def train(X_test,y_test_lable):
x_ = tf.placeholder(tf.float32, [None, INPUT_NODE],name='x-input')
x = tf.reshape(x_, shape=[-1, 28, 28, 1])
y_ = tf.placeholder(tf.float32, [None,OUTPUT_NODE], name='y-input')
regularizer = tf.contrib.layers.l2_regularizer(REGULARIZATION_RATE)
cosine,loss = lenet5_infernece.inference(x,False,regularizer,tf.argmax(y_,1))
# Evaluate model
pred_max=tf.argmax(cosine,1)
y_max=tf.argmax(y_,1)
correct_pred = tf.equal(pred_max,y_max)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
# 初始化TensorFlow持久化類。
saver = tf.train.Saver()
with tf.Session() as sess:
saver.restore(sess,"lenet5/lenet5_model")
X_test = np.reshape(X_test, (
X_test.shape[0],
IMAGE_SIZE,
IMAGE_SIZE,
NUM_CHANNELS))
acc = sess.run(accuracy, feed_dict={x: X_test, y_:y_test_lable})
print('Test accuracy: %.2f%%' % (acc * 100))
def initVarible():
shuffle = True
batch_idx = 0
train_acc = []
# 定義輸出為4維矩陣的placeholder
# x_ = tf.placeholder(tf.float32, [None, INPUT_NODE],name='x-input')
# x = tf.reshape(x_, shape=[-1, 28, 28, 1])
x = tf.placeholder(tf.float32, [None, 28, 28, 1], name='x-input')
y_ = tf.placeholder(tf.float32, [None, OUTPUT_NODE], name='y-input')
regularizer = tf.contrib.layers.l2_regularizer(REGULARIZATION_RATE)
y = lenet5_infernece.inference(x, True, regularizer,
tf.AUTO_REUSE) #tf.AUTO_REUSE
global_step = tf.Variable(0, trainable=False)
# Evaluate model
pred_max = tf.argmax(y, 1)
y_max = tf.argmax(y_, 1)
correct_pred = tf.equal(pred_max, y_max)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
# 定義損失函數、學習率、及訓練過程。
cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(
logits=y, labels=tf.argmax(y_, 1))
cross_entropy_mean = tf.reduce_mean(cross_entropy)
loss = cross_entropy_mean + tf.add_n(tf.get_collection('losses'))
return global_step, shuffle, batch_idx, loss, x, y_, accuracy, train_acc
def train(X_train,y_train_lable,X_test,y_test_lable):
shuffle=True
batch_idx=0
batch_len =int( X_train.shape[0]/BATCH_SIZE)
test_batch_len =int( X_test.shape[0]/BATCH_SIZE)
test_acc=[]
train_acc=[]
train_idx=np.random.permutation(batch_len)#打散btach_len=600 group
x_ = tf.placeholder(tf.float32, [None, INPUT_NODE],name='x-input')
x = tf.reshape(x_, shape=[-1, 28, 28, 1])
y_ = tf.placeholder(tf.float32, [None,NUM_LABELS], name='y-input')
regularizer = tf.contrib.layers.l2_regularizer(REGULARIZATION_RATE)
cosine,loss= lenet5_infernece.inference(x,True,regularizer,tf.argmax(y_,1))
global_step = tf.Variable(0, trainable=False)
pred_max=tf.argmax(cosine,1)
y_max=tf.argmax(y_,1)
correct_pred = tf.equal(pred_max,y_max)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
train_step = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss, global_step=global_step)
saver = tf.train.Saver()
with tf.Session() as sess:
tf.global_variables_initializer().run()
step = 1
print ("Start training!")
while step < TRAINING_STEPS:
batch_shuffle_idx=train_idx[batch_idx]
batch_xs=X_train[batch_shuffle_idx*BATCH_SIZE:batch_shuffle_idx*BATCH_SIZE+BATCH_SIZE]
batch_ys=y_train_lable[batch_shuffle_idx*BATCH_SIZE:batch_shuffle_idx*BATCH_SIZE+BATCH_SIZE]
if batch_idx<batch_len:
batch_idx+=1
if batch_idx==batch_len:
batch_idx=0
else:
batch_idx=0
reshaped_xs = np.reshape(batch_xs, (
BATCH_SIZE,
IMAGE_SIZE,
IMAGE_SIZE,
NUM_CHANNELS))
_, loss_value, step = sess.run([train_step, loss, global_step], feed_dict={x: reshaped_xs, y_: batch_ys})
#print loss_value,step
if step % display_step == 0:
X_test=np.reshape(X_test[:3000],(3000,28,28,1))
loss_value,acc=sess.run([loss,accuracy] , feed_dict={x:X_test,y_:y_test_lable[:3000]})
print("After %d training step(s), loss on test data is %g,acc is %g" % (step, loss_value,acc))
saver.save(sess, os.path.join(MODEL_SAVE_PATH, MODEL_NAME),global_step=global_step)
step += 1
print ("Optimization Finished!")
