SPPAS 4.20

Module sppas.src.imgdata

Class sppasImageSightsReader

Description

Read&create sights from a CSV/XRA file.

The CSV file must have the following columns:

  • 1
  • index of the sights in the image;
  • success -- like in OpenFace2 CSV results
  • number of sights
  • all x values
  • all y values
  • all z values -- if not None
  • image name

Constructor

Set the list of sights defined in the given file.

Parameters
  • input_file: (str) Coords from a sppasCoordsImageWriter
  • csv_separator: (char) Optional parameter, ';' character by default Columns separator in the CSV file
Raises
  • sppasTypeError: If one of the two parameters are not a string object
  • sppasIOError: If the input file is not found
  • sppasExtensionReadError: If the extension of the input file is not 'xra' or 'csv'
View Source 
def __init__(self, input_file, csv_separator=';'):
    """Set the list of sights defined in the given file.

    :param input_file: (str) Coords from a sppasCoordsImageWriter
    :param csv_separator: (char) Optional parameter, ';' character by default
                                 Columns separator in the CSV file

    :raises sppasTypeError: If one of the two parameters are not a string object
    :raises sppasIOError: If the input file is not found
    :raises sppasExtensionReadError: If the extension of the input file is not 'xra' or 'csv'

    """
    if not isinstance(input_file, str):
        raise sppasTypeError(input_file, 'str')
    if not os.path.exists(input_file):
        raise sppasIOError(input_file)
    self.sights = list()
    _, file_extension = os.path.splitext(input_file)
    if file_extension.lower() == '.csv':
        if not isinstance(csv_separator, str):
            raise sppasTypeError(csv_separator, 'str')
        self.__load_from_csv(input_file, csv_separator)
    elif file_extension.lower() == '.xra':
        self.__load_from_xra(input_file)
    else:
        raise sppasExtensionReadError('Unrecognized extension, expected .csv or .xra.Got {0} instead.'.format(file_extension))

Protected functions

__load_from_csv

Initialize all sights data from the csv file data loaded.

Parameters
  • input_file: (str) Coords from a sppasCoordsImageWriter
  • separator: (char) Columns separator in the CSV file
Raises
  • sppasTypeError: If the csv has a bad value in this data
View Source 
def __load_from_csv(self, input_file, separator):
    """Initialize all sights data from the csv file data loaded.

        :param input_file: (str) Coords from a sppasCoordsImageWriter
        :param separator: (char) Columns separator in the CSV file

        :raises sppasTypeError: If the csv has a bad value in this data

        """
    with codecs.open(input_file, 'r') as csv:
        lines = csv.readlines()
    if len(lines) == 0:
        return
    for line in lines:
        content = line.split(separator)
        if content[2] == '1':
            nb_sights = sppasCoords.to_dtype(content[3])
            current_sights = sppasSights(nb_sights)
            for i in range(3, 3 + nb_sights):
                x = sppasCoords.to_dtype(content[i - 3], unsigned=False)
                y = sppasCoords.to_dtype(content[i - 3 + nb_sights], unsigned=False)
                current_sights.set_sight(i - 3, x, y)
            self.sights.append(current_sights)
__load_from_xra

Initialize all sights data from the xra file data loaded.

Parameters
  • input_file: (str) Coords from a sppasCoordsImageWriter
Raises
  • sppasError: If the annotation file doesn't contain any sights tier.
View Source 
def __load_from_xra(self, input_file):
    """Initialize all sights data from the xra file data loaded.

        :param input_file: (str) Coords from a sppasCoordsImageWriter

        :raises sppasError: If the annotation file doesn't contain any sights tier.

        """
    transcription = sppasXRA('HandImageSights')
    transcription.read(input_file)
    tier = None
    for current_tier in transcription:
        if current_tier.is_point() is True and current_tier.is_fuzzypoint() is True:
            tier = current_tier
            break
    if tier is None:
        raise sppasError('No valid tier in XRA. Cant load sights.')
    for annotation in tier:
        current_sights = sppasSights(nb=len(annotation.get_labels()))
        for label in annotation.get_labels():
            tag = label.get_best()
            score = label.get_score(tag)
            fuzzy_point = tag.get_typed_content()
            x, y = fuzzy_point.get_midpoint()
            sight_index = label.get_key()[-2:]
            try:
                sight_index = int(sight_index)
                current_sights.set_sight(sight_index, x, y, z=None, score=score)
            except ValueError:
                logging.error('The key of the sight is incorrect : ' + sight_index)
        self.sights.append(current_sights)