generate_tf_records.py

"""
It generates the train and tf records that are set by dataset_costants.py file.
"""

from __future__ import division
from __future__ import print_function
from __future__ import absolute_import

from dataset_costants import TABLE_DICT

import os
import io
import pandas as pd
import tensorflow as tf

from PIL import Image
from object_detection.utils import dataset_util
from collections import namedtuple
from dataset_costants import \
    TF_TRAIN_RECORD_TO_PATH, \
    TF_TRAIN_RECORD_NAME, \
    TF_TEST_RECORD_TO_PATH, \
    TF_TEST_RECORD_NAME, \
    PATH_TO_IMAGES, \
    TRAIN_CSV_TO_PATH, \
    TEST_CSV_TO_PATH, \
    TRAIN_CSV_NAME, \
    TEST_CSV_NAME

import logging
from logger import TimeHandler

logger = logging.getLogger(__name__)
logger.setLevel(logging.INFO)
logger.addHandler(TimeHandler().handler)


def class_text_to_int(row_label):
    """
    Replace text with a int. Zero is for showing no boxes at all
    :param row_label:
    :return: returns corresponding int
    """
    return 1 if row_label == TABLE_DICT['name'] else 0


def split(df, group):
    """
    Splits name, sys function
    :param df:
    :param group:
    :return:
    """
    data = namedtuple('data', ['filename', 'object'])
    gb = df.groupby(group)
    return [data(filename, gb.get_group(x)) for filename, x in zip(gb.groups.keys(), gb.groups)]


def create_tf_example(group, path):
    """
    Creates the effective tf example given a group of data and a path.
    :param group: all the information from csv of a single image
    :param path: to images
    :return: a tf example (a single image tensorflow image)
    """
    # looking for filename image of csv in path folder
    logger.debug('Now creating a single-page tf_example...')
    with tf.gfile.GFile(os.path.join(path, '{}'.format(group.filename)), 'rb') as fid:
        encoded_jpeg = fid.read()
        logger.debug('Image found in {}'.format(path))
    # encode it as a pillow image
    encoded_jpeg_io = io.BytesIO(encoded_jpeg)
    image = Image.open(encoded_jpeg_io)
    width, height = image.size

    filename = group.filename.encode('utf8')
    image_format = b'jpeg'
    xmins = []
    xmaxs = []
    ymins = []
    ymaxs = []
    classes_text = []  # class text that are written in costants.TABLE_DICT. In our case we have a single table lable
    classes = []  # class number

    # it now append the coordinates of the boxes inside csv file per image
    logger.debug('Appending csv {fn} infos to page tf_example...'.format(fn=filename))
    for index, row in group.object.iterrows():
        xmins.append(row['xmin'] / width)
        xmaxs.append(row['xmax'] / width)
        ymins.append(row['ymin'] / height)
        ymaxs.append(row['ymax'] / height)
        classes_text.append(row['class'].encode('utf8'))
        classes.append(class_text_to_int(row['class']))

    # create single tf_example object
    tf_example = tf.train.Example(features=tf.train.Features(feature={
        'image/height': dataset_util.int64_feature(height),
        'image/width': dataset_util.int64_feature(width),
        'image/filename': dataset_util.bytes_feature(filename),
        'image/source_id': dataset_util.bytes_feature(filename),
        'image/encoded': dataset_util.bytes_feature(encoded_jpeg),
        'image/format': dataset_util.bytes_feature(image_format),
        'image/object/bbox/xmin': dataset_util.float_list_feature(xmins),
        'image/object/bbox/xmax': dataset_util.float_list_feature(xmaxs),
        'image/object/bbox/ymin': dataset_util.float_list_feature(ymins),
        'image/object/bbox/ymax': dataset_util.float_list_feature(ymaxs),
        'image/object/class/text': dataset_util.bytes_list_feature(classes_text),
        'image/object/class/label': dataset_util.int64_list_feature(classes),
    }))
    logger.debug('Successfully created single-page tf_example')
    return tf_example


def main(_):
    # write train record:
    csv_input = os.path.join(TRAIN_CSV_TO_PATH, TRAIN_CSV_NAME)
    output_path = os.path.join(TF_TRAIN_RECORD_TO_PATH, TF_TRAIN_RECORD_NAME)
    writer = tf.python_io.TFRecordWriter(output_path)
    path = os.path.join(os.getcwd(), PATH_TO_IMAGES)
    # examples is the csv file
    examples = pd.read_csv(csv_input)
    grouped = split(examples, 'filename')
    logger.info('Now creating {}:'.format(TRAIN_CSV_NAME))
    for group in grouped:
        tf_example = create_tf_example(group, path)
        # append tf_example to the writer
        writer.write(tf_example.SerializeToString())
    writer.close()
    # output_path = os.path.join(os.getcwd(), TF_TRAIN_RECORD_TO_PATH)
    logger.info('Successfully created the TF train records: {}'.format(output_path))

    # write test record:
    csv_input = os.path.join(TEST_CSV_TO_PATH, TEST_CSV_NAME)
    output_path = os.path.join(TF_TEST_RECORD_TO_PATH, TF_TEST_RECORD_NAME)
    writer = tf.python_io.TFRecordWriter(output_path)
    path = os.path.join(os.getcwd(), PATH_TO_IMAGES)
    examples = pd.read_csv(csv_input)
    grouped = split(examples, 'filename')
    logger.info('Now creating {}:'.format(TEST_CSV_NAME))
    for group in grouped:
        tf_example = create_tf_example(group, path)
        writer.write(tf_example.SerializeToString())
    writer.close()
    # output_path = os.path.join(os.getcwd(), TF_TEST_RECORD_TO_PATH)
    logger.info('Successfully created the TF test record: {}'.format(output_path))


if __name__ == '__main__':
    tf.app.run()