xml_parse

xml_parse.py

res_dim = 1024

if __name__ == "__main__":
    """loading the data, 
    reading the file annotations,
    appending the tabular coordinates to formulate a dataframe
    """
    df_org = pd.DataFrame()  
    directory = '/content/drive/MyDrive/data_cs2'
    final_col_directory = '/content/drive/MyDrive/cs2_col'
    final_table_directory = '/content/drive/MyDrive/cs2_table'

    for file in os.listdir(directory):
        filename = os.fsdecode(file)        
        if filename.endswith(".xml"):
            filename = filename[:-4]
            #https://docs.python.org/3/library/xml.etree.elementtree.html            
            tree = ET.parse(directory +'//'+ filename + '.xml')
            root = tree.getroot()            
            f_name = root.find('./filename').text          
            depth = root.find('./size/depth').text
            height = root.find('./size/height').text
            width = root.find('./size/width').text
            
            # Empty lists to append the coordinates as shown in the previous sample case
            xmin=[]
            ymin=[]
            xmax=[]
            ymax=[]
            for i in root.findall('object'):
                bndbox = i.find('bndbox')
                xmin_ = int(bndbox.find('xmin').text)
                xmin.append(xmin_)
                xmax_ = int(bndbox.find('xmax').text)
                xmax.append(xmax_)
                ymin_ = int(bndbox.find('ymin').text) 
                ymin.append(ymin_)               
                ymax_ = int(bndbox.find('ymax').text)
                ymax.append(ymax_)