page alto label generation activated for textline

src/eynollah/training/extract_line_gt.py

@@ -92,7 +92,7 @@ def linegt_cli(
             tree = ET.parse(dir_xml)
             root = tree.getroot()
 
-            NS = {"alto": "http://www.loc.gov/standards/alto/ns-v4#"}
+            NS = {'alto': root.tag.split('}')[0].strip('{')}#{"alto": "http://www.loc.gov/standards/alto/ns-v4#"}
 
             results = []
             

src/eynollah/training/generate_gt_for_training.py

@@ -73,8 +73,14 @@ def main():
     is_flag=True,
     help="if this parameter set to true, generated labels and in the case of provided org images cropping will be imposed and cropped labels and images will be written in output directories.",
 )
+@click.option(
+    "--page_alto",
+    "-alto",
+    is_flag=True,
+    help="If this parameter is set to True, textline label generation is performed using PAGE/ALTO files. Otherwise, the default method for PAGE XML files is used.",
+)
 
-def pagexml2label(dir_xml,dir_out,type_output,config, printspace, dir_images, dir_out_images):
+def pagexml2label(dir_xml,dir_out,type_output,config, printspace, dir_images, dir_out_images, page_alto):
     if config:
         with open(config) as f:
             config_params = json.load(f)
@@ -82,7 +88,7 @@ def pagexml2label(dir_xml,dir_out,type_output,config, printspace, dir_images, di
         print("passed")
         config_params = None
     gt_list = get_content_of_dir(dir_xml)
-    get_images_of_ground_truth(gt_list,dir_xml,dir_out,type_output, config, config_params, printspace, dir_images, dir_out_images)
+    get_images_of_ground_truth(gt_list,dir_xml,dir_out,type_output, config, config_params, printspace, dir_images, dir_out_images, page_alto)
     
 @main.command()
 @click.option(

src/eynollah/training/gt_gen_utils.py

@@ -686,7 +686,7 @@ def get_layout_contours_for_visualization(xml_file):
                 co_noise.append(np.array(c_t_in))
     return co_text, co_graphic, co_sep, co_img, co_table, co_map, co_music, co_noise, y_len, x_len
     
-def get_images_of_ground_truth(gt_list, dir_in, output_dir, output_type, config_file, config_params, printspace, dir_images, dir_out_images):
+def get_images_of_ground_truth(gt_list, dir_in, output_dir, output_type, config_file, config_params, printspace, dir_images, dir_out_images, page_alto=False):
     """
     Reading the page xml files and write the ground truth images into given output directory.
     """
@@ -696,8 +696,22 @@ def get_images_of_ground_truth(gt_list, dir_in, output_dir, output_type, config_
         ls_org_imgs = os.listdir(dir_images)
         ls_org_imgs_stem = [os.path.splitext(item)[0] for item in ls_org_imgs]
     for index in tqdm(range(len(gt_list))):
-        #try:
         print(gt_list[index])
+        
+        try:
+            if page_alto:
+                tree = ET.parse(dir_in+'/'+gt_list[index])
+                root = tree.getroot()
+
+                NS = {'alto': root.tag.split('}')[0].strip('{')}#{"alto": "http://www.loc.gov/standards/alto/ns-v4#"}
+                x_len, y_len = 0, 0
+                
+                page = root.find('.//alto:Page', NS)
+                
+                x_len = int( page.get("WIDTH") )
+                y_len = int( page.get("HEIGHT") )
+                
+            else:
                 tree1 = ET.parse(dir_in+'/'+gt_list[index], parser = ET.XMLParser(encoding='utf-8'))
                 root1=tree1.getroot()
                 alltags=[elem.tag for elem in root1.iter()]
@@ -783,6 +797,24 @@ def get_images_of_ground_truth(gt_list, dir_in, output_dir, output_type, config_
                     
                 textline_rgb_color = (255, 0, 0)
                     
+                if page_alto:
+                    co_use_case = []
+                    for line in root.findall(".//alto:TextLine", NS):
+                        string_el = line.find("alto:String", NS)
+                        textline_text = string_el.attrib["CONTENT"] if string_el is not None else None
+
+                        polygon_el = line.find("alto:Shape/alto:Polygon", NS)
+                        if polygon_el is None:
+                            continue
+
+                        points = polygon_el.attrib["POINTS"].split()
+                        coords = [
+                            (int(points[i]), int(points[i + 1]))
+                            for i in range(0, len(points), 2)
+                        ]
+                        
+                        co_use_case.append( np.array(coords, dtype=np.int32) )
+                else:
                     if config_params['use_case']=='textline':
                         region_tags = np.unique([x for x in alltags if x.endswith('TextLine')])
                     elif config_params['use_case']=='word':
@@ -879,6 +911,8 @@ def get_images_of_ground_truth(gt_list, dir_in, output_dir, output_type, config_
                         
                     cv2.imwrite(os.path.join(dir_out_images, org_image_name), img_org)
 
+        except:
+            pass
         
         if config_file and config_params['use_case']=='layout':
             keys = list(config_params.keys())