saver.save(sess, os.path.join(MODEL_SAVE_PATH, MODEL_NAME))
def evaluate(X_test):
with tf.Graph().as_default() as g:
# 定義輸出為4維矩陣的placeholder
x_ = tf.placeholder(tf.float32, [None, lenet5_train.INPUT_NODE],
name='x-input')
x = tf.reshape(x_, shape=[-1, 28, 28, 1])
y_ = tf.placeholder(tf.float32, [None, lenet5_train.OUTPUT_NODE],
name='y-input')
regularizer = tf.contrib.layers.l2_regularizer(
lenet5_train.REGULARIZATION_RATE)
y = lenet5_infernece.inference(x, False, regularizer)
global_step = tf.Variable(0, trainable=False)
# Evaluate model
pred_max = tf.argmax(y, 1)
y_max = tf.argmax(y_, 1)
correct_pred = tf.equal(pred_max, y_max)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
test_batch_len = int(X_test.shape[0] / lenet5_train.BATCH_SIZE)
test_acc = []
kaggle_pred = np.array([])
test_xs = np.reshape(
X_test, (X_test.shape[0], lenet5_train.IMAGE_SIZE,
lenet5_train.IMAGE_SIZE, lenet5_train.NUM_CHANNELS))
batchsize = lenet5_train.BATCH_SIZE
# 'Saver' op to save and restore all the variables
saver = tf.train.Saver()
with tf.Session() as sess:
saver.restore(sess, "./lenet5/lenet5_model")
for i in range(test_batch_len):
pred_result = sess.run(pred_max,
feed_dict={
x:
test_xs[batchsize *
i:batchsize * i + batchsize]
})
kaggle_pred = np.append(kaggle_pred, pred_result)
kaggle_pred = kaggle_pred.astype(int)
kaggle_pred = kaggle_pred.tolist()
print("pred_result.length:", len(kaggle_pred))
#print("pred_result=",kaggle_pred)
print("Save prediction result...")
saveResult(kaggle_pred)
return
def evaluate(X_test, y_test_lable, My_Yd):
with tf.Graph().as_default() as g:
# 定義輸出為4維矩陣的placeholder
x_ = tf.placeholder(tf.float32, [None, lenet5_train.INPUT_NODE],
name='x-input')
x = tf.reshape(x_, shape=[-1, 28, 28, 1])
y_ = tf.placeholder(tf.float32, [None, lenet5_train.OUTPUT_NODE],
name='y-input')
regularizer = tf.contrib.layers.l2_regularizer(
lenet5_train.REGULARIZATION_RATE)
y = lenet5_infernece.inference(x, False, regularizer)
global_step = tf.Variable(0, trainable=False)
# Evaluate model
pred_max = tf.argmax(y, 1)
y_max = tf.argmax(y_, 1)
correct_pred = tf.equal(pred_max, y_max)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
batchsize = 20
test_batch_len = int(X_test.shape[0] / batchsize)
test_acc = []
test_xs = np.reshape(
X_test, (X_test.shape[0], lenet5_train.IMAGE_SIZE,
lenet5_train.IMAGE_SIZE, lenet5_train.NUM_CHANNELS))
# 'Saver' op to save and restore all the variables
saver = tf.train.Saver()
#saver = tf.train.import_meta_graph("./mnist/mnist_model.meta")
with tf.Session() as sess:
saver.restore(sess, "./lenet5/lenet5_model")
My_test_pred = sess.run(pred_max, feed_dict={x: test_xs[:20]})
print("期望值：", My_Yd)
print("預測值：", My_test_pred)
My_acc = sess.run(accuracy,
feed_dict={
x: test_xs,
y_: y_test_lable
})
print('Test accuracy: %.2f%%' % (My_acc * 100))
display_result(My_test_pred)
return
def evaluate(X_test,y_test_lable):
with tf.Graph().as_default() as g:
# 定義輸出為4維矩陣的placeholder
x_ = tf.placeholder(tf.float32, [None, lenet5_train.INPUT_NODE],name='x-input')
x = tf.reshape(x_, shape=[-1, 28, 28, 1])
y_ = tf.placeholder(tf.float32, [None, lenet5_train.OUTPUT_NODE], name='y-input')
regularizer = tf.contrib.layers.l2_regularizer(lenet5_train.REGULARIZATION_RATE)
y = lenet5_infernece.inference(x,False,regularizer)
global_step = tf.Variable(0, trainable=False)
# Evaluate model
pred_max=tf.argmax(y,1)
y_max=tf.argmax(y_,1)
correct_pred = tf.equal(pred_max,y_max)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
