版权声明：转载请说明来源，谢谢 https://blog.csdn.net/wsp_1138886114/article/details/83055982
文章目录
运行环境：
Anaconda | python3.7 --pycharm
使用Anaconda 集成环境
pip install tensorflow , pip install keras
keras的核心数据结构就是模型。模型是用来组织网络层的方式。
有Sequential 和 Model 两种模型。
在 ~/keras/keras.json 下修改最后一行backend对应的值即可。
{
"image_dim_ordering": "tf",
"epsilon": 1e-07,
"floatx": "float32",
"backend": "tensorflow"/"theano" //二选一
}
两种模型传参顺序:
1. Sequential 模型:(conv_dim1,channels,conv_dim2,conv_dim3)
X_train = X_train.reshape(X_train.shape[0],1,img_rows,img_cols)
input_shape = (1,img_rows,img_cols)
2.model 模型 :(conv_dim1,conv_dim2,conv_dim3，channels)
X_train = X_train.reshape(X_train.shape[0],img_rows,img_cols,1)
input_shape = (img_rows,img_cols,1)
一、获取图片数据
1.1 读取图片文件 read_data.py
import os
import cv2
import numpy as np
def read_file(path):
img_list = []
label_list = []
dir_counter = 0
#对路径下的所有子文件夹中的所有jpg文件进行读取并存入到一个list中
for child_dir in os.listdir(path):
child_path = path+"/"+child_dir
for dir_image in os.listdir(child_path):
print(child_path+"/"+dir_image)
Suffix_name = ['.png','.jpg','.jpeg']
if dir_image.endswith(tuple(Suffix_name)):
img = cv2.imread(child_path+"/"+dir_image)
img =cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
img_list.append(img)
label_list.append(dir_counter)
dir_counter += 1
img_list = np.array(img_list)
return img_list,label_list,dir_counter
#读取训练数据集的文件夹，把他们的名字返回给一个list
def read_name_list(path):
name_list = []
for child_dir in os.listdir(path):
name_list.append(child_dir)
return name_list
if __name__ == '__main__':
img_list,label_lsit,counter = read_file('./face_reg/dataset_faces')
print (counter)
1.2 数据集整理 dataset.py
import random
import sys
from sklearn.model_selection import train_test_split
class DataSet(object):
def __init__(self,path):
self.num_classes = None
self.X_train = None
self.X_test = None
self.Y_train = None
self.Y_test = None
self.extract_data(path)
#在这个类初始化的过程中读取path下的训练数据
def extract_data(self,path):
imgs,labels,counter = read_file(path) #根据指定路径读取出图片、标签和类别数
print("输出标记")
print(labels)
#拆分数据集
X_train,X_test,y_train,y_test = train_test_split(imgs,labels,test_size=0.4,random_state=random.randint(0, 100))
print("输出训练标记:{0} 训练集长度: {1}:".format(y_train,len(X_train)))
print("输出测试集标记和长度".format(y_test,len(X_test)))
print("输出和:" ,counter)
#重新格式化和标准化 #本案例是基于thano的，如果基于tensorflow的backend需要进行修改
X_train = X_train.reshape(X_train.shape[0], 174, 212, 1) # X_train.shape[0] 训练集图片数量
X_test = X_test.reshape(X_test.shape[0], 174, 212,1)
X_train = X_train.astype('float32')/255
X_test = X_test.astype('float32')/255
print(X_train[1])
#将labels转成 binary class matrices
Y_train = np_utils.to_categorical(y_train, num_classes=counter)
Y_test = np_utils.to_categorical(y_test, num_classes=counter)
print(Y_train)
-----------------------多分类-start----------------------------
def Label_one_hot(labels_dense, num_classes):
num_labels = labels_dense.shape[0] # label数量
index_offset = np.arange(num_labels) * num_classes
labels_one_hot = np.zeros((num_labels, num_classes))
labels_one_hot.flat[index_offset + labels_dense.ravel()] = 1
return labels_one_hot
Y_train = Label_one_hot(y_train, num_classes=counter) #标签类别：counter
Y_test = Label_one_hot(y_test, num_classes=counter)
-----------------------多分类-end----------------------------
#将格式化后的数据赋值给类的属性上
self.X_train = X_train
self.X_test = X_test
self.Y_train = Y_train
self.Y_test = Y_test
self.num_classes = counter
二、训练模型 train_model.py
#建立一个基于CNN的识别模型 class Model(object):
class Model(object):
FILE_PATH = "./model.h5" #模型进行存储和读取的地方
def __init__(self):
self.model = None
def read_trainData(self,dataset): #读取实例化后的DataSet类作为进行训练的数据源
