tensorflow 1.0 學習：模型的保存與恢復(Saver)
將訓練好的模型參數保存起來，以便以后進行驗證或測試，這是我們經常要做的事情。tf里面提供模型保存的是tf.train.Saver()模塊。
模型保存，先要創建一個Saver對象：如
saver=tf.train.Saver()
在創建這個Saver對象的時候，有一個參數我們經常會用到，就是 max_to_keep 參數，這個是用來設置保存模型的個數，默認為5，即 max_to_keep=5，保存最近的5個模型。如果你想每訓練一代（epoch)就想保存一次模型，則可以將 max_to_keep設置為None或者0，如：
saver=tf.train.Saver(max_to_keep=0)
但是這樣做除了多占用硬盤，並沒有實際多大的用處，因此不推薦。
當然，如果你只想保存最后一代的模型，則只需要將max_to_keep設置為1即可，即
saver=tf.train.Saver(max_to_keep=1)
創建完saver對象后，就可以保存訓練好的模型了，如：
saver.save(sess,'ckpt/mnist.ckpt',global_step=step)
第一個參數sess,這個就不用說了。第二個參數設定保存的路徑和名字，第三個參數將訓練的次數作為后綴加入到模型名字中。
saver.save(sess, 'my-model', global_step=0) ==> filename: 'my-model-0'
...
saver.save(sess, 'my-model', global_step=1000) ==> filename: 'my-model-1000'
看一個mnist實例：
# -*- coding: utf-8 -*-
"""
Created on Sun Jun 4 10:29:48 2017
@author: Administrator
"""
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=False)
x = tf.placeholder(tf.float32, [None, 784])
y_=tf.placeholder(tf.int32,[None,])
dense1 = tf.layers.dense(inputs=x,
units=1024,
activation=tf.nn.relu,
kernel_initializer=tf.truncated_normal_initializer(stddev=0.01),
kernel_regularizer=tf.nn.l2_loss)
dense2= tf.layers.dense(inputs=dense1,
units=512,
activation=tf.nn.relu,
kernel_initializer=tf.truncated_normal_initializer(stddev=0.01),
kernel_regularizer=tf.nn.l2_loss)
logits= tf.layers.dense(inputs=dense2,
units=10,
activation=None,
kernel_initializer=tf.truncated_normal_initializer(stddev=0.01),
kernel_regularizer=tf.nn.l2_loss)
loss=tf.losses.sparse_softmax_cross_entropy(labels=y_,logits=logits)
train_op=tf.train.AdamOptimizer(learning_rate=0.001).minimize(loss)
correct_prediction = tf.equal(tf.cast(tf.argmax(logits,1),tf.int32), y_)
acc= tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
sess=tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
saver=tf.train.Saver(max_to_keep=1)
for i in range(100):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_op, feed_dict={x: batch_xs, y_: batch_ys})
val_loss,val_acc=sess.run([loss,acc], feed_dict={x: mnist.test.images, y_: mnist.test.labels})
print('epoch:%d, val_loss:%f, val_acc:%f'%(i,val_loss,val_acc))
saver.save(sess,'ckpt/mnist.ckpt',global_step=i+1)
sess.close()
代碼中紅色部分就是保存模型的代碼，雖然我在每訓練完一代的時候，都進行了保存，但后一次保存的模型會覆蓋前一次的，最終只會保存最后一次。因此我們可以節省時間，將保存代碼放到循環之外（僅適用max_to_keep=1,否則還是需要放在循環內).
