queue_runner_B.py

import tensorflow as tf
import matplotlib.pyplot as plt
import os

tf.reset_default_graph()
image_dir = './tfrecord_dataset/images_png/
label_name = './tfrecord_dataset/label_csv/label.csv'
image_names = [os.path.join(image_dir, n) for n in os.listdir(image_dir)]

img_name_queue = tf.train.string_input_producer(image_names, seed=7777)
label_name_queue = tf.train.string_input_producer([label_name], seed=7777)

img_reader = tf.WholeFileReader()
tf.TextLineReader()

key, value = img_reader.read(img_name_queue)

img_png = tf.image.decode_png(value)
img_png = tf.reduce_mean(img_png, axis=-1)

with tf.Session() as sess:
    coord = tf.train.Coordinator()
    thread = tf.train.start_queue_runners(sess, coord)
    _img, _label = sess.run([img_png, label_csv])
    print(_label)
    print(_img.shape)
    plt.imshow(_img)
    plt.show()
    coord.request_stop()
    coord.join(thread)

reading_tfrecord.py

import tensorflow as tf
import matplotlib.pyplot as plt


tf.reset_default_graph()

def parser(serialized_example):
    feature = {
        'age': tf.FixedLenFeature([1], tf.int64),
        'img': tf.FixedLenFeature([61*49], tf.int64)
    }
    parsed_feature = tf.parse_single_example(serialized_example, feature)
    age = tf.cast(parsed_feature['age'], tf.int32)
    img = tf.cast(parsed_feature['img'], tf.float32)
    return age, img
    

dataset_dir = './cnn_dataset/face_train.tfrecord'
dataset = tf.contrib.data.TFRecordDataset(dataset_dir).map(parser)
dataset = dataset.batch(32)
dataset = dataset.shuffle(7777)

itr = dataset.make_one_shot_iterator()

age, img = itr.get_next()
img = tf.reshape(img, [-1, 61, 49])

with tf.Session() as sess:
    _age, _img = sess.run([age, img])
    for i in range(32):
        print(_age[i])
        plt.imshow(_img[i])
        plt.show()

tfrecord_writer.py

import tensorflow as tf
import matplotlib.pyplot as plt
import os

tf.reset_default_graph()
image_dir = './tfrecord_dataset/images_png/'
label_name = './tfrecord_dataset/label_csv/label.csv'
image_names = [os.path.join(image_dir, n) for n in os.listdir(image_dir)]
image_names = sorted(image_names)

img_name_queue = tf.train.string_input_producer(image_names, num_epochs=1, shuffle=False)
label_name_queue = tf.train.string_input_producer([label_name], num_epochs=1, shuffle=False)

img_reader = tf.WholeFileReader()
text_reader = tf.TextLineReader()

img_key, img_value = img_reader.read(img_name_queue)
txt_key, txt_value = text_reader.read(label_name_queue)

img_png = tf.image.decode_png(img_value)
img_png = tf.reduce_mean(img_png, axis=-1)
img_png = tf.reshape(img_png, [-1])

txt_csv = tf.decode_csv(txt_value, record_defaults=[[0]])

with tf.Session() as sess:
    sess.run(tf.local_variables_initializer())
    coord = tf.train.Coordinator()
    thread = tf.train.start_queue_runners(sess, coord)

    face_train_dir = './cnn_dataset/face_train.tfrecord'
    face_test_dir = './cnn_dataset/face_test.tfrecord'
    
    train_writer = tf.python_io.TFRecordWriter(face_train_dir)
    
    for i in range(9999999999999):
        try:
            _img, _age = sess.run([img_png, txt_csv])

            example = tf.train.Example()
            example.features.feature['age'].int64_list.value.append(_age[0])
            example.features.feature['img'].int64_list.value.extend(_img.tolist())
            train_writer.write(example.SerializeToString())
            print('{}th record is written'.format(i))
            
        except tf.errors.OutOfRangeError:
            print('end of record')
            break
            
    train_writer.close()

    coord.request_stop()
    coord.join(thread)