keras-team/keras · error · ValueError

`encoded_format` should be 'center_xywh' or 'center_yxhw', b

Error message

`encoded_format` should be 'center_xywh' or 'center_yxhw', but got '{encoded_format}'.

What it means

decode_deltas_to_boxes accepts only two encoded formats: 'center_xywh' and 'center_yxhw'. The encoded_format must match the format used when the deltas were produced; any other string raises this ValueError.

Source

Thrown at keras/src/layers/preprocessing/image_preprocessing/bounding_boxes/converters.py:404

    Returns:
        Decoded box coordinates. The return type matches the `box_format`.

    Raises:
        ValueError: If `variance` is not None and its length is not 4.
        ValueError: If `encoded_format` is not `"center_xywh"` or
            `"center_yxhw"`.

    """
    if variance is not None:
        variance = ops.convert_to_tensor(variance, "float32")
        var_len = variance.shape[-1]

        if var_len != 4:
            raise ValueError(f"`variance` must be length 4, got {variance}")

    if encoded_format not in ["center_xywh", "center_yxhw"]:
        raise ValueError(
            f"`encoded_format` should be 'center_xywh' or 'center_yxhw', "
            f"but got '{encoded_format}'."
        )

    if image_shape is None:
        height, width = None, None
    else:
        height, width, _ = image_shape

    