Converts CVAT annotation samples from *checkbox tags* to *object tag attributes*

label2attr.py

import bs4

source_path = "annotations_source.xml"
output_xml_path = "annotations_out.xml"

LABELS = [
    "no-category", "valid", "non-centered", "occluded", "spoof",
    "rotated", "blurred"]


print("Loading XML")
with open(source_path) as fp:
    xml_obj = bs4.BeautifulSoup(
        fp.read(), features="lxml")

print("Editing annotations")
invalid_annotations = set()
for image in xml_obj.select("image"):
    # new CVAT attribute from scratch
    new_box = bs4.Tag(name="box")
    for k, v in image.find("box").attrs.items():
        new_box[k] = v

    # tag -> attribute reformat
    tags = set(map(
        lambda tag: tag["label"], image.find_all("tag", recursive=False)))
    for label in LABELS:
        label_tag = bs4.Tag(name="attribute", attrs={"name": label})
        label_tag.string = "true" if label in tags else "false"
        new_box.contents.append(label_tag)

    # erases other boxes (to avoid duplicate) and insert new
    [ box.decompose() for box in image.find_all("box") ]
    image.contents.append(new_box)

print("Outputting fixed annotations")
with open(output_xml_path, "w") as fp:
    xml_text = str(xml_obj)
    xml_text = xml_text.replace("&lt;", "<").replace("&gt;", ">")
    fp.write(xml_text)