# encoding=utf-8
import tensorflow as tf
import numpy as np
from tensorflow.examples.tutorials.mnist import input_data

mnist = input_data.read_data_sets('MNIST_data', one_hot=True)


def weight_variable(shape):
    initial = tf.truncated_normal(shape, stddev=0.1)  # 定義Weights的函式  使用tf.trunncated_normal  標準值為0.1
    return tf.Variable(initial)


def bias_variable(shape):                             #定義biase函式，tf.constant
    initial = tf.constant(0.1, shape=shape)
    return tf.Variable(initial)

                             
def conv2d(x, W):                                       #定義卷積層
    return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')  # strides第0位和第3為一定為1，剩下的是卷積的橫向和縱向步長


def max_pool_2x2(x):                       #定義最大化池化層pooling
    return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')  # 引數同上，ksize是池化塊的大小


x = tf.placeholder(tf.float32,[None,784])
y_ = tf.placeholder(tf.float32, [None, 10])

# 影象轉化為一個四維張量，第一個引數代表樣本數量，-1表示不定，第二三引數代表影象尺寸，最後一個引數代表影象通道數
#x_image = tf.reshape(x, [-1, 28, 28, 1])

# 第一層卷積加池化
w_conv1 = weight_variable([5, 5, 1, 32])  # 第一二引數值得卷積核尺寸大小，即patch，第三個引數是影象通道數，第四個引數是卷積核的數目，代表會出現多少個卷積特徵 其實就是輸出的尺寸
b_conv1 = bias_variable([32])

h_conv1 = tf.nn.relu(conv2d(x, w_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)

# 第二層卷積加池化
w_conv2 = weight_variable([5, 5, 32, 64])  # 多通道卷積，卷積出64個特徵
b_conv2 = bias_variable([64])

h_conv2 = tf.nn.relu(conv2d(h_pool1, w_conv2) + b_conv2)
h_pool2 = max_pool_2x2(h_conv2)

# 原影象尺寸28*28，第一輪影象縮小為14*14，共有32張，第二輪後圖像縮小為7*7，共有64張

w_fc1 = weight_variable([7 * 7 * 64, 1024])
b_fc1 = bias_variable([1024])

h_pool2_flat = tf.reshape(h_pool2, [-1, 7 * 7 * 64])  # 展開，第一個引數為樣本數量，-1未知
f_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, w_fc1) + b_fc1)

# dropout操作，減少過擬合
keep_prob = tf.placeholder(tf.float32)
h_fc1_drop = tf.nn.dropout(f_fc1, keep_prob)

w_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
y_conv = tf.nn.softmax(tf.matmul(h_fc1_drop, w_fc2) + b_fc2)

cross_entropy = -tf.reduce_sum(y_ * tf.log(y_conv))  # 定義交叉熵為loss函式
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)  # 呼叫優化器優化

def compute_accuracy(v_xs,v_ys):

    correct_prediction=tf.equal(tf.argmax(y_conv,1),tf.argmax(v_ys,1))#比較你預測的y和真實y那個1所在的位置。是一個布林型量
    print(sess.run(correct_prediction))

    accuracy=tf.reduce_mean(tf.cast(correct_prediction,tf.float32))#轉化為float32格式然後求其正確率  tf.cast轉換格式
    result=sess.run(accuracy,feed_dict={x:v_xs,y_:v_ys})
    return result


sess = tf.InteractiveSession()
sess.run(tf.initialize_all_variables())
for i in range(2000):
    batch = mnist.train.next_batch(50)
    if i % 100 == 0:
        train_accuracy = compute_accuracy(batch[0], batch[1])
        print(sess.run(train_accuracy))
   # train_step.run(feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})
print(sess.run(compute_accuracy(mnist.test.images[0:50], mnist.test.labels[0:50])))

計算輸出特徵通道數，針對polling方式