Where can I find/generate the tfrecords for training?

[GalDude33]

Where can I find/generate the tfrecords for training?

The tfrecords aren't provided.

[sugyan]

Sorry, I don't provide data file for training. You need to create it yourself...

If you tell me good dataset of face images, I'll add a script to create tfrecords files based on that dataset.

[ayuLiao]

I use my tfrecords, but is not work. Can you provide program about create tfrecords？

This is my error when I use my tfrecords : tensorflow.python.framework.errors_impl.InvalidArgumentError: Expected image (JPEG, PNG, or GIF), got unknown format starting with '\212\211\t?\204\203\003?\206\205\005?\330\326\326>'

This is my code to create tfrecords

import os
import matplotlib.image as mpimg
import tensorflow as tf
from PIL import Image

SAVE_PATH = r'D:\LearnPython\DCGAN\face-generator-master\data/dataset.tfrecords'
FACE_PATH = r'D:\LearnPython\DCGAN\face-generator-master\data\face2'

def resize(img_data, width, high, method=0):
    return tf.image.resize_images(img_data, [width, high], method)

def _int64_feature(value):
    return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))

def _bytes_feature(value):
    return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))

def load_data(width, high, method=0, save=False):
    train_list = os.listdir(FACE_PATH)
    writer = tf.python_io.TFRecordWriter(SAVE_PATH)
    with tf.Session() as sess:
        for line in train_list:
            img_path = FACE_PATH+'/'+line
            image = tf.gfile.FastGFile(img_path, 'rb').read()
            image = tf.image.decode_jpeg(image)
            image = tf.image.convert_image_dtype(image, dtype=tf.float32)
            image = resize(image, width, high)
            image = sess.run(image)
            image_raw = image.tostring()
            example = tf.train.Example(features=tf.train.Features(feature={
                'image_raw': _bytes_feature(image_raw),
            }))
            writer.write(example.SerializeToString())
    writer.close()

load_data(224, 224)

I am a novice, thank you.

[sugyan]

@ayuLiao Maybe you should encode to jpeg binary after resize. My program try to decode tfrecord's image_raw data as jpeg binary.