在實驗中，最后一代可能並不是驗證精度最高的一代，因此我們並不想默認保存最后一代，而是想保存驗證精度最高的一代，則加個中間變量和判斷語句就可以了。
saver=tf.train.Saver(max_to_keep=1)
max_acc=0
for i in range(100):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_op, feed_dict={x: batch_xs, y_: batch_ys})
val_loss,val_acc=sess.run([loss,acc], feed_dict={x: mnist.test.images, y_: mnist.test.labels})
print('epoch:%d, val_loss:%f, val_acc:%f'%(i,val_loss,val_acc))
if val_acc>max_acc:
max_acc=val_acc
saver.save(sess,'ckpt/mnist.ckpt',global_step=i+1)
sess.close()
如果我們想保存驗證精度最高的三代，且把每次的驗證精度也隨之保存下來，則我們可以生成一個txt文件用於保存。
saver=tf.train.Saver(max_to_keep=3)
max_acc=0
f=open('ckpt/acc.txt','w')
for i in range(100):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_op, feed_dict={x: batch_xs, y_: batch_ys})
val_loss,val_acc=sess.run([loss,acc], feed_dict={x: mnist.test.images, y_: mnist.test.labels})
print('epoch:%d, val_loss:%f, val_acc:%f'%(i,val_loss,val_acc))
f.write(str(i+1)+', val_acc: '+str(val_acc)+'\n')
if val_acc>max_acc:
max_acc=val_acc
saver.save(sess,'ckpt/mnist.ckpt',global_step=i+1)
f.close()
sess.close()
模型的恢復用的是restore()函數，它需要兩個參數restore(sess, save_path)，save_path指的是保存的模型路徑。我們可以使用tf.train.latest_checkpoint（）來自動獲取最后一次保存的模型。如：
model_file=tf.train.latest_checkpoint('ckpt/')
saver.restore(sess,model_file)
則程序后半段代碼我們可以改為：
sess=tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
is_train=False
saver=tf.train.Saver(max_to_keep=3)
#訓練階段
if is_train:
max_acc=0
f=open('ckpt/acc.txt','w')
for i in range(100):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_op, feed_dict={x: batch_xs, y_: batch_ys})
val_loss,val_acc=sess.run([loss,acc], feed_dict={x: mnist.test.images, y_: mnist.test.labels})
print('epoch:%d, val_loss:%f, val_acc:%f'%(i,val_loss,val_acc))
f.write(str(i+1)+', val_acc: '+str(val_acc)+'\n')
if val_acc>max_acc:
max_acc=val_acc
saver.save(sess,'ckpt/mnist.ckpt',global_step=i+1)
f.close()
#驗證階段
else:
model_file=tf.train.latest_checkpoint('ckpt/')
saver.restore(sess,model_file)
val_loss,val_acc=sess.run([loss,acc], feed_dict={x: mnist.test.images, y_: mnist.test.labels})
print('val_loss:%f, val_acc:%f'%(val_loss,val_acc))
sess.close()
標紅的地方，就是與保存、恢復模型相關的代碼。用一個bool型變量is_train來控制訓練和驗證兩個階段。
整個源程序：
# -*- coding: utf-8 -*-
"""
Created on Sun Jun 4 10:29:48 2017
@author: Administrator
"""
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=False)
x = tf.placeholder(tf.float32, [None, 784])
y_=tf.placeholder(tf.int32,[None,])
dense1 = tf.layers.dense(inputs=x,
units=1024,
activation=tf.nn.relu,
kernel_initializer=tf.truncated_normal_initializer(stddev=0.01),
kernel_regularizer=tf.nn.l2_loss)
dense2= tf.layers.dense(inputs=dense1,
units=512,
activation=tf.nn.relu,
kernel_initializer=tf.truncated_normal_initializer(stddev=0.01),
kernel_regularizer=tf.nn.l2_loss)
logits= tf.layers.dense(inputs=dense2,
units=10,
activation=None,
kernel_initializer=tf.truncated_normal_initializer(stddev=0.01),
kernel_regularizer=tf.nn.l2_loss)
loss=tf.losses.sparse_softmax_cross_entropy(labels=y_,logits=logits)
train_op=tf.train.AdamOptimizer(learning_rate=0.001).minimize(loss)
correct_prediction = tf.equal(tf.cast(tf.argmax(logits,1),tf.int32), y_)
acc= tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
sess=tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
is_train=True
saver=tf.train.Saver(max_to_keep=3)
#訓練階段
if is_train:
max_acc=0
f=open('ckpt/acc.txt','w')
for i in range(100):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_op, feed_dict={x: batch_xs, y_: batch_ys})
