练集文件夹
create_dir(test_dogs_dir)#创建狗的测试集文件夹

"""
True
True
True
True
"""

将总的猫狗图像按9：1分成训练集和测试集，猫和狗各12500张
最终temp/train/cats和temp/train/dogs两个文件夹下各12500 * 0.9=11250张
temp/test/cats和temp/test/dogs这两个文件夹下各12500 * 0.1=1250张
cats和dogs为总共的猫狗图像
test和train为准备的数据集文件

import os
import zipfile
import random
import shutil
import tensorflow as tf
from tensorflow.keras.optimizers import RMSprop
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from shutil import copyfile


def split_data(source,train,test,split_size):
    files = []
    for filename in os.listdir(source):
        file = source + filename
        if os.path.getsize(file)>0:
            files.append(filename)
        else:
            print(filename + "is zero file,please ignoring")
            
        train_length = int(len(files)*split_size)
        test_length = int(len(files)-train_length)
        shuffled_set = random.sample(files,len(files))
        train_set = shuffled_set[0:train_length]
        test_set = shuffled_set[-test_length:]
        
        for filename in train_set:
            this_file = source + filename
            destination = train + filename
            copyfile(this_file,destination)
            
        for filename in test_set:
            this_file = source + filename
            destination = test + filename
            copyfile(this_file,destination)


cat_source_dir = "./temp/cats/"
train_cats_dir = "./temp/train/cats/"
test_cats_dir = "./temp/test/cats/"

dot_source_dir = "./temp/dogs/"
train_dogs_dir = "./temp/train/dogs/"
test_dogs_dir = "./temp/test/dogs/"


split_size = 0.9
split_data(cat_source_dir,train_cats_dir,test_cats_dir,split_size)
split_data(dog_source_dir,train_dogs_dir,test_dogs_dir,split_size)

先对数据进行归一化操作，预处理进行优化一下

import os
import zipfile
import random
import shutil
import tensorflow as tf
from tensorflow.keras.optimizers import RMSprop
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from shutil import copyfile


train_dir = "./temp/train/"
train_datagen = ImageDataGenerator(rescale=1.0/255.0)#优化网络，先进行归一化操作
train_generator = train_datagen.flow_from_directory(train_dir,batch_size=100,class_mode='binary',target_size=(150,150))#二分类，训练样本的输入的要一致

validation_dir = "./temp/test/"
validation_datagen = ImageDataGenerator(rescale=1.0/255.0)
validation_generator = validation_datagen.flow_from_directory(validation_dir,batch_size=100,class_mode='binary',target_size=(150,150))
"""
Found 22500 images belonging to 2 classes.
Found 2500 images belonging to 2 classes.
"""

搭建模型架构

model = tf.keras.models.Sequential([
    tf.keras.layers.Conv2D(16,(3,3),activation='relu',input_shape=(150,150,3)),
    tf.keras.layers.MaxPooling2D(2,2),
    tf.keras.layers.Conv2D(32,(3,3),activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),
    tf.keras.layers.Conv2D(64,(3,3),activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),

    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(512,activation='relu'),
    tf.keras.layers.Dense(1,activation='sigmoid')
])
model.compile(optimizer=RMSprop(lr=0.001),loss='binary_crossentropy',metrics=['acc'])

训练模型
225：因为数据一共22500张，猫和狗各12500张，其对于训练集个11250张，故训练集共22500张，在预处理第一段代码中，batch_size=100设置了一批100个，故总共应该有225批
epochs=2：两轮，也就是所有的样本全部训练一次
每轮包含225批，每一批有100张样本

history = model.fit_generator(train_generator,
                             epochs=2,#进行2轮训练，每轮255批
                             verbose=1,#要不记录每次训练的日志，1表示记录
                             validation_data=validation_generator)
                             
"""
Instructions for updating:
Use tf.cast instead.
Epoch 1/2
131/225 [================>.............] - ETA: 2:03 - loss: 0.7204 - acc: 0.6093
"""

history是模型运行过程的结果

import matplotlib.image as mpimg
import matplotlib.pyplot as plt

acc = history.history['acc']
val_acc = history.history['val_acc']
loss = history.history['loss']
val_loss = history.history['val_loss']

epochs = range(len(acc))

epoch太少了，导致是直线，多训练几轮实际应该是折线图
准确率

plt.plot(epochs,acc,'r',"training accuracy")
plt.plot(epochs,val_acc,'b',"validation accuracy")
plt.title("training and validation accuracy")
plt.figure()

损失值

plt.plot(epochs,loss,'r',"training loss")
plt.plot(epochs,val_loss,'b',"validation loss")
plt.figure()

import numpy as np
from google.colab import files
from tensorflow.keras.preprocessing import image

uploaded = files.upload()
for fn in uploaded.keys():
    path = 'G:/Juptyer_workspace/Tensorflow_mooc/sjj/test/' + fn#该路径为要用模型测试的路径
    img = image.load_img(path,target_size=(150,150))
    x = image.img_to_array(img)#多维数组
    x = np.expand_dims(x,axis=0)#拉伸
    
    images = np.vstack([x])#水平方向拉直
    classes = model.predict(images,batch_size=10)
    print(classes[0])
    if classes[0]>0.5:
        print(fn + "it is a dog")
    else:
        print(fn + "it is a cat")