Tensorflow&CNN：裂纹分类

版权声明：本文为博主原创文章，转载 请注明出处：https://blog.csdn.net/sc2079/article/details/90478551

- 写在前面

---

　　本科毕业设计终于告一段落了。特写博客记录做毕业设计（路面裂纹识别）期间的踩过的坑和收获。希望对你有用。


　　目前有：


　　　　1.Tensorflow&CNN：裂纹分类


　　　　2.Tensorflow&CNN：验证集预测与模型评价


　　　　3.PyQt5多个GUI界面设计

​
　　本篇讲CNN的训练与预测（以裂纹分类为例）。任务目标：将裂纹图片数据集自动分类：纵向裂纹、横向裂纹、块状裂纹、龟裂裂纹、无裂纹共五类。

​ ​
　　本篇主要参照博客tensorflow: 花卉分类。

- 环境配置安装

---

​ ​
　　运行环境：Python3.6、Spyder

​ ​
　　依赖模块：Skimage、Tensorflow（CPU）、Numpy 、Matlpotlib、Cv2等

- 开始工作

---

1.CNN架构

​ ​
　　所使用的CNN架构如下：

​ ​
　　一共有十三层。

2.训练

​ ​
　　所使用的训练代码如下：

from skimage import io,transform
import glob
import os
import tensorflow as tf
import numpy as np
import time
import matplotlib.pyplot as plt
import pandas as pd

start_time = time.time()
tf.reset_default_graph()   #清除过往tensorflow数据记录
#训练图片集地址
path='..//img5//'

#将所有的图片resize成100*100
w=100
h=100
c=3
#归一化
def normlization(img):
    X=img.copy()
    X1= np.mean(X, axis = 0) # 减去均值，使得以0为中心
    X2=X-X1
    X3= np.std(X2, axis = 0) # 归一化
    X4=X2/X3
    return X4

#读取图片
def read_img(path):
    cate=[path+x for x in os.listdir(path)]
    imgs=[]
    labels=[]
    for idx,folder in enumerate(cate):
        for im in glob.glob(folder+'/*.jpg'):
            #print('reading the images:%s'%(im))
            img=io.imread(im)
            img=transform.resize(img,(w,h))
            #img=normlization(img)
            imgs.append(img)
            labels.append(idx)
    return np.asarray(imgs,np.float32),np.asarray(labels,np.int32)
data,label=read_img(path)

#打乱顺序
num_example=data.shape[0]
arr=np.arange(num_example)
np.random.shuffle(arr)
data=data[arr]
label=label[arr]

#将所有数据分为训练集和验证集
ratio=0.8
s=np.int(num_example*ratio)
x_train=data[:s]
y_train=label[:s]
x_val=data[s:]
y_val=label[s:]

#-----------------构建网络----------------------
#占位符
x=tf.placeholder(tf.float32,shape=[None,w,h,c],name='x')
y_=tf.placeholder(tf.int32,shape=[None,],name='y_')

def inference(input_tensor, train, regularizer):
    with tf.variable_scope('layer1-conv1'):
        conv1_weights = tf.get_variable("weight",[5,5,3,32],initializer=tf.truncated_normal_initializer(stddev=0.1))
        conv1_biases = tf.get_variable("bias", [32], initializer=tf.constant_initializer(0.0))
        conv1 = tf.nn.conv2d(input_tensor, conv1_weights, strides=[1, 1, 1, 1], padding='SAME')
        relu1 = tf.nn.relu(tf.nn.bias_add(conv1, conv1_biases))

    with tf.name_scope("layer2-pool1"):
        pool1 = tf.nn.max_pool(relu1, ksize = [1,2,2,1],strides=[1,2,2,1],padding="VALID")

    with tf.variable_scope("layer3-conv2"):
        conv2_weights = tf.get_variable("weight",[5,5,32,64],initializer=tf.truncated_normal_initializer(stddev=0.1))
        conv2_biases = tf.get_variable("bias", [64], initializer=tf.constant_initializer(0.0))
        conv2 = tf.nn.conv2d(pool1, conv2_weights, strides=[1, 1, 1, 1], padding='SAME')
        relu2 = tf.nn.relu(tf.nn.bias_add(conv2, conv2_biases))

    with tf.name_scope("layer4-pool2"):
        pool2 = tf.nn.max_pool(relu2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')

