roboflow/supervision · error · ValueError

{role} class ids must be finite integers.

Error message

{role} class ids must be finite integers.

What it means

Error "{role} class ids must be finite integers." thrown in roboflow/supervision.

Source

Thrown at src/supervision/metrics/detection.py:51

    if metric_target == MetricTarget.MASKS:
        raise ValueError(
            "MetricTarget.MASKS is not currently supported for ConfusionMatrix."
        )


def _validated_class_ids(
    values: npt.NDArray[np.float32],
    num_classes: int,
    role: str,
) -> npt.NDArray[np.int64]:
    """Return class ids that are safe to use as confusion-matrix indexes."""
    if values.size == 0:
        return np.asarray(values, dtype=np.int64)

    finite = np.isfinite(values)
    integral = values == np.floor(values)
    if not np.all(finite & integral):
        raise ValueError(f"{role} class ids must be finite integers.")

    class_ids = values.astype(np.int64)
    invalid = (class_ids < 0) | (class_ids >= num_classes)
    if np.any(invalid):
        invalid_values = np.unique(class_ids[invalid]).tolist()
        raise ValueError(
            f"{role} class ids must be in [0, {num_classes - 1}], got {invalid_values}."
        )
    return class_ids


def detections_to_tensor(
    detections: Detections,
    with_confidence: bool = False,
    metric_target: MetricTarget = MetricTarget.BOXES,
) -> npt.NDArray[np.float32]:
    """
    Convert Supervision Detections to a numpy tensor for metric computation.

View on GitHub (pinned to 7f254d9784)

Solutions

1. Ensure class ids are finite integers (no NaN, inf, or float values) for both predictions and targets.
2. Cast class ids with arr.astype(int) after validating they contain no NaN/inf.

When it happens

Trigger: Thrown at src/supervision/metrics/detection.py:51 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of roboflow/supervision@7f254d9784 (2026-08-15). Data as JSON: /api/errors/5e13b641bc7fec7f. Report an issue: GitHub.