keras-team/keras · error · ValueError

Found bounding_boxes['boxes'].shape={boxes_shape} and expect

Error message

Found bounding_boxes['boxes'].shape={boxes_shape} and expected bounding_boxes['labels'] to have rank 1 or 2, but received: bounding_boxes['labels'].shape={labels_shape} 

What it means

For dense tensors, when boxes has rank 2 (single image, shape (num_boxes, 4)), labels must have rank 1 (num_boxes,) or rank 2 (1, num_boxes). Any other rank (e.g. a scalar or rank-3 labels tensor) breaks the box-to-label correspondence.

Source

Thrown at keras/src/layers/preprocessing/image_preprocessing/bounding_boxes/validation.py:159

                "`bounding_boxes['labels']` are both lists, "
                "they must have the same length. Received: "
                f"len(bounding_boxes['boxes'])={len(boxes)} and "
                f"len(bounding_boxes['labels'])={len(labels)} and "
            )
    elif tf_utils.is_ragged_tensor(boxes):
        if not tf_utils.is_ragged_tensor(labels):
            raise ValueError(
                "If `bounding_boxes['boxes']` is a Ragged tensor, "
                " `bounding_boxes['labels']` must also be a "
                "Ragged tensor. "
                f"Received: bounding_boxes['labels']={labels}"
            )
    else:
        boxes_shape = current_backend.shape(boxes)
        labels_shape = current_backend.shape(labels)
        if len(boxes_shape) == 2:  # (boxes, 4)
            if len(labels_shape) not in {1, 2}:
                raise ValueError(
                    "Found "
                    f"bounding_boxes['boxes'].shape={boxes_shape} "
                    "and expected bounding_boxes['labels'] to have "
                    "rank 1 or 2, but received: "
                    f"bounding_boxes['labels'].shape={labels_shape} "
                )
        elif len(boxes_shape) == 3:
            if len(labels_shape) not in {2, 3}:
                raise ValueError(
                    "Found "
                    f"bounding_boxes['boxes'].shape={boxes_shape} "
                    "and expected bounding_boxes['labels'] to have "
                    "rank 2 or 3, but received: "
                    f"bounding_boxes['labels'].shape={labels_shape} "
                )
        else:
            raise ValueError(
                "Expected `bounding_boxes['boxes']` "

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Reshape labels to (num_boxes,) — e.g. labels.reshape(-1) or keras.ops.reshape(labels, (-1,))
2. If one-hot, keep rank 2 as (num_boxes, num_classes)
3. Check boxes actually is rank 2; a stray leading dim on boxes also shifts the expected labels rank

Example fix

# before
bbs = {'boxes': boxes, 'labels': labels[None, None, :]}
# after
bbs = {'boxes': boxes, 'labels': keras.ops.reshape(labels, (-1,))}

if len(bbs['boxes'].shape) == 2:
    assert len(bbs['labels'].shape) in (1, 2), bbs['labels'].shape

def labels_rank_ok(boxes, labels):
    r = len(boxes.shape)
    lr = len(labels.shape)
    return (r == 2 and lr in (1, 2)) or (r == 3 and lr in (2, 3))

When it happens

Trigger: densify_bounding_boxes with boxes shape (num_boxes, 4) and labels of rank 0, 3+, e.g. labels shape (1, 1, num_boxes) or a scalar class id.

Common situations: Squeezing/reshaping labels incorrectly during preprocessing; carrying extra leading dims (e.g. (1, batch, boxes)) from a previous stage; using one-hot labels with extra dims.

Related errors

• Expected `bounding_boxes['boxes']` to have rank 2 or 3, with
• If `bounding_boxes['boxes']` is a list, then `bounding_boxes
• Found bounding_boxes['boxes'].shape={boxes_shape} and expect
• Expected as input a list/tuple of 2 tensors. Received input_
• `height` and `width` must be set if `format='xyxy'`.

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/6800754edb650abd. Report an issue: GitHub.