import tensorflow as tf
import numpy as np
import os
import pickle
# 数据载入参数
def load_data(filename):
with open(filename,'rb') as f:
data = pickle.load(f,encoding='bytes')
return data[b'data'],data[b'labels']
# 建立数据处理类
# 方便从cifar10取出图片,打乱顺序等操作。
class CifarData:
def __init__(self,filenames,need_shuffle):
all_data = []
all_labels = []
for filename in filenames:
data,label = load_data(filename)
all_data.append(data)
all_labels.append(label)
self._data = np.vstack(all_data)
self._data = self._data / 127.5 - 1
self._labels = np.hstack(all_labels)
self._num_examples = self._data.shape[0]
self._need_shauffle = need_shuffle
self._start_index = 0
if self._need_shauffle:
self.shuffle_data()
def shuffle_data(self):
o = np.random.permutation(self._num_examples)
self._data = self._data[o]
self._labels = self._labels[o]
def next_batch(self,batch_size):
end_index = self._start_index + batch_size
if end_index > self._num_examples:
if self._need_shauffle:
self.shuffle_data()
self._start_index = 0
end_index = batch_size
else:
raise Exception('没有过多的样本')
if end_index > self._num_examples:
raise Exception('尺寸过大')
batch_data = self._data[self._start_index:end_index]
batch_labels = self._labels[self._start_index:end_index]
self._start_index = end_index
return batch_data,batch_labels
# 文件夹路径
cifar_dir = './cifar-10-batches-py'
print(os.listdir(cifar_dir))
# 构建文件组
train_filenames = [os.path.join(cifar_dir,'data_batch_%d'%i)for i in range(1,6)]
test_filenames = [os.path.join(cifar_dir,'test_batch')]
# 实例化
train_data = CifarData(train_filenames,True)
test_data = CifarData(test_filenames,False)
# 模型
X = tf.placeholder(dtype=tf.float32,shape=[None,3072])
Y = tf.placeholder(tf.int64)
# 构建神经网络模型
hidden_1 = tf.layers.dense(X,400,activation=tf.nn.relu)
hidden_2 = tf.layers.dense(hidden_1,120,activation=tf.nn.relu)
hidden_3 = tf.layers.dense(hidden_2,84,activation=tf.nn.relu)
y_ = tf.layers.dense(hidden_3,10)
loss = tf.losses.sparse_softmax_cross_entropy(labels=Y,logits=y_)
predict = tf.argmax(y_,1)
accuracy = tf.reduce_mean(tf.cast(tf.equal(Y,predict),dtype=tf.float32))
with tf.name_scope('train_op'):
train_op = tf.train.AdamOptimizer(learning_rate=0.001).minimize(loss)
# 超参数
batch_size = 20
train_steps = 10000
test_steps = 100
# 开启会话
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
for i in range(train_steps):
x_train,y_train = train_data.next_batch(batch_size)
c,acc,_ = sess.run([loss,accuracy,train_op],feed_dict={X:x_train,Y:y_train})
if (i+1) % 500 == 0:
print('代价:',c)
print('准确率: ',acc)
if (i + 1) % 5000 == 0:
# 测试
test_data = CifarData(test_filenames,False)
acc_all = []
for j in range(test_steps):
x_test, yh_test = test_data.next_batch(batch_size)
acc_val = sess.run(accuracy,feed_dict={X:x_test,Y:yh_test})
acc_all.append(acc_val)
print('测试准确率:',sess.run(tf.reduce_mean(acc_all)))
深度学习之神经网络图像分类cifar10数据集
最新推荐文章于 2024-05-09 09:45:00 发布