def _get_output_data()

in ocr/utils/iam_dataset.py [0:0]

• 33 lines of code
• 7 McCabe index (conditional complexity)

    def _get_output_data(self, item, height, width):
        ''' Function to obtain the output data (both text and bounding boxes).
        Note that the bounding boxes are rescaled based on the rescale_ratio parameter.

        Parameter
        ---------
        item: xml.etree 
            XML object for a word/line/form.

        height: int
            Height of the form to calculate percentages of bounding boxes

        width: int
            Width of the form to calculate percentages of bounding boxes

        Returns
        -------

        np.array
            A numpy array ouf the output requested (text or the bounding box)
        '''

        output_data = []
        if self._output_data == "text":
            if self._parse_method in ["form", "form_bb", "form_original"]:
                text = ""
                for line in item.iter('line'):
                    text += line.attrib["text"] + "\n"
                output_data.append(text)
            else:
                output_data.append(item.attrib['text'])
        else:
            for item_output in item.iter(self._output_parse_method):
                bb = self._get_bb_of_item(item_output, height, width)
                if bb == None: # Account for words with no letters
                    continue
                output_data.append(bb)
        output_data = np.array(output_data)
        return output_data