val_loss,val_acc=sess.run([loss,acc], feed_dict={x: mnist.test.images, y_: mnist.test.labels})
print('epoch:%d, val_loss:%f, val_acc:%f'%(i,val_loss,val_acc))
f.write(str(i+1)+', val_acc: '+str(val_acc)+'\n')
if val_acc>max_acc:
max_acc=val_acc
saver.save(sess,'ckpt/mnist.ckpt',global_step=i+1)
f.close()
#驗證階段
else:
model_file=tf.train.latest_checkpoint('ckpt/')
saver.restore(sess,model_file)
val_loss,val_acc=sess.run([loss,acc], feed_dict={x: mnist.test.images, y_: mnist.test.labels})
print('val_loss:%f, val_acc:%f'%(val_loss,val_acc))
sess.close()
參考文章：http://blog.csdn.net/u011500062/article/details/51728830
tensorflow 1.0 學習：用別人訓練好的模型來進行圖像分類
谷歌在大型圖像數據庫ImageNet上訓練好了一個Inception-v3模型，這個模型我們可以直接用來進來圖像分類。
下載地址：https://storage.googleapis.com/download.tensorflow.org/models/inception_dec_2015.zip
下載完解壓后，得到幾個文件：
其中的classify_image_graph_def.pb 文件就是訓練好的Inception-v3模型。
imagenet_synset_to_human_label_map.txt是類別文件。
隨機找一張圖片：如
對這張圖片進行識別，看它屬於什么類？
代碼如下：先創建一個類NodeLookup來將softmax概率值映射到標簽上。
然后創建一個函數create_graph()來讀取模型。
最后讀取圖片進行分類識別：
# -*- coding: utf-8 -*-
import tensorflow as tf
import numpy as np
import re
import os
model_dir='D:/tf/model/'
image='d:/cat.jpg'
#將類別ID轉換為人類易讀的標簽
class NodeLookup(object):
def __init__(self,
label_lookup_path=None,
uid_lookup_path=None):
if not label_lookup_path:
label_lookup_path = os.path.join(
model_dir, 'imagenet_2012_challenge_label_map_proto.pbtxt')
if not uid_lookup_path:
uid_lookup_path = os.path.join(
model_dir, 'imagenet_synset_to_human_label_map.txt')
self.node_lookup = self.load(label_lookup_path, uid_lookup_path)
def load(self, label_lookup_path, uid_lookup_path):
if not tf.gfile.Exists(uid_lookup_path):
tf.logging.fatal('File does not exist %s', uid_lookup_path)
if not tf.gfile.Exists(label_lookup_path):
tf.logging.fatal('File does not exist %s', label_lookup_path)
# Loads mapping from string UID to human-readable string
proto_as_ascii_lines = tf.gfile.GFile(uid_lookup_path).readlines()
uid_to_human = {}
p = re.compile(r'[n\d]*[ \S,]*')
for line in proto_as_ascii_lines:
parsed_items = p.findall(line)
uid = parsed_items[0]
human_string = parsed_items[2]
uid_to_human[uid] = human_string
# Loads mapping from string UID to integer node ID.
node_id_to_uid = {}
proto_as_ascii = tf.gfile.GFile(label_lookup_path).readlines()
for line in proto_as_ascii:
if line.startswith(' target_class:'):
target_class = int(line.split(': ')[1])
if line.startswith(' target_class_string:'):
target_class_string = line.split(': ')[1]
node_id_to_uid[target_class] = target_class_string[1:-2]
# Loads the final mapping of integer node ID to human-readable string
node_id_to_name = {}
for key, val in node_id_to_uid.items():
if val not in uid_to_human:
tf.logging.fatal('Failed to locate: %s', val)
name = uid_to_human[val]
node_id_to_name[key] = name
return node_id_to_name
def id_to_string(self, node_id):
if node_id not in self.node_lookup:
return ''
return self.node_lookup[node_id]
#讀取訓練好的Inception-v3模型來創建graph
def create_graph():
with tf.gfile.FastGFile(os.path.join(
model_dir, 'classify_image_graph_def.pb'), 'rb') as f:
graph_def = tf.GraphDef()
graph_def.ParseFromString(f.read())
tf.import_graph_def(graph_def, name='')
#讀取圖片
image_data = tf.gfile.FastGFile(image, 'rb').read()
#創建graph
create_graph()
sess=tf.Session()
#Inception-v3模型的最后一層softmax的輸出
softmax_tensor= sess.graph.get_tensor_by_name('softmax:0')
#輸入圖像數據，得到softmax概率值（一個shape=(1,1008)的向量）
predictions = sess.run(softmax_tensor,{'DecodeJpeg/contents:0': image_data})
#(1,1008)->(1008,)
predictions = np.squeeze(predictions)
