roboflow/supervision · error · ValueError

LabelMe shape of type {shape_type} (label={label}) has malfo

Error message

LabelMe shape of type {shape_type} (label={label}) has malformed points: expected an (N, 2) array, got shape {points.shape}.

What it means

Raised when a LabelMe shape's points cannot be interpreted as an (N, 2) numeric array — np.array(points_raw, dtype=np.float32) yields a wrong ndim or second dimension. LabelMe stores points as a list of [x, y] pairs; a flat list of numbers, a list of triples, or ragged data fails this structural check with the actual shape reported.

Source

Thrown at src/supervision/dataset/formats/labelme.py:112

    polygons: list[npt.NDArray[np.float32]] = []
    skipped_types: set[str] = set()

    for shape in shapes:
        shape_type = shape.get("shape_type")
        if shape_type not in SUPPORTED_SHAPE_TYPES:
            skipped_types.add(str(shape_type))
            continue
        label = shape.get("label")
        points_raw = shape.get("points")
        if label is None or points_raw is None:
            missing = "label" if label is None else "points"
            raise ValueError(
                f"LabelMe shape of type {shape_type!r} is missing the "
                f"required {missing!r} field."
            )
        points = np.array(points_raw, dtype=np.float32)
        if points.ndim != 2 or points.shape[1] != 2:
            raise ValueError(
                f"LabelMe shape of type {shape_type!r} (label={label!r}) has "
                f"malformed points: expected an (N, 2) array, got shape "
                f"{points.shape}."
            )
        if shape_type == "rectangle":
            if len(points) < 2:
                raise ValueError(
                    f"LabelMe rectangle shape (label={label!r}) has "
                    f"{len(points)} point(s); expected at least 2."
                )
            xyxy = _rectangle_to_xyxy(points)
            polygon = _xyxy_to_polygon(xyxy)
        else:
            if len(points) < 3:
                raise ValueError(
                    f"LabelMe polygon shape (label={label!r}) has "
                    f"{len(points)} point(s); expected at least 3."
                )

View on GitHub (pinned to 7f254d9784)

Solutions

Look at the reported shape in the error: (N,) means flat pairs -> reshape to [[x, y], ...]; (N, 3) means extra columns -> drop them.
Fix the generator/converter to emit a list of [x, y] pairs.
If only a few shapes are bad, repair or delete them in the JSON directly.

Example fix

// before
"points": [10, 20, 30, 40]
// after
"points": [[10, 20], [30, 40]]

Defensive patterns

Strategy: validation

Validate before calling

def points_are_pairs(points: object) -> bool:
    """LabelMe points must be a list of [x, y] pairs."""
    return (isinstance(points, list) and len(points) > 0
            and all(isinstance(p, (list, tuple)) and len(p) == 2 for p in points))

Try / catch

try:
    dataset = sv.DetectionDataset.from_labelme(images_dir, ann_dir)
except ValueError as e:
    if 'malformed points' in str(e):
        raise SystemExit(f'Reshape points to [[x, y], ...] in the named file: {e}') from e
    raise

Prevention

Emit points as nested [x, y] lists from converters, never flat arrays.
Check for extra columns when converting from 3D tools.
Validate a sample of generated JSON with a schema check before bulk runs.

When it happens

Trigger: DetectionDataset.from_labelme where points is e.g. [1, 2, 3, 4] (flat), [[x, y, z]] (triples), or contains non-numeric entries causing an unexpected array shape after coercion.

Common situations: Scripts that emit flattened coordinates to save space; unit conversion code that flattens pairs; copy-paste from CSV where each point became a single string; 3D annotation tools writing extra columns.

Understand the failure class

Parsing and encoding errors: unexpected token, malformed input — why parsers reject input and how to find the real culprit.

Related errors

AI-assisted analysis of roboflow/supervision@7f254d9784 (2026-08-15). Data as JSON: /api/errors/54ac5cea1c8bd9cb. Report an issue: GitHub.