tensorflow学习（6.Alexnet实现及猫狗分类）

有关Alexnet介绍的部分可以见：https://blog.csdn.net/qq_26499769/article/details/82928164

然后自己结合教程尝试着实现了一下：

from skimage import io,transform
import glob
import numpy as np
import tensorflow as tf
#from alexnet import alexnetdef alexnet(x, keep_prob, num_classes):# conv1with tf.name_scope('conv1') as scope:kernel = tf.Variable(tf.truncated_normal([11, 11, 3, 96], dtype=tf.float32,stddev=1e-1), name='weights')conv = tf.nn.conv2d(x, kernel, [1, 4, 4, 1], padding='SAME')biases = tf.Variable(tf.constant(0.0, shape=[96], dtype=tf.float32),trainable=True, name='biases')bias = tf.nn.bias_add(conv, biases)conv1 = tf.nn.relu(bias, name=scope)# lrn1with tf.name_scope('lrn1') as scope:lrn1 = tf.nn.local_response_normalization(conv1,alpha=1e-4,beta=0.75,depth_radius=2,bias=2.0)# pool1with tf.name_scope('pool1') as scope:pool1 = tf.nn.max_pool(lrn1,ksize=[1, 3, 3, 1],strides=[1, 2, 2, 1],padding='VALID')# conv2with tf.name_scope('conv2') as scope:pool1_groups = tf.split(axis=3, value=pool1, num_or_size_splits=2)kernel = tf.Variable(tf.truncated_normal([5, 5, 48, 256], dtype=tf.float32,stddev=1e-1), name='weights')kernel_groups = tf.split(axis=3, value=kernel, num_or_size_splits=2)conv_up = tf.nn.conv2d(pool1_groups[0], kernel_groups[0], [1, 1, 1, 1], padding='SAME')conv_down = tf.nn.conv2d(pool1_groups[1], kernel_groups[1], [1, 1, 1, 1], padding='SAME')biases = tf.Variable(tf.constant(0.0, shape=[256], dtype=tf.float32),trainable=True, name='biases')biases_groups = tf.split(axis=0, value=biases, num_or_size_splits=2)bias_up = tf.nn.bias_add(conv_up, biases_groups[0])bias_down = tf.nn.bias_add(conv_down, biases_groups[1])bias = tf.concat(axis=3, values=[bias_up, bias_down])conv2 = tf.nn.relu(bias, name=scope)# lrn2with tf.name_scope('lrn2') as scope:lrn2 = tf.nn.local_response_normalization(conv2,alpha=1e-4,beta=0.75,depth_radius=2,bias=2.0)# pool2with tf.name_scope('pool2') as scope:pool2 = tf.nn.max_pool(lrn2,ksize=[1, 3, 3, 1],strides=[1, 2, 2, 1],padding='VALID')# conv3with tf.name_scope('conv3') as scope:kernel = tf.Variable(tf.truncated_normal([3, 3, 256, 384],dtype=tf.float32,stddev=1e-1), name='weights')conv = tf.nn.conv2d(pool2, kernel, [1, 1, 1, 1], padding='SAME')biases = tf.Variable(tf.constant(0.0, shape=[384], dtype=tf.float32),trainable=True, name='biases')bias = tf.nn.bias_add(conv, biases)conv3 = tf.nn.relu(bias, name=scope)# conv4with tf.name_scope('conv4') as scope:conv3_groups = tf.split(axis=3, value=conv3, num_or_size_splits=2)kernel = tf.Variable(tf.truncated_normal([3, 3, 192, 384],dtype=tf.float32,stddev=1e-1), name='weights')kernel_groups = tf.split(axis=3, value=kernel, num_or_size_splits=2)conv_up = tf.nn.conv2d(conv3_groups[0], kernel_groups[0], [1, 1, 1, 1], padding='SAME')conv_down = tf.nn.conv2d(conv3_groups[1], kernel_groups[1], [1, 1, 1, 1], padding='SAME')biases = tf.Variable(tf.constant(0.0, shape=[384], dtype=tf.float32),trainable=True, name='biases')biases_groups = tf.split(axis=0, value=biases, num_or_size_splits=2)bias_up = tf.nn.bias_add(conv_up, biases_groups[0])bias_down = tf.nn.bias_add(conv_down, biases_groups[1])bias = tf.concat(axis=3, values=[bias_up, bias_down])conv4 = tf.nn.relu(bias, name=scope)# conv5with tf.name_scope('conv5') as scope:conv4_groups = tf.split(axis=3, value=conv4, num_or_size_splits=2)kernel = tf.Variable(tf.truncated_normal([3, 3, 192, 256],dtype=tf.float32,stddev=1e-1), name='weights')kernel_groups = tf.split(axis=3, value=kernel, num_or_size_splits=2)conv_up = tf.nn.conv2d(conv4_groups[0], kernel_groups[0], [1, 1, 1, 1], padding='SAME')conv_down = tf.nn.conv2d(conv4_groups[1], kernel_groups[1], [1, 1, 1, 1], padding='SAME')biases = tf.Variable(tf.constant(0.0, shape=[256], dtype=tf.float32),trainable=True, name='biases')biases_groups = tf.split(axis=0, value=biases, num_or_size_splits=2)bias_up = tf.nn.bias_add(conv_up, biases_groups[0])bias_down = tf.nn.bias_add(conv_down, biases_groups[1])bias = tf.concat(axis=3, values=[bias_up, bias_down])conv5 = tf.nn.relu(bias, name=scope)# pool5with tf.name_scope('pool5') as scope:pool5 = tf.nn.max_pool(conv5,ksize=[1, 3, 3, 1],strides=[1, 2, 2, 1],padding='VALID', )# flattened6with tf.name_scope('flattened6') as scope:flattened = tf.reshape(pool5, shape=[-1, 6 * 6 * 256])# fc6with tf.name_scope('fc6') as scope:weights = tf.Variable(tf.truncated_normal([6 * 6 * 256, 4096],dtype=tf.float32,stddev=1e-1), name='weights')biases = tf.Variable(tf.constant(0.0, shape=[4096], dtype=tf.float32),trainable=True, name='biases')bias = tf.nn.xw_plus_b(flattened, weights, biases)fc6 = tf.nn.relu(bias)# dropout6with tf.name_scope('dropout6') as scope:dropout6 = tf.nn.dropout(fc6, keep_prob)# fc7with tf.name_scope('fc7') as scope:weights = tf.Variable(tf.truncated_normal([4096, 4096],dtype=tf.float32,stddev=1e-1), name='weights')biases = tf.Variable(tf.constant(0.0, shape=[4096], dtype=tf.float32),trainable=True, name='biases')bias = tf.nn.xw_plus_b(dropout6, weights, biases)fc7 = tf.nn.relu(bias)# dropout7with tf.name_scope('dropout7') as scope:dropout7 = tf.nn.dropout(fc7, keep_prob)# fc8with tf.name_scope('fc8') as scope:weights = tf.Variable(tf.truncated_normal([4096, num_classes],dtype=tf.float32,stddev=1e-1), name='weights')biases = tf.Variable(tf.constant(0.0, shape=[num_classes], dtype=tf.float32),trainable=True, name='biases')fc8 = tf.nn.xw_plus_b(dropout7, weights, biases)return fc8#将所有的图片重新设置尺寸为227*227*3
w = 227
h = 227
c = 3
num_classes=2
learning_rate=0.5#mnist数据集中训练数据和测试数据保存地址
train_path = "train/"
test_path = "test/"#读取图片及其标签函数
def read_image(path,start_num,end_num):images = []labels = []for animal in ['cat','dog']:for img_num in range(start_num,end_num,1):#获取指定目录下的所有图片img=path + '/' + animal + '.'+str(img_num+1)+'.jpg'print("reading the image:%s"%img)image = io.imread(img)image = transform.resize(image,(w,h,c))images.append(image)if animal=='cat':labels.append([0,1])if animal == 'dog':labels.append([1,0])return np.asarray(images,dtype=np.float32),np.asarray(labels,dtype=np.int32)#array和asarray都可以将结构数据转化为ndarray，但是主要区别就是当数据源是ndarray时，array仍然会copy出一个副本，占用新的内存，但asarray不会#读取训练数据及测试数据