test_batch_len =int( X_test.shape[0]/lenet5_train.BATCH_SIZE)
test_acc=[]
test_xs = np.reshape(X_test, (
X_test.shape[0],
lenet5_train.IMAGE_SIZE,
lenet5_train.IMAGE_SIZE,
lenet5_train.NUM_CHANNELS))
batchsize=lenet5_train.BATCH_SIZE
# 'Saver' op to save and restore all the variables
saver = tf.train.Saver()
with tf.Session() as sess:
saver.restore(sess,"./lenet5/lenet5_model")
for i in range(test_batch_len):
temp_acc= sess.run(accuracy, feed_dict={x: test_xs[batchsize*i:batchsize*i+batchsize], y_: y_test_lable[batchsize*i:batchsize*i+batchsize]})
test_acc.append(temp_acc)
print ("Test batch ",i,":Testing Accuracy:",temp_acc)
t_acc=tf.reduce_mean(tf.cast(test_acc, tf.float32))
print("Average Testing Accuracy=",sess.run(t_acc))
return
def evaluate(X_test, y_test_lable, My_Yd):
with tf.Graph().as_default() as g:
# 定義輸出為4維矩陣的placeholder
x_ = tf.placeholder(tf.float32, [None, lenet5_train.INPUT_NODE],
name='x-input')
x = tf.reshape(x_, shape=[-1, 28, 28, 1])
y_ = tf.placeholder(tf.float32, [None, lenet5_train.OUTPUT_NODE],
name='y-input')
regularizer = tf.contrib.layers.l2_regularizer(
lenet5_train.REGULARIZATION_RATE)
y = lenet5_infernece.inference(x, False, regularizer,
False) #tf.AUTO_REUSE
global_step = tf.Variable(0, trainable=False)
# Evaluate model
pred_max = tf.argmax(y, 1)
y_max = tf.argmax(y_, 1)
correct_pred = tf.equal(pred_max, y_max)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
batchsize = 20
test_batch_len = int(X_test.shape[0] / batchsize)
test_acc = []
test_xs = np.reshape(
X_test, (X_test.shape[0], lenet5_train.IMAGE_SIZE,
lenet5_train.IMAGE_SIZE, lenet5_train.NUM_CHANNELS))
# 'Saver' op to save and restore all the variables
saver = tf.train.Saver()
# saver = tf.train.import_meta_graph("./mnist/mnist_model.meta")
with tf.Session() as sess:
if MODEL_SELECT == 0:
MODEL_DIR = MODEL_DIR_NUMBER
else:
MODEL_DIR = MODEL_DIR_LETTER
saver.restore(sess, MODEL_DIR)
My_test_pred = sess.run(pred_max, feed_dict={x: test_xs[:20]})
print("期望值：", My_Yd)
print("before cnvert My_test_pred =")
print(My_test_pred)
My_test_pred1 = []
for i in range(len(My_test_pred)):
print("My_test_pred[i] =")
print(My_test_pred[i])
if MODEL_SELECT == 0:
if My_test_pred[i] == 0:
My_test_pred1.append(0)
elif My_test_pred[i] == 1:
My_test_pred1.append(1)
print("A=")
elif My_test_pred[i] == 2:
My_test_pred1.append(2)
elif My_test_pred[i] == 3:
My_test_pred1.append(3)
elif My_test_pred[i] == 4:
My_test_pred1.append(4)
elif My_test_pred[i] == 5:
My_test_pred1.append(5)
elif My_test_pred[i] == 6:
My_test_pred1.append(6)
elif My_test_pred[i] == 7:
My_test_pred1.append(7)
elif My_test_pred[i] == 8:
My_test_pred1.append(8)
elif My_test_pred[i] == 9:
My_test_pred1.append(9)
elif My_test_pred[i] == 10:
My_test_pred1.append('A')
elif My_test_pred[i] == 11:
My_test_pred1.append('B')
elif My_test_pred[i] == 12:
My_test_pred1.append('C')
elif My_test_pred[i] == 13:
My_test_pred1.append('D')
elif My_test_pred[i] == 14:
My_test_pred1.append('E')
elif My_test_pred[i] == 15:
My_test_pred1.append('F')
elif My_test_pred[i] == 16:
My_test_pred1.append('G')
elif My_test_pred[i] == 17:
My_test_pred1.append('H')
elif My_test_pred[i] == 18:
My_test_pred1.append('I')
elif My_test_pred[i] == 19:
My_test_pred1.append('J')
else:
My_test_pred1.append('--')
else:
if My_test_pred[i] == 0:
My_test_pred1.append('A')
elif My_test_pred[i] == 1:
My_test_pred1.append('B')
print("A=")
elif My_test_pred[i] == 2:
My_test_pred1.append('C')