    with tf.variable_scope("layer5-conv3"):
        conv3_weights = tf.get_variable("weight",[3,3,64,128],initializer=tf.truncated_normal_initializer(stddev=0.1))
        conv3_biases = tf.get_variable("bias", [128], initializer=tf.constant_initializer(0.0))
        conv3 = tf.nn.conv2d(pool2, conv3_weights, strides=[1, 1, 1, 1], padding='SAME')
        relu3 = tf.nn.relu(tf.nn.bias_add(conv3, conv3_biases))

    with tf.name_scope("layer6-pool3"):
        pool3 = tf.nn.max_pool(relu3, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')

    with tf.variable_scope("layer7-conv4"):
        conv4_weights = tf.get_variable("weight",[3,3,128,128],initializer=tf.truncated_normal_initializer(stddev=0.1))
        conv4_biases = tf.get_variable("bias", [128], initializer=tf.constant_initializer(0.0))
        conv4 = tf.nn.conv2d(pool3, conv4_weights, strides=[1, 1, 1, 1], padding='SAME')
        relu4 = tf.nn.relu(tf.nn.bias_add(conv4, conv4_biases))

    with tf.name_scope("layer8-pool4"):
        pool4 = tf.nn.max_pool(relu4, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
        nodes = 6*6*128
        reshaped = tf.reshape(pool4,[-1,nodes])

    with tf.variable_scope('layer9-fc1'):
        fc1_weights = tf.get_variable("weight", [nodes, 1024],
                                      initializer=tf.truncated_normal_initializer(stddev=0.1))
        if regularizer != None: tf.add_to_collection('losses', regularizer(fc1_weights))
        fc1_biases = tf.get_variable("bias", [1024], initializer=tf.constant_initializer(0.1))

        fc1 = tf.nn.relu(tf.matmul(reshaped, fc1_weights) + fc1_biases)
        if train: fc1 = tf.nn.dropout(fc1, 0.5)

    with tf.variable_scope('layer10-fc2'):
        fc2_weights = tf.get_variable("weight", [1024, 512],
                                      initializer=tf.truncated_normal_initializer(stddev=0.1))
        if regularizer != None: tf.add_to_collection('losses', regularizer(fc2_weights))
        fc2_biases = tf.get_variable("bias", [512], initializer=tf.constant_initializer(0.1))

        fc2 = tf.nn.relu(tf.matmul(fc1, fc2_weights) + fc2_biases)
        if train: fc2 = tf.nn.dropout(fc2, 0.5)

    with tf.variable_scope('layer11-fc3'):
        fc3_weights = tf.get_variable("weight", [512, 5],
                                      initializer=tf.truncated_normal_initializer(stddev=0.1))
        if regularizer != None: tf.add_to_collection('losses', regularizer(fc3_weights))
        fc3_biases = tf.get_variable("bias", [5], initializer=tf.constant_initializer(0.1))
        logit = tf.matmul(fc2, fc3_weights) + fc3_biases

    return logit

#训练参数
n_epoch=14
batch_size=32
batch_size2=32
learning_rate=0.001

#---------------------------网络结束---------------------------
regularizer = tf.contrib.layers.l2_regularizer(0.0001)
logits = inference(x,False,regularizer)

#(小处理)将logits乘以1赋值给logits_eval，定义name，方便在后续调用模型时通过tensor名字调用输出tensor
b = tf.constant(value=1,dtype=tf.float32)
logits_eval = tf.multiply(logits,b,name='logits_eval') 

# 利用交叉熵定义损失
loss=tf.nn.sparse_softmax_cross_entropy_with_logits(logits=logits, labels=y_)
mean_loss = tf.reduce_mean(loss)  # 平均损失
train_op=tf.train.AdamOptimizer(learning_rate).minimize(loss)
correct_prediction = tf.equal(tf.cast(tf.argmax(logits,1),tf.int32), y_)
acc= tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

