Merge pull request #9 from shashwat1998/master

Extended tool for subdirectories

README.md

@@ -1,5 +1,6 @@
 # XmlToTxt
 ImageNet file xml format to [Darknet](https://github.com/pjreddie/darknet) text format.
+Works well with directories and subdirectories.
 
 ### Installation
 ```bash
@@ -41,3 +42,4 @@ Output text file.
 ### Motivation
 
 I used [Darknet](https://github.com/pjreddie/darknet) for real-time object detection and classification. Sometimes you need to collect your own training dataset for train your model. I collected training dataset images and fine awesome [tool](https://github.com/tzutalin/labelImg) for labeling images. But it generates xml files. So I needed to implement tool which translates from ImageNet xml format to Darknet text format.
+Also compatible with latest [YOLOv5](https://github.com/ultralytics/yolov5) by Ultralytics.

objectmapper.py

@@ -1,5 +1,5 @@
 import logging
-
+import os
 import declxml as xml
 
 
@@ -25,14 +25,16 @@ def __init__(self):
             xml.string("filename")
         ])
 
-    def bind(self, xml_file_path):
-        return xml.parse_from_file(self.processor, xml_file_path=xml_file_path)
+    def bind(self, xml_file_path, xml_dir):
+        ann = xml.parse_from_file(self.processor, xml_file_path=os.path.join(xml_dir, xml_file_path))
+        ann.filename = xml_file_path
+        return ann
 
-    def bind_files(self, xml_file_paths):
+    def bind_files(self, xml_file_paths, xml_dir):
         result = []
         for xml_file_path in xml_file_paths:
             try:
-                result.append(self.bind(xml_file_path=xml_file_path))
+                result.append(self.bind(xml_file_path=xml_file_path, xml_dir=xml_dir))
             except Exception as e:
                 logging.error("%s", e.args)
         return result

reader.py

@@ -7,9 +7,12 @@ def __init__(self, xml_dir):
 
     def get_xml_files(self):
         xml_filenames = []
-        for xml_filename in os.listdir(self.xml_dir):
-            if xml_filename.endswith(".xml"):
-                xml_filenames.append(os.path.join(self.xml_dir, xml_filename))
+        for root, subdirectories, files in os.walk(self.xml_dir):
+            for filename in files:
+                if filename.endswith(".xml"):
+                    file_path = os.path.join(root, filename)
+                    file_path = os.path.relpath(file_path, start=self.xml_dir)
+                    xml_filenames.append(file_path)    
         return xml_filenames
 
     @staticmethod

transformer.py

@@ -14,12 +14,15 @@ def transform(self):
         xml_files = reader.get_xml_files()
         classes = reader.get_classes()
         object_mapper = ObjectMapper()
-        annotations = object_mapper.bind_files(xml_files)
+        annotations = object_mapper.bind_files(xml_files, xml_dir=self.xml_dir)
         self.write_to_txt(annotations, classes)
 
     def write_to_txt(self, annotations, classes):
         for annotation in annotations:
-            with open(os.path.join(self.out_dir, self.darknet_filename_format(annotation.filename)), "w+") as f:
+            output_path = os.path.join(self.out_dir, self.darknet_filename_format(annotation.filename))
+            if not os.path.exists(os.path.dirname(output_path)):
+                os.makedirs(os.path.dirname(output_path))
+            with open(output_path, "w+") as f:
                 f.write(self.to_darknet_format(annotation, classes))
 
     def to_darknet_format(self, annotation, classes):