启动此代码时，出现错误：
FailedPreconditionError (see above for traceback): Attempting to use uninitialized value model0/W5
已执行声明，
sess.run(tf.global_variables_initializer())
什么是问题？
import os
import cv2
import numpy as np
import sklearn.preprocessing as skp
import tensorflow as tf
将mnist图像转换为numpy后，
if __name__ == "__main__" :
Train_dir = "c:\\image\\MNIST\\trainingSet\\"
Test_dir = "c:\\image\\MNIST\\testSet\\"
train_folder_list = np.array(os.listdir(Train_dir))
test_folder_list = np.array(os.listdir(Test_dir))
integer_encoder = skp.LabelEncoder()
integer_encoded = integer_encoder.fit_transform(train_folder_list).reshape(len(train_folder_list), 1)
test_label_encoded = integer_encoder.fit_transform(test_folder_list).reshape(len(test_folder_list), 1)
onehot_encoder = skp.OneHotEncoder(sparse = False)
onehot_encoded = onehot_encoder.fit_transform(integer_encoded)
test_onehot_encoded = onehot_encoder.fit_transform(test_label_encoded
train_image_set = []
train_label_set = []
test_image_set = []
test_label_set = []
for index in range(len(train_folder_list)) :
path_train = os.path.join(Train_dir, train_folder_list[index]) + '\\'
path_test = os.path.join(Test_dir, test_folder_list[index]) + '\\'
train_image_list = np.array(os.listdir(path_train))
test_image_list = np.array(os.listdir(path_test))
for image in zip(train_image_list, test_image_list) :
train_image_dir = os.path.join(path_train, image[0])
test_image_dir = os.path.join(path_test, image[1])
train_img = cv2.imread(train_image_dir, cv2.IMREAD_GRAYSCALE)
test_img = cv2.imread(test_image_dir, cv2.IMREAD_GRAYSCALE)
train_image_set.append([np.array(train_img)]) #-> 4dim, list
train_label_set.append([np.array(onehot_encoded[index])])
test_image_set.append([np.array(test_img)])
test_label_set.append([np.array(test_onehot_encoded[index])])
train_image_set = np.reshape(train_image_set, (-1, 784)) #->2dim, np.ndar
train_label_set = np.reshape(train_label_set, (-1, 10))
test_image_set = np.reshape(test_image_set, (-1, 784)) #->2dim, np.ndar
test_label_set = np.reshape(test_label_set, (-1, 10))
train_image_set = np.array(train_image_set).astype(np.float32)
train_label_set = np.array(train_label_set).astype(np.float32)
test_image_set = np.array(test_image_set).astype(np.float32)
test_label_set = np.array(test_label_set).astype(np.float32)
print("\n--------------------Image transforming to numpy has completed--------------------\n")
开始了cnn
class CNN_model :
def __init__(self, sess, name) :
self.sess = sess
self.name = name
self._build_net()
def _build_net(self) :
with tf.variable_scope(self.name) :
self.X = tf.placeholder(tf.float32, shape = [None, 784])
X_img = tf.reshape(self.X, [-1, 28, 28, 1])
self.Y = tf.placeholder(tf.float32, shape = [None, 10])
self.keep_prob = tf.placeholder(tf.float32)
W1 = tf.Variable(tf.random_normal([3, 3, 1, 32], stddev = 0.01), name = 'Weight1')
L1 = tf.nn.conv2d(X_img, W1, strides = [1, 1, 1, 1], padding = 'SAME')
L1 = tf.nn.relu(L1)
L1 = tf.nn.max_pool(L1, ksize = [1, 2, 2, 1], strides = [1, 2, 2, 1], padding = 'SAME')
W2 = tf.Variable(tf.random_normal([3, 3, 32, 64], stddev = 0.01), name = 'Weight2')