elif My_test_pred[i] == 3:
My_test_pred1.append('D')
elif My_test_pred[i] == 4:
My_test_pred1.append('E')
elif My_test_pred[i] == 5:
My_test_pred1.append('F')
elif My_test_pred[i] == 6:
My_test_pred1.append('G')
elif My_test_pred[i] == 7:
My_test_pred1.append('H')
elif My_test_pred[i] == 8:
My_test_pred1.append('I')
elif My_test_pred[i] == 9:
My_test_pred1.append('J')
else:
My_test_pred1.append('--')
print("預測值：", My_test_pred1)
My_acc = sess.run(accuracy,
feed_dict={
x: test_xs,
y_: y_test_lable
})
print('Test accuracy: %.2f%%' % (My_acc * 100))
display_result(My_test_pred, My_test_pred1)
return
def train(X_train, y_train_lable, X_test, y_test_lable):
shuffle = True
batch_idx = 0
batch_len = int(X_train.shape[0] / BATCH_SIZE)
test_batch_len = int(X_test.shape[0] / BATCH_SIZE)
test_acc = []
train_acc = []
train_idx = np.random.permutation(batch_len) #打散btach_len=600 group
# 定義輸出為4維矩陣的placeholder
x_ = tf.placeholder(tf.float32, [None, INPUT_NODE], name='x-input')
x = tf.reshape(x_, shape=[-1, 28, 28, 1])
y_ = tf.placeholder(tf.float32, [None, OUTPUT_NODE], name='y-input')
regularizer = tf.contrib.layers.l2_regularizer(REGULARIZATION_RATE)
y = lenet5_infernece.inference(x, True, regularizer)
global_step = tf.Variable(0, trainable=False)
# Evaluate model
pred_max = tf.argmax(y, 1)
y_max = tf.argmax(y_, 1)
correct_pred = tf.equal(pred_max, y_max)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
# 定義損失函數、學習率、及訓練過程。
cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(
logits=y, labels=tf.argmax(y_, 1))
cross_entropy_mean = tf.reduce_mean(cross_entropy)
loss = cross_entropy_mean + tf.add_n(tf.get_collection('losses'))
if learning_rate_flag == True:
learning_rate = tf.train.exponential_decay(LEARNING_RATE_BASE,
global_step,
X_train.shape[0] /
BATCH_SIZE,
LEARNING_RATE_DECAY,
staircase=True)
else:
learning_rate = 0.001 #Ashing test
train_step = tf.train.GradientDescentOptimizer(learning_rate).minimize(
loss, global_step=global_step)
# 初始化TensorFlow持久化類。
saver = tf.train.Saver()
with tf.Session() as sess:
tf.global_variables_initializer().run()
step = 1
print("Start training!")
while step < TRAINING_STEPS:
#batch_xs, batch_ys = mnist.train.next_batch(BATCH_SIZE)
if shuffle == True:
batch_shuffle_idx = train_idx[batch_idx]
batch_xs = X_train[batch_shuffle_idx *
BATCH_SIZE:batch_shuffle_idx * BATCH_SIZE +
BATCH_SIZE]
batch_ys = y_train_lable[batch_shuffle_idx *
BATCH_SIZE:batch_shuffle_idx *
BATCH_SIZE + BATCH_SIZE]
else:
batch_xs = X_train[batch_idx *
BATCH_SIZE:batch_idx * BATCH_SIZE +
BATCH_SIZE]
batch_ys = y_train_lable[batch_idx *
BATCH_SIZE:batch_idx * BATCH_SIZE +
BATCH_SIZE]
if batch_idx < batch_len:
batch_idx += 1
if batch_idx == batch_len:
batch_idx = 0
else:
batch_idx = 0
reshaped_xs = np.reshape(
batch_xs, (BATCH_SIZE, IMAGE_SIZE, IMAGE_SIZE, NUM_CHANNELS))
# Fit training using batch data
_, loss_value, step = sess.run([train_step, loss, global_step],
feed_dict={
x: reshaped_xs,
y_: batch_ys
})
acc = sess.run(accuracy, feed_dict={x: reshaped_xs, y_: batch_ys})
train_acc.append(acc)
if step % display_step == 0:
print(
"After %d training step(s), loss on training batch is %g,Training Accuracy=%g"
% (step, loss_value, acc))
step += 1
print("Optimization Finished!")
train_acc_avg = tf.reduce_mean(tf.cast(train_acc, tf.float32))
print("Average Training Accuracy=", sess.run(train_acc_avg))
print("Save model...")
#saver.save(sess, "./lenet5/lenet5_model")
saver.save(sess, os.path.join(MODEL_SAVE_PATH, MODEL_NAME))