roboflow/supervision · error · ValueError

from_inference() operates on a single result at a time.You c

Error message

from_inference() operates on a single result at a time.You can retrieve it like so:  inference_result = model.infer(image)[0]

What it means

Inside Precision.compute()'s main matching branch, the code selects the IoU function by metric target: box_iou_batch for BOXES, mask_iou_batch for MASKS, oriented_box_iou_batch for ORIENTED_BOUNDING_BOXES. Any other MetricTarget value reaches the else-raise. With the current enum this is defensive dead code; it fires only on version skew or injected enum values.

Source

Thrown at src/supervision/key_points/core.py:428

            ```

            ```python
            from supervision import _cv2 as cv2
            import supervision as sv
            from inference_sdk import InferenceHTTPClient

            image = cv2.imread("<SOURCE_IMAGE_PATH>")
            client = InferenceHTTPClient(
                api_url="https://detect.roboflow.com",
                api_key="<ROBOFLOW_API_KEY>"
            )

            result = client.infer(image, model_id="<POSE_MODEL_ID>")
            key_points = sv.KeyPoints.from_inference(result)
            ```
        """
        if isinstance(inference_result, list):
            raise ValueError(
                "from_inference() operates on a single result at a time."
                "You can retrieve it like so:  inference_result = model.infer(image)[0]"
            )

        if hasattr(inference_result, "dict"):
            inference_result = inference_result.dict(exclude_none=True, by_alias=True)
        elif hasattr(inference_result, "json"):
            inference_result = inference_result.json()
        if not inference_result.get("predictions"):
            return cls.empty()

        xy = []
        confidence = []
        class_id = []
        class_names = []

        for prediction in inference_result["predictions"]:
            prediction_xy = []

View on GitHub (pinned to 7f254d9784)

Solutions

1. Reinstall/upgrade supervision so the enum and metric modules come from one version (pip install --force-reinstall supervision)
2. Restrict metric_target to the three supported members
3. Remove enum monkey-patches before computing metrics

SUPPORTED_IOU_TARGETS = {sv.MetricTarget.BOXES, sv.MetricTarget.MASKS, sv.MetricTarget.ORIENTED_BOUNDING_BOXES}
assert metric_target in SUPPORTED_IOU_TARGETS, f'{metric_target} has no IoU kernel here'

def has_iou_kernel(target: sv.MetricTarget) -> bool:
    return target in {sv.MetricTarget.BOXES, sv.MetricTarget.MASKS, sv.MetricTarget.ORIENTED_BOUNDING_BOXES}

When it happens

Trigger: Running a supervision version where MetricTarget gained a member (e.g. a future target) that this Precision build does not handle — typically a mixed/partial install or a monkey-patched enum — while evaluating images with both predictions and targets.

Common situations: Partial upgrades, vendored/copied metric modules paired with a newer installed package, or custom enum extension experiments.

Related errors

• 2D boolean mask row count {mask.shape[0]} does not match obj
• 2D boolean mask column count {mask.shape[1]} does not match
• edges is a dict but class_id is None; KeyPoints must have cl
• No edges defined for class_id={class_id}.
• 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/39c8cac2de0e6b03. Report an issue: GitHub.