self.dataset = dataset
"""
CNN模型:
卷积 -> 批归一化 -> 激活 -> 池化
卷积 -> 批归一化 -> 激活 -> 池化 -> dropout
卷积 -> 批归一化 -> 激活 -> 池化 -> dropout
展平 -> 全连接 -> 批归一化 -> 激活 -> dropout
-> 全连接 -> 批归一化 -> 激活 -> dropout
-> 全连接 -> 批归一化 -> 激活 -> summary
"""
def build_model(self):
self.model = Sequential()
self.model.add(
Convolution2D(
filters=32,
kernel_size=(5, 5),
padding='same',
dim_ordering='tf',
input_shape=self.dataset.X_train.shape[1:],
)
)
self.model.add( BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(
MaxPooling2D(
pool_size=(2, 2),
strides=(2, 2),
padding='same'
)
)
self.model.add(Convolution2D(filters=64, kernel_size=(5, 5), padding='same'))
self.model.add(BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2), padding='same'))
self.model.add(Dropout(0.15))
self.model.add(Convolution2D(filters=64, kernel_size=(5, 5), padding='same'))
self.model.add(BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2), padding='same'))
self.model.add(Dropout(0.15))
self.model.add(Flatten())
self.model.add(Dense(512))
self.model.add(BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(Dropout(0.5))
self.model.add(Dense(128))
self.model.add(BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(Dropout(0.5))
self.model.add(Dense(self.dataset.num_classes))
self.model.add(BatchNormalization())
self.model.add(Activation('softmax'))
self.model.summary()
plot_model(model,to_file='./model.png')
"""
进行模型训练的函数
optimizer 可选: RMSprop,Adagrad，adadelta
loss：可选：categorical_crossentropy squared_hinge
"""
def train_model(self):
self.model.compile(optimizer='adadelta',loss='squared_hinge', metrics=['accuracy'])
self.model.fit(self.dataset.X_train,self.dataset.Y_train,epochs=12,batch_size=20,callbacks=[TensorBoard(log_dir='./tmp/log')])
def evaluate_model(self):
print('\nTesting---------------')
loss, accuracy = self.model.evaluate(self.dataset.X_test, self.dataset.Y_test)
print('test loss;', loss)
print('test accuracy:', accuracy)
def save(self, file_path=FILE_PATH):
print('Model Saved.')
self.model.save(file_path)
def load(self, file_path=FILE_PATH):
print('Model Loaded.')
self.model = load_model(file_path)
def predict(self,img):
img = img.reshape((1, 174, 212,1)) #1：灰度图
img = img.astype('float32')
img = img/255.0
result = self.model.predict_proba(img) #测算一下该img属于某个label的概率
max_index = np.argmax(result) #找出概率最高的
return max_index,result[0][max_index] #第一个参数为概率最高的label的index,第二个参数为对应概率
if __name__ == '__main__':
datast = DataSet('./pictures/train') # 获取数据集
model = Model() # 调用keras模型
model.read_trainData(datast) # 读取训练数据
model.build_model() # 创建训练网络
model.train_model() # 训练模型
model.evaluate_model() # 验证模型
model.save() # 保存模型
三、模型测试 test_model.py
from read_data import read_file
from train_model import Model
import cv2
import os
import numpy as np
from dataSet import DataSet
from keras import backend as K
from keras.utils import np_utils
K.clear_session() #清空上一次会话
#读取训练数据集的文件夹，把他们的名字返回给一个list
def read_name_list(path):
name_list = []
for child_dir in os.listdir(path):
name_list.append(child_dir)
return name_list
#读取文件夹下子文件夹中所有图片进行识别
def test_onBatch(path): #传入需要分类图片的路径
model= Model()
model.load()
index = 0
img_list, label_list, counter = read_file(path)
for img in img_list:
picType,prob = model.predict(img)
if picType != -1:
index += 1
name_list = read_name_list('./pictures/test')
print(name_list)
print (name_list[picType])
cv2.imwrite("img_"+index+".png",img) #对分类的图像保存
else:
print (" error")
return index
if __name__ == '__main__':
pro_predict=test_onBatch('./pictures/test/') #传入需要分类图片的路径
特别鸣谢：
https://blog.csdn.net/u011268787/article/details/79891284