roboflow/supervision · error · ValueError

Missing '{tag}' in Pascal VOC annotation.

Error message

Missing '{tag}' in Pascal VOC annotation.

What it means

Raised by _get_required_text in the Pascal VOC loader when a required child element (e.g. name within object, or width/height within size) is missing from the XML or has empty text. It is the generic required-field accessor used while parsing VOC XML annotation objects.

Source

Thrown at src/supervision/dataset/formats/pascal_voc.py:398

    return obj.find("polygon") is not None


def parse_polygon_points(polygon: Element) -> npt.NDArray[np.int_]:
    coordinates: list[int] = []
    for coord in polygon.findall(".//*"):
        if coord.text is None:
            raise ValueError("Missing polygon coordinate value in Pascal VOC.")
        coordinates.append(int(coord.text))
    return np.array(
        [(coordinates[i], coordinates[i + 1]) for i in range(0, len(coordinates), 2)],
        dtype=int,
    )


def _get_required_text(element: Element, tag: str) -> str:
    child = element.find(tag)
    if child is None or child.text is None:
        raise ValueError(f"Missing '{tag}' in Pascal VOC annotation.")
    return child.text


def save_pascal_voc_annotations(
    dataset: "DetectionDataset",
    annotations_directory_path: str,
    min_image_area_percentage: float = 0.0,
    max_image_area_percentage: float = 1.0,
    approximation_percentage: float = 0.75,
    show_progress: bool = False,
) -> None:
    """Write Pascal VOC XML annotation files for every image in *dataset*.

    Args:
        dataset: Dataset whose annotations are saved.
        annotations_directory_path: Destination directory for ``.xml`` files;
            created automatically if it does not exist.
        min_image_area_percentage: Minimum detection area as a fraction of the

View on GitHub (pinned to 7f254d9784)

Solutions

1. Open the reported XML file and add the missing element with its text value (e.g. <name>dog</name> inside <object>).
2. If generating VOC XML yourself, use supervision's object_to_pascal_voc / save_pascal_voc_annotations which always emit required fields.
3. Validate files with a quick lxml/ElementTree pass checking for empty required tags before bulk loading.

Example fix

// before
<object><bndbox><xmin>1</xmin>...</bndbox></object>

// after
<object><name>dog</name><bndbox><xmin>1</xmin>...</bndbox></object>

from pathlib import Path
from xml.etree import ElementTree as ET

def validate_voc_xml(file_path: str) -> list[str]:
    """Return list of missing required elements in a VOC XML file."""
    root = ET.parse(file_path).getroot()
    problems = []
    for tag in ("size/width", "size/height", "size/depth"):
        el = root.find(tag)
        if el is None or not (el.text or "").strip():
            problems.append(tag)
    for obj in root.findall("object"):
        name = obj.find("name")
        if name is None or not (name.text or "").strip():
            problems.append(f"object[{obj.findtext('name', default='?')}]/name")
    return problems

def is_complete_voc_object(obj: ET.Element) -> bool:
    """True when an <object> has non-empty name and bndbox coordinates."""
    name = obj.find("name")
    bnd = obj.find("bndbox")
    return (
        name is not None and bool((name.text or "").strip())
        and bnd is not None
        and all((bnd.findtext(t) or "").strip() for t in ("xmin", "ymin", "xmax", "ymax"))
    )

try:
    sv.DetectionDataset.from_pascal_voc(images_directory_path=d, annotations_directory_path=a)
except ValueError as exc:
    if "Missing" in str(exc) and "Pascal VOC" in str(exc):
        problems = validate_voc_xml(str(exc_file))  # identify and repair the file
        raise ValueError(f"Incomplete VOC XML: {problems}") from exc
    raise

When it happens

Trigger: Loading a Pascal VOC XML file via load_pascal_voc_annotations / detections_from_xml_obj where an <object> lacks <name>, or <size> lacks <width>/<height>/<depth>, or any other element fetched through _get_required_text has no text content.

Common situations: Truncated or hand-edited XML; converters that omit <name> for objects; XML files from non-standard VOC dialects; elements written as <name/> self-closing (text is None).

Related errors

• Malformed CreateML annotation entry (missing or non-string '
• CreateML annotation entry is missing the required 'image' ke
• Malformed CreateML annotation entry {annotation}: {exc}
• Failed to parse XML root from {annotation_path}
• Missing bndbox in Pascal VOC annotation.

AI-assisted analysis of roboflow/supervision@7f254d9784 (2026-08-15). Data as JSON: /api/errors/592fc06c761d5583. Report an issue: GitHub.