L2 = tf.nn.conv2d(L1, W2, strides = [1, 1, 1, 1], padding = 'SAME')
L2 = tf.nn.relu(L2)
L2 = tf.nn.max_pool(L2, ksize = [1, 2, 2, 1], strides = [1, 2, 2, 1], padding = 'SAME')
W3 = tf.Variable(tf.random_normal([3, 3, 64, 128], stddev = 0.01), name = 'Weight3')
L3 = tf.nn.conv2d(L2, W3, strides = [1, 1, 1, 1], padding = 'SAME')
L3 = tf.nn.relu(L3)
L3 = tf.nn.max_pool(L3, ksize = [1, 2, 2, 1], strides = [1, 2, 2, 1], padding = 'SAME')
L3 = tf.reshape(L3, [-1, 4*4*128])
W4 = tf.get_variable("W4", shape = [4*4*128, 625], initializer = tf.contrib.layers.xavier_initializer())
b4 = tf.Variable(tf.random_normal([625]), name = 'bias4')
L4 = tf.nn.relu(tf.matmul(L3, W4) + b4)
L4 = tf.nn.dropout(L4, keep_prob = self.keep_prob)
W5 = tf.get_variable("W5", shape = [625, 10], initializer = tf.contrib.layers.xavier_initializer())
b5 = tf.Variable(tf.random_normal([10]), name = 'bais5')
self.logits = tf.matmul(L4, W5) + b5
#hypothesis = tf.nn.softmax(logits)
self.cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits = self.logits, labels = self.Y))
#self.cost = tf.reduce_mean(-tf.reduce_sum(self.Y*tf.log(hypothesis), axis = 1))
self.optimizer = tf.train.AdamOptimizer(learning_rate = 0.001).minimize(self.cost)
predict_correct = tf.equal(tf.argmax(tf.nn.softmax(self.logits), 1), tf.argmax(self.Y, 1))
self.Accuracy = tf.reduce_sum(tf.cast(predict_correct, tf.float32))
def train(self, X_data, Y_data, keep_prob = 0.7) :
return self.sess.run([self.cost, self.optimizer], feed_dict = { self.X : X_data, self.Y : Y_data, self.keep_prob : keep_prob })
def test(self, X_data, Y_data, keep_prob = 1.0) :
return self.sess.run(self.Accuracy, feed_dict = { self.X : X_data, self.Y : Y_data, self.keep_prob : keep_prob })
def predict(self, X_data, keep_prob = 1.0) :
return self.ses.run(self.logits, feed_dict = { self.X : X_data, self.keep_prob : keep_prob })
tf.reset_default_graph()
sess = tf.Session()
sess.run(tf.global_variables_initializer())
total_epoch = 5
batch_size = 100
model_num = 3
model_list = []
train_data = train_image_set
train_label = train_label_set
test_data = test_image_set
test_label = test_label_set
for i in range(model_num) :
model_list.append(CNN_model(sess, "model" + str(i)))
print("\n--------------------CNN_Learning has started--------------------\n")
for epoch in range(total_epoch) :
avg_cost_list = np.zeros(len(model_list))
total_batch = int(len(train_data)/ batch_size)
for i in range(total_batch) :
start = ((i+1)*batch_size) - batch_size
end = (i+1)*batch_size
batch_xs = train_data[start : end]
batch_ys = train_label[start : end]
for m_idx, m in enumerate(model_list) :
c, _ = m.train(batch_xs, batch_ys)
avg_cost_list[m_idx] += c/total_batch
print("Epoch : ", epoch + 1, "Cost : ", avg_cost_list)
print("\n--------------------CNN_Learning has completed--------------------\n")
但我得到一个例外：
FailedPreconditionError (see above for traceback): Attempting to use uninitialized value model0/beta1_power
[[Node: model0/beta1_power/read = Identity[T=DT_FLOAT, _class=["loc:@model0/Adam/Assign"], _device="/job:localhost/replica:0/task:0/device:CPU:0"](model0/beta1_power)]]