train_data,train_label = read_image(train_path,1,1+500)
test_data,test_label = read_image(train_path,1001,1001+500)#打乱训练数据及测试数据
train_image_num = len(train_data)
train_image_index = np.arange(train_image_num)#arange(start，stop, step, dtype=None)根据start与stop指定的范围以及step设定的步长，生成一个 ndarray。
np.random.shuffle(train_image_index)#乱序函数，多维时只对一维乱序，说明见https://blog.csdn.net/jasonzzj/article/details/53932645
train_data = train_data[train_image_index]#乱序后的数据
train_label = train_label[train_image_index]test_image_num = len(test_data)
test_image_index = np.arange(test_image_num)
np.random.shuffle(test_image_index)
test_data = test_data[test_image_index]
test_label = test_label[test_image_index]#alexnet调用x = tf.placeholder(tf.float32,[None,w,h,c],name='x')
y_ = tf.placeholder(tf.int32,[None,num_classes],name='y_')
keep_prob = tf.placeholder(tf.float32)
fc8 = alexnet(x, keep_prob, num_classes)# loss
with tf.name_scope('loss'):loss_op = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits_v2(logits=fc8,labels=y_))
# optimizer
with tf.name_scope('optimizer'):optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)train_op = optimizer.minimize(loss_op)# accuracy
with tf.name_scope("accuracy"):correct_pred = tf.equal(tf.argmax(fc8, 1), tf.argmax(y_, 1))accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))init = tf.global_variables_initializer()# Tensorboard
filewriter_path='alexnet_tmp/tensorboard'
tf.summary.scalar('loss', loss_op)
tf.summary.scalar('accuracy', accuracy)
merged_summary = tf.summary.merge_all()
writer = tf.summary.FileWriter(filewriter_path)# saver
saver = tf.train.Saver()#每次获取batch_size个样本进行训练或测试
def get_batch(data,label,batch_size):for start_index in range(0,len(data)-batch_size+1,batch_size):slice_index = slice(start_index,start_index+batch_size)yield data[slice_index],label[slice_index]#创建Session会话
with tf.Session() as sess:#初始化所有变量(权值，偏置等)#将所有样本训练10次，每次训练中以64个为一组训练完所有样本。#train_num可以设置大一些。train_num = 10batch_size = 64sess.run(init)j = 0for i in range(train_num):print("batchnum:",i)train_loss,train_acc,batch_num = 0, 0, 0for train_data_batch,train_label_batch in get_batch(train_data,train_label,batch_size):_,err,acc = sess.run([train_op,loss_op,accuracy],feed_dict={x:train_data_batch,y_:train_label_batch,keep_prob:0.5})train_loss+=err;train_acc+=acc;batch_num+=1j=j+1result=sess.run(merged_summary,feed_dict={x: train_data_batch, y_: train_label_batch, keep_prob: 0.5})writer.add_summary(result, j)print("train loss:",train_loss/(batch_num))print("train acc:",train_acc/(batch_num))test_loss,test_acc,batch_num = 0, 0, 0for test_data_batch,test_label_batch in get_batch(test_data,test_label,batch_size):err,acc = sess.run([loss_op,accuracy],feed_dict={x:test_data_batch,y_:test_label_batch,keep_prob:1})test_loss+=err;test_acc+=acc;batch_num+=1print("test loss:",test_loss/(batch_num))print("test acc:",test_acc/(batch_num))

