What causes "Unsupported index type: {type(index)}" in roboflow/supervision?

Convert the index to int first: idx = np.asarray(idx).astype(int) or int(idx)

roboflow/supervision · error · TypeError

Unsupported index type: {type(index)}

Error message

Unsupported index type: {type(index)}

What it means

Raised by get_data_item when a Detections is indexed with an index whose type is not slice, list of ints, int, or integer/boolean np.ndarray, and the Detections has at least one list-typed data field. ndarray data values would instead raise a NumPy indexing error, so this guard fires only on the list-comprehension path.

Source

Thrown at src/supervision/detection/utils/internal.py:686

    for key, value in data.items():
        if isinstance(value, np.ndarray):
            subset_data[key] = value[index]
        elif isinstance(value, list):
            if isinstance(index, slice):
                subset_data[key] = value[index]
            elif isinstance(index, list):
                subset_data[key] = [value[i] for i in index]
            elif isinstance(index, np.ndarray):
                if index.dtype == bool:
                    subset_data[key] = [
                        value[i] for i, index_value in enumerate(index) if index_value
                    ]
                else:
                    subset_data[key] = [value[i] for i in index]
            elif isinstance(index, int):
                subset_data[key] = [value[index]]
            else:
                raise TypeError(f"Unsupported index type: {type(index)}")
        else:
            raise TypeError(f"Unsupported data type for key '{key}': {type(value)}")

    return subset_data


def cross_product(
    anchors: npt.NDArray[np.number], vector: Vector
) -> npt.NDArray[np.number]:
    """Get signed z-component of cross product (2-D determinant) per anchor.

    Replaces the deprecated `np.cross` 2-D path (NumPy 2.0) with an explicit
    determinant: ``a[..., 0] * b[..., 1] - a[..., 1] * b[..., 0]``.

    Args:
        anchors: Array of anchors of shape (number of anchors, detections, 2).
        vector: Vector to calculate cross product with.

View on GitHub (pinned to 7f254d9784)

Solutions

Convert the index to int first: idx = np.asarray(idx).astype(int) or int(idx)
Use canonical index forms: a Python int, list[int], slice, or np.ndarray of integer/bool dtype
Check type of computed indices with isinstance before indexing Detections

Example fix

# before
sub = detections[np.nonzero(mask)[0] * 1.0]  # float index
# after
sub = detections[np.nonzero(mask)[0].astype(int)]

Defensive patterns

Strategy: type-guard

Validate before calling

import numpy as np

def coerce_index(idx):
    if isinstance(idx, np.ndarray):
        if idx.dtype == bool:
            return idx
        return idx.astype(int)
    if isinstance(idx, (int, np.integer)):
        return int(idx)
    if isinstance(idx, (list, slice)):
        return idx
    raise TypeError(f"bad index type: {type(idx)}")

Type guard

def is_valid_detections_index(idx) -> bool:
    import numpy as np
    if isinstance(idx, (int, slice)) or isinstance(idx, list):
        return True
    return isinstance(idx, np.ndarray) and idx.dtype.kind in ('i', 'u', 'b')

Try / catch

try:
    sub = detections[idx]
except TypeError as e:
    if "Unsupported index type" in str(e):
        sub = detections[np.asarray(idx).astype(int)]
    else:
        raise

Prevention

Convert indices to int explicitly after np.where/argwhere arithmetic
Never index Detections with floats or tuples
Centralize index computation in one helper that always returns int arrays

When it happens

Trigger: detections[0.0], detections[(0, 1)], detections[np.array([0.5])] or any exotic index type applied to a Detections whose data contains a list value. Float indices from np.argwhere-derived code (forgetting .astype(int) or .tolist() of ints) are typical.

Common situations: Passing a float or float array produced by filtering code (e.g. np.where result passed through arithmetic) directly as an index; using a tuple index copied from ndarray idioms; wrapping an index in a dtype=object array.

Related errors

AI-assisted analysis of roboflow/supervision@7f254d9784 (2026-08-15). Data as JSON: /api/errors/0d6b22b9b3d8b268. Report an issue: GitHub.