#定义一个函数，按批次取数据
def minibatches(inputs=None, targets=None, batch_size=None, shuffle=False):
    assert len(inputs) == len(targets)
    if shuffle:
        indices = np.arange(len(inputs))
        np.random.shuffle(indices)
    for start_idx in range(0, len(inputs) - batch_size + 1, batch_size):
        if shuffle:
            excerpt = indices[start_idx:start_idx + batch_size]
        else:
            excerpt = slice(start_idx, start_idx + batch_size)
        yield inputs[excerpt], targets[excerpt]

saver=tf.train.Saver()
sess=tf.Session()
sess.run(tf.global_variables_initializer())
traloss,traacc,valloss,valacc=[],[],[],[]
for epoch in range(n_epoch):
    #training
    train_loss, train_acc, n_batch = [],[], 0
    for x_train_a, y_train_a in minibatches(x_train, y_train, batch_size, shuffle=True):
        _,err,ac=sess.run([train_op,mean_loss,acc], feed_dict={x: x_train_a, y_: y_train_a})
        train_loss.append(err); train_acc.append(ac); n_batch += 1
    tra_loss=round(np.sum(train_loss)/ n_batch,3)
    tra_acc=round(np.sum(train_acc)/ n_batch,3)
    traloss.append(tra_loss)
    traacc.append(tra_acc)
    print("epoch: %d    train loss: %.3f    train acc: %.3f"%(epoch,tra_loss,tra_acc))

    #validation
    validation_loss, validation_acc, n_batch = [], [], 0
    for x_val_a, y_val_a in minibatches(x_val, y_val, batch_size2, shuffle=False):
        err, ac = sess.run([mean_loss,acc], feed_dict={x: x_val_a, y_: y_val_a})
        validation_loss.append(err); validation_acc.append(ac); n_batch += 1
    val_loss=round(np.sum(validation_loss)/ n_batch,3)
    val_acc=round(np.sum(validation_acc)/ n_batch,3)
    valloss.append(val_loss)
    valacc.append(val_acc)
    print("epoch: %d    validation loss: %.3f    validation acc: %.3f"%(epoch,val_loss,val_acc))

end_time = time.time()
print("   train loss: %f" %tra_loss)
print("   train acc: %f" %tra_acc)
print("   validation loss: %f" %val_loss)
print("   validation acc: %f" %val_acc)
print("   consume: %f s" %(end_time-start_time))
timeArray = time.localtime(end_time)
now=time.strftime("%Y_%m_%d", timeArray)  #时间
saver.save(sess,".//model//model-" + str(epoch)+'-'+now)
sess.close()

3.训练记录存储

​ ​
　　训练过程存储，主要是训练批次、训练集损失率、训练集准确率、验证集损失率、验证集准确率。

​ ​
　　将其存入csv方法如下：

#字典中的key值即为csv中列名
dataframe = pd.DataFrame({'traloss':traloss,'traacc':traacc,'valloss':valloss,'valacc':valacc})

#将DataFrame存储为csv,index表示是否显示行名，default=True
dataframe.to_csv("test.csv",index=['traloss','traacc','valloss','valacc'],sep=',')

​ ​
　　另外为了记录每次训练更详细信息，便于选择最合适的训练参数，需要将训练时的参数一并加以保存。使用txt方法保存：

#数据记录
with open('log.txt','a+')as file:
    file.write('\n'+now+'\n')
    file.write('n_epoch:'+str(n_epoch)+'  '+
           'batch_size:'+str(batch_size)+'  '+
           'batch_size2:'+str(batch_size2)+'  '+
           'learning_rate:'+str(learning_rate)+'\n')
    for i in range(len(traloss)):
        file.write(str(traloss[i])+','+
               str(traacc[i])+','+
               str(valloss[i])+','+
               str(valacc[i])+'\n')

4.绘制训练集和验证集的损失准确率曲线

def map_loss_acc(_type,loss,acc):
    plt.figure()
    fig, ax1 = plt.subplots()
    ax2 = ax1.twinx()
    lns1 = ax1.plot(np.arange(n_epoch), loss, label="Loss")
    lns2 = ax2.plot(np.arange(n_epoch), acc, 'r', label="Accuracy")
    ax1.set_xlabel('epoch')
    ax1.set_ylabel(_type+'loss')
    ax2.set_ylabel(_type+'accuracy')
    # 合并图例
    lns = lns1 + lns2
    labels = ["Loss", "Accuracy"]
    plt.legend(lns, labels, loc=7)

​ ​
　　直接调用即可。

map_loss_acc('training',traloss,traacc)
map_loss_acc('validation',valloss,valacc)

- 结果展示

---

​ ​
　　可以看出验证集的准确率达到了92.1%，对于在数据集不足、计算力有限的情况下还是挺不错的。