from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Activation, Flatten
from keras.layers.convolutional import Convolution2D, MaxPooling2D
from keras.optimizers import SGD  def Net_model(nb_classes, lr=0.001,decay=1e-6,momentum=0.9):  model = Sequential()  model.add(Convolution2D(filters=10, kernel_size=(5,5),padding='valid',  input_shape=(200, 200, 3)))  model.add(Activation('tanh'))  model.add(MaxPooling2D(pool_size=(2, 2)))  model.add(Convolution2D(filters=20, kernel_size=(10,10)))model.add(Activation('tanh'))  model.add(MaxPooling2D(pool_size=(2, 2)))  model.add(Dropout(0.25))  model.add(Flatten())  model.add(Dense(1000))model.add(Activation('tanh'))  model.add(Dropout(0.5))  model.add(Dense(nb_classes))  model.add(Activation('softmax'))  sgd = SGD(lr=lr, decay=decay, momentum=momentum, nesterov=True)  model.compile(loss='categorical_crossentropy', optimizer=sgd)  return model

import cv2
import os
import numpy as np
import kerasdef loadImages():imageList=[]labelList=[]rootdir="d:\\books\\0"list =os.listdir(rootdir)for item in list:path=os.path.join(rootdir,item)if(os.path.isfile(path)):f=cv2.imread(path)f=cv2.resize(f, (200, 200))#resize到网络input的shapeimageList.append(f)labelList.append(0)#类别0rootdir="d:\\books\\1"list =os.listdir(rootdir)for item in list:path=os.path.join(rootdir,item)if(os.path.isfile(path)):f=cv2.imread(path)f=cv2.resize(f, (200, 200))#resize到网络input的shapeimageList.append(f)labelList.append(1)#类别1return np.asarray(imageList), keras.utils.to_categorical(labelList, 2)

nb_classes = 2
nb_epoch = 30
nb_step = 6
batch_size = 3x,y=loadImages()from keras.preprocessing.image import ImageDataGenerator
dataGenerator=ImageDataGenerator()
dataGenerator.fit(x)
data_generator=dataGenerator.flow(x, y, batch_size, True)#generator函数，用来生成批处理数据（从loadImages中）model=NetModule.Net_model(nb_classes=nb_classes, lr=0.0001) #加载网络模型history=model.fit_generator(data_generator, epochs=nb_epoch, steps_per_epoch=nb_step, shuffle=True)#训练网络，并且返回每次epoch的损失valuemodel.save_weights('D:\\Documents\\Visual Studio 2017\\Projects\\ConsoleApp9\\PythonApplication1\\书本识别\\trained_model_weights.h5')#保存权重
print("DONE, model saved in path-->D:\\Documents\\Visual Studio 2017\\Projects\\ConsoleApp9\\PythonApplication1\\书本识别\\trained_model_weights.h5")

import matplotlib.pyplot as pltplt.plot(history.history['loss'])
plt.show()

def loadImages():imageList=[]rootdir="d:\\books\\validate"list =os.listdir(rootdir)for item in list:path=os.path.join(rootdir,item)if(os.path.isfile(path)):f=cv2.imread(path)f=cv2.resize(f, (200, 200))imageList.append(f)return np.asarray(imageList)x=loadImages()x=np.asarray(x)model=NetModule.Net_model(nb_classes=2, lr=0.0001)
model.load_weights('D:\\Documents\\Visual Studio 2017\\Projects\\ConsoleApp9\\PythonApplication1\\书本识别\\trained_model_weights.h5')print(model.predict(x))
print(model.predict_classes(x))
y=convert2label(model.predict_classes(x))
print(y)

def convert2label(vector):string_array=[]for v in vector:if v==1:string_array.append('BOOK')else:string_array.append('NOT BOOK')return string_array

for i in range(len(x)):cv2.putText(x[i], y[i], (50,50), cv2.FONT_HERSHEY_SIMPLEX, 1, 255, 2)cv2.imshow('image'+str(i), x[i])cv2.waitKey(-1)