input.py

import tensorflow as tf
import os

IMAGE_SIZE = 24
NUM_CLASSES = 10
NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN = 50000
NUM_EXAMPLES_PER_EPOCH_FOR_TEST = 10000


def read_cifar10(filename_queue):
    class CIFAR10Record(object):
        pass
    result = CIFAR10Record()

    label_bytes = 1
    result.height = 32
    result.width = 32
    result.depth = 3
    image_bytes = result.height * result.width * result.depth
    record_bytes = label_bytes + image_bytes

    reader = tf.FixedLengthRecordReader(record_bytes)
    result.key, value = reader.read(filename_queue)

    record_bytes = tf.decode_raw(value, tf.uint8)
    result.label = tf.cast(tf.strided_slice(record_bytes, [0], [label_bytes]), tf.int32)
    depth_major = tf.reshape(tf.strided_slice(record_bytes, [label_bytes], [label_bytes + image_bytes]),
                             [result.depth, result.height, result.width])
    result.uint8image = tf.transpose(depth_major, [1, 2, 0])

    return result


def _generate_image_and_label_batch(image, label, min_queue_example, batch_size, shuffle):
    num_preprocess_threads = 16
    if shuffle:
        images, label_batch = tf.train.shuffle_batch([image, label],
                                                     batch_size=batch_size,
                                                     num_threads=num_preprocess_threads,
                                                     capacity=min_queue_example + 3 * batch_size,
                                                     min_after_dequeue=min_queue_example)
    else:
        images, label_batch = tf.train.batch([image, label],
                                             batch_size=batch_size,
                                             num_threads=num_preprocess_threads,
                                             capacity=min_queue_example + 3 * batch_size)
    tf.summary.image('images', images)
    return images, tf.reshape(label_batch, [batch_size])


def distorted_inputs(data_dir, batch_size):
    filenames = [os.path.join(data_dir, 'data_batch_%d.bin' % i) for i in range(1, 6)]
    for f in filenames:
        if not tf.gfile.Exists(f):
            raise ValueError('Failed to find file:' + f)

    filename_queue = tf.train.string_input_producer(filenames)
    with tf.name_scope('data_augmentation'):
        read_input = read_cifar10(filename_queue)
        reshaped_image = tf.cast(read_input.uint8image, tf.float32)

        height = IMAGE_SIZE
        width = IMAGE_SIZE

        distorted_images = tf.random_crop(reshaped_image, [height, width, 3])
        distorted_images = tf.image.random_flip_left_right(distorted_images)
        distorted_images = tf.image.random_brightness(distorted_images, max_delta=63)
        distorted_images = tf.image.random_contrast(distorted_images, lower=0.2, upper=1.8)
        float_image = tf.image.per_image_standardization(distorted_images)

        float_image.set_shape([height, width, 3])
        read_input.label.set_shape([1])

        min_fraction_of_example_in_queue = 0.4
        min_queue_examples = int(NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN *
                                 min_fraction_of_example_in_queue)

    return _generate_image_and_label_batch(float_image, read_input.label, min_queue_examples,
                                           batch_size, shuffle=True)


def inputs(eval_data, data_dir, batch_size):
    if not eval_data:
        filenames = [os.path.join(data_dir, 'data_batch_%d.bin' % i) for i in range(1, 6)]
        num_examples_per_epoch = NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN
    else:
        filenames = [os.path.join(data_dir, 'test_batch.bin')]
        num_examples_per_epoch = NUM_EXAMPLES_PER_EPOCH_FOR_TEST

    for f in filenames:
        if not tf.gfile.Exists(f):
            raise ValueError('Failed to find file： ' + f)

    with tf.name_scope('input'):
        filename_queue = tf.train.string_input_producer(filenames)

        read_input = read_cifar10(filename_queue)
        reshaped_image = tf.cast(read_input.uint8image, tf.float32)

        height = IMAGE_SIZE
        width = IMAGE_SIZE

        resized_image = tf.image.resize_image_with_crop_or_pad(reshaped_image, height, width)
        float_image = tf.image.per_image_standardization(resized_image)

        float_image.set_shape([height, width, 3])
        read_input.label.set_shape([1])

        min_fraction_of_examples_in_queue = 0.4
        min_queue_examples = int(num_examples_per_epoch * min_fraction_of_examples_in_queue)

    return _generate_image_and_label_batch(float_image, read_input.label,
                                           min_queue_examples, batch_size, shuffle=False)