tensorflow製作tfrecord格式資料集

#encoding=utf-8

import os
import tensorflow as tf
from pil import image
cwd = os.getcwd()
classes = 
#製作二進位制資料
defcreate_record
():    writer = tf.python_io.tfrecordwriter("train.tfrecords")
for index, name in enumerate(classes):
class_path = cwd +"/"+ name+"/"
for img_name in os.listdir(class_path):
img_path = class_path + img_name
img = image.open(img_path)
img = img.resize((64, 64))
img_raw = img.tobytes() #將轉化為原生bytes
print index,img_raw
example = tf.train.example(
features=tf.train.features(feature=))
writer.write(example.serializetostring())
writer.close()
data = create_record()
#讀取二進位制資料
defread_and_decode
(filename):
# 建立檔案佇列,不限讀取的數量
filename_queue = tf.train.string_input_producer([filename])
# create a reader from file queue
reader = tf.tfrecordreader()
# reader從檔案佇列中讀入乙個序列化的樣本
_, serialized_example = reader.read(filename_queue)
# get feature from serialized example
# 解析符號化的樣本
features = tf.parse_single_example(
serialized_example,
features=
)label = features['label']
img = features['img_raw']
img = tf.decode_raw(img, tf.uint8)
img = tf.reshape(img, [64, 64, 3])
img = tf.cast(img, tf.float32) * (1. / 255) - 0.5
label = tf.cast(label, tf.int32)
return img, label
if __name__ == '__main__':
if0:        data = create_record("train.tfrecords")
else:
img, label = read_and_decode("train.tfrecords")
print
"tengxing",img,label
#使用shuffle_batch可以隨機打亂輸入 next_batch挨著往下取
# shuffle_batch才能實現[img,label]的同步,也即特徵和label的同步,不然可能輸入的特徵和label不匹配
# 比如只有這樣使用,才能使img和label一一對應,每次提取乙個image和對應的label
# shuffle_batch返回的值就是randomshufflequeue.dequeue_many()的結果
# shuffle_batch構建了乙個randomshufflequeue，並不斷地把單個的[img,label],送入佇列中
img_batch, label_batch = tf.train.shuffle_batch([img, label],
batch_size=4, capacity=2000,
min_after_dequeue=1000)
# 初始化所有的op
init = tf.initialize_all_variables()
with tf.session() as sess:
sess.run(init)
# 啟動佇列
threads = tf.train.start_queue_runners(sess=sess)
for i in range(5):
print img_batch.shape,label_batch
val, l = sess.run([img_batch, label_batch])
# l = to_categorical(l, 12)
print(val.shape, l)

#製作二進位制資料

defcreate_record
():    cwd = os.getcwd()
classes = 
writer = tf.python_io.tfrecordwriter("train.tfrecords")
for index, name in enumerate(classes):
class_path = cwd +"/"+ name+"/"
for img_name in os.listdir(class_path):
img_path = class_path + img_name
img = image.open(img_path)
img = img.resize((28, 28))
img_raw = img.tobytes() #將轉化為原生bytes
#print index,img_raw
example = tf.train.example(
features=tf.train.features(
feature=))
writer.write(example.serializetostring())
writer.close()

#讀取二進位制資料

defread_and_decode
(filename):
# 建立檔案佇列,不限讀取的數量
filename_queue = tf.train.string_input_producer([filename])
# create a reader from file queue
reader = tf.tfrecordreader()
# reader從檔案佇列中讀入乙個序列化的樣本
_, serialized_example = reader.read(filename_queue)
# get feature from serialized example
# 解析符號化的樣本
features = tf.parse_single_example(
serialized_example,
features=
)label = features['label']
img = features['img_raw']
img = tf.decode_raw(img, tf.uint8)
img = tf.reshape(img, [64, 64, 3])
img = tf.cast(img, tf.float32) * (1. / 255) - 0.5
label = tf.cast(label, tf.int32)
return img, label

with tf.session() as sess:

sess.run(init)
# 啟動佇列
threads = tf.train
.start_queue_runners(sess=sess)
for i in range(5):
print img_batch.shape,label_batch
val, l = sess.run([img_batch, label_batch])
# l = to_categorical(l, 12)
print(val.shape, l)

算交叉熵時候：cross_entropy=tf.nn.sparse_softmax_cross_entropy_with_logits(logits,labels)算交叉熵 

最後評估的時候用tf.nn.in_top_k(logits,labels,1)選logits最大的數的索引和label比較

cross_entropy = -tf.reduce_sum(y_*tf.log(y_conv))算交叉熵，所以label必須轉成one-hot向量

tensorflow高效讀取資料之tfrecord

tensorflow提供了一種統一的格式來儲存資料，這個格式就是tfrecord，接下來介紹如何使用tfrecord來統一輸入資料的格式。tfrcord檔案中的資料都是通過tf.train.example protocol buffer的格式來儲存的，以下 給出了tf.train.example的定...

Tensorflow中製作tfrecord資料集

tensorflow提供統一的格式來儲存資料，tfrecord。使得在之後的系統中可以更加方便的處理，實際問題的資料往往有很多格式和屬性，用tfrecord格式可以統一不同原始資料格式，並且更加有效的管理不同屬性。import tensorflow as tf import os import cv...

php mysql製作 php MYSQL製作分頁

header content type text html charset utf 8 傳入頁碼 p get p 每頁數量 pagesize 1 鏈結資料庫 host localhost user root pass db test conn mysqli connect host,user,pas...