# ID --> English string label.
node_lookup = NodeLookup()
#取出前5個概率最大的值（top-5)
top_5 = predictions.argsort()[-5:][::-1]
for node_id in top_5:
human_string = node_lookup.id_to_string(node_id)
score = predictions[node_id]
print('%s (score = %.5f)' % (human_string, score))
sess.close()
最后輸出：
tiger cat (score = 0.40316)
Egyptian cat (score = 0.21686)
tabby, tabby cat (score = 0.21348)
lynx, catamount (score = 0.01403)
Persian cat (score = 0.00394)
tensorflow中保存模型、加載模型做預測（不需要再定義網絡結構）
旭旭_哥 2017-10-11 23:11:21 20282 收藏 11
展開
下面用一個線下回歸模型來記載保存模型、加載模型做預測
參考文章:
http://blog.csdn.net/thriving_fcl/article/details/71423039
訓練一個線下回歸模型並保存看代碼：
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
money=np.array([[109],[82],[99], [72], [87], [78], [86], [84], [94], [57]]).astype(np.float32)
click=np.array([[11], [8], [8], [6],[ 7], [7], [7], [8], [9], [5]]).astype(np.float32)
x_test=money[0:5].reshape(-1,1)
y_test=click[0:5]
x_train=money[5:].reshape(-1,1)
y_train=click[5:]
x=tf.placeholder(tf.float32,[None,1],name='x') #保存要輸入的格式
w=tf.Variable(tf.zeros([1,1]))
b=tf.Variable(tf.zeros([1]))
y=tf.matmul(x,w)+b
tf.add_to_collection('pred_network', y) #用於加載模型獲取要預測的網絡結構
y_=tf.placeholder(tf.float32,[None,1])
cost=tf.reduce_sum(tf.pow((y-y_),2))
train_step=tf.train.GradientDescentOptimizer(0.000001).minimize(cost)
init=tf.global_variables_initializer()
sess=tf.Session()
sess.run(init)
cost_history=[]
saver = tf.train.Saver()
for i in range(100):
feed={x:x_train,y_:y_train}
sess.run(train_step,feed_dict=feed)
cost_history.append(sess.run(cost,feed_dict=feed))
# 輸出最終的W,b和cost值
print("109的預測值是:",sess.run(y, feed_dict={x: [[109]]}))
print("W_Value: %f" % sess.run(w), "b_Value: %f" % sess.run(b), "cost_Value: %f" % sess.run(cost, feed_dict=feed))
saver_path = saver.save(sess, "/Users/shuubiasahi/Desktop/tensorflow/modelsave/model.ckpt",global_step=100)
print("model saved in file: ", saver_path)
模型訓練結果：
2017-10-11 23:05:12.606557: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.
109的預測值是: [[ 9.84855175]]
保存模型文件分析：
一、 saver.restore()時填的文件名，因為在saver.save的時候，每個checkpoint會保存三個文件，如
model.ckpt-100.meta, model.ckpt-100.index, model.ckpt-100.data-00000-of-00001
在import_meta_graph時填的就是meta文件名，我們知道權值都保存在model.ckpt-100.data-00000-of-00001這個文件中，但是如果在restore方法中填這個文件名，就會報錯，應該填的是前綴，這個前綴可以使用tf.train.latest_checkpoint(checkpoint_dir)這個方法獲取。
二、模型的y中有用到placeholder，在sess.run()的時候肯定要feed對應的數據，因此還要根據具體placeholder的名字，從graph中使用get_operation_by_name方法獲取。
加載訓練好的模型：
import tensorflow as tf
with tf.Session() as sess:
new_saver=tf.train.import_meta_graph('/Users/shuubiasahi/Desktop/tensorflow/modelsave/model.ckpt-100.meta')
new_saver.restore(sess,"/Users/shuubiasahi/Desktop/tensorflow/modelsave/model.ckpt-100")
graph = tf.get_default_graph()
x=graph.get_operation_by_name('x').outputs[0]
y=tf.get_collection("pred_network")[0]
print("109的預測值是:",sess.run(y, feed_dict={x: [[109]]}))
加載模型后的預測結果:
2017-10-11 23:07:33.176523: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.
109的預測值是: [[ 9.84855175]]
Process finished with exit code 0
————————————————
版權聲明：本文為CSDN博主「旭旭_哥」的原創文章，遵循CC 4.0 BY-SA版權協議，轉載請附上原文出處鏈接及本聲明。
原文鏈接：https://blog.csdn.net/luoyexuge/java/article/details/78209670