数据集的来源：

https://www.kaggle.com/c/dogs-vs-cats-redux-kernels-edition/data（感觉把资源放在网盘根本下不下来）

训练结果准确度只有50%，可能是数据使用太少的原因

简单说明一下问题：

1.代码的注释不是很好，有时间会加上详细的注释

2.由于电脑本身的限制，没有使用很多数据去训练，结果其实是很差的，没有起到验证结果，有机会会验证代码

3.分析代码本身的一些问题，因为数据量是比较多的，所以说一次训练或者数据的读入有时是内存是不够的，正确的做法应该是将数据分次输入，每次结束训练后，将参数保存，再次读入，利用新的数据训练

4.Alexnet的论文没有进行阅读，找时间会将论文详细进行阅读

由于平时时间不是很多，博客的更新内容是平时想到的一些问题，如果感觉讲解很乱请谅解。以后有机会可能会对自己关注的问题进行博客的更新，例如，变尺寸图像的输入，不均衡样本的问题，少label的样本问题，小目标检测，网络的优化及过拟合问题，拟合可变参数的系统（如PSF去卷积问题，核在变化的情况），在不进行大量样本扩充和计算量提升的情况下，提高网络结果的准确性和抑制过拟合情形，实际项目的实时性要求和轻量化工作。如果有相应的见解，很希望能共同学习。

tensorflow学习（6.Alexnet实现及猫狗分类）相关推荐

Python深度学习（4）：猫狗分类
这个项目使用卷积神经网络,<Python深度学习>中使用了两个方法实现.一个是自己搭建卷积网络,另一个是直接使用VGG16.其中直接使用VGG16又可以分为抽取特征和微调模型两种方法. 1 ...
基于MATLAB的Alexnet迁移学习进行猫狗分类（数据集：Kaggle）
基本介绍软件:Matlab R2018b 数据集:Kaggle猫狗数据集网络:AlexNet 前期准备数据集 Kaggle猫狗数据集猫与狗用于训练的图片(train)分别12500张,每张图片的 ...
CV之IC之AlexNet：基于tensorflow框架采用CNN卷积神经网络算法(改进的AlexNet,训练/评估/推理)实现猫狗分类识别案例应用
CV之IC之AlexNet:基于tensorflow框架采用CNN卷积神经网络算法(改进的AlexNet,训练/评估/推理)实现猫狗分类识别案例应用目录基于tensorflow框架采用CNN(改进 ...
Tensorflow 学习之猫狗分类案例
Tensorflow 学习之猫狗分类案例本人一直在Cousera上学习Ng Andrew老师的Tensorflow课程,在本次猫狗分类案例当中,我对课程做了相应的记录,呈现在此,一方面加深学习的印象 ...
AlexNet 实现猫狗分类（keras and pytorch）
AlexNet 实现猫狗分类前言在训练网络过程中遇到了很多问题,先在这里抱怨一下,没有硬件条件去使用庞大的ImageNet2012 数据集 .所以在选择合适的数据集上走了些弯路,最后选择有kagg ...
TensorFlow 2.0 - tf.data.Dataset 数据预处理猫狗分类
文章目录 1 tf.data.Dataset.from_tensor_slices() 数据集建立 2. Dataset.map(f) 数据集预处理 3. Dataset.prefetch() 并行处 ...
tensorflow.js在nodejs训练猫狗分类模型在浏览器上使用
目录本人系统环境注意事项前言数据集准备处理数据集数据集初步处理将每一张图片数据转换成张量数据(tensor) 将图片转换成张量数组的代码和运行效果将图片的标注转换成张量数据(tenso ...
使用Tensorflow 2进行猫狗分类识别
使用Tensorflow 2进行猫狗分类识别本文参照了大佬Andrew Ng的所讲解的Tensorflow 2视频所写,本文将其中只适用于Linux的功能以及只适用于Google Colab的功能改 ...
PyTorch深度学习实战 | 猫狗分类
本文内容使用TensorFlow和Keras建立一个猫狗图片分类器. 图1 猫狗图片 01.安装TensorFlow和Keras库 TensorFlow是一个采用数据流图(data flow grap ...

tensorflow学习（6.Alexnet实现及猫狗分类）

tensorflow学习（6.Alexnet实现及猫狗分类）相关推荐

最新文章

热门文章