roboflow/supervision · error · ValueError

F1Score metric requires `class_id` and `confidence` on predi

Error message

F1Score metric requires `class_id` and `confidence` on predictions.

What it means

Raised by F1Score.update() when an image contains only predictions and no targets (e.g. a background image with false positives), but the predictions lack class_id or confidence. The metric needs class_id to bucket false positives per class and confidence to rank predictions across thresholds. Without these fields, per-class F1 statistics cannot be accumulated.

Source

Thrown at src/supervision/metrics/f1_score.py:186

            prediction_size_mask = np.ones(len(predictions), dtype=bool)
            target_size_mask = np.ones(len(targets), dtype=bool)
            if size_category != ObjectSizeCategory.ANY:
                if len(predictions) > 0:
                    prediction_size_mask = (
                        get_detection_size_category(predictions, self._metric_target)
                        == size_category.value
                    )
                if len(targets) > 0:
                    target_size_mask = (
                        get_detection_size_category(targets, self._metric_target)
                        == size_category.value
                    )

            if len(targets) == 0 and len(predictions) > 0:
                # Only predictions are present (e.g. a background image); every
                # prediction is a false positive.
                if predictions.class_id is None or predictions.confidence is None:
                    raise ValueError(
                        "F1Score metric requires `class_id` and `confidence` "
                        "on predictions."
                    )
                prediction_class_ids = np.asarray(predictions.class_id, dtype=np.int32)[
                    prediction_size_mask
                ]
                prediction_confidence = np.asarray(
                    predictions.confidence, dtype=np.float32
                )[prediction_size_mask]
                if len(prediction_class_ids) == 0:
                    continue
                stats.append(
                    (
                        np.zeros(
                            (len(prediction_class_ids), iou_thresholds.size),
                            dtype=np.bool_,
                        ),
                        np.zeros(

View on GitHub (pinned to 7f254d9784)

Solutions

1. Attach class_id and confidence to the predictions Detections: Detections(xyxy=..., class_id=np.array([0]), confidence=np.array([0.9]))
2. If your model connector drops these fields, keep them: most from_* connectors populate both automatically
3. Skip images with no targets before calling update() if you do not want background images scored

Example fix

# before
preds = sv.Detections(xyxy=boxes)  # no class_id / confidence
f1.update(targets=sv.Detections.empty(), predictions=preds)

# after
preds = sv.Detections(
    xyxy=boxes,
    class_id=np.zeros(len(boxes), dtype=np.int32),
    confidence=scores,
)
f1.update(targets=sv.Detections.empty(), predictions=preds)

def has_f1_fields(preds: sv.Detections) -> bool:
    return preds.class_id is not None and preds.confidence is not None

if not has_f1_fields(predictions):
    raise ValueError('predictions need class_id and confidence before F1')

def is_f1_ready(dets: sv.Detections) -> bool:
    """True when class_id and confidence are populated."""
    return dets.class_id is not None and dets.confidence is not None

try:
    f1.update(targets=targets, predictions=predictions)
except ValueError as e:
    if 'class_id and confidence' in str(e):
        predictions.class_id = predictions.class_id or np.zeros(len(predictions), dtype=np.int32)
    else:
        raise

When it happens

Trigger: Calling F1Score().update(targets=empty_detections, predictions=Detections(xyxy=..., class_id=None)) or predictions without a confidence array, on an image where len(targets)==0 and len(predictions)>0.

Common situations: Hand-built Detections objects (e.g. from a custom model connector) that omit confidence; predictions crafted from trackers that drop confidence; test fixtures with empty target sets but populated predictions.

Related errors

• F1Score metric requires `class_id` on both predictions and t
• F1Score metric requires `confidence` on predictions.
• 2D boolean mask row count {mask.shape[0]} does not match obj
• 2D boolean mask column count {mask.shape[1]} does not match
• Cannot filter keypoints with a 2D boolean mask where rows ha

AI-assisted analysis of roboflow/supervision@7f254d9784 (2026-08-15). Data as JSON: /api/errors/0a13574a2a1b9011. Report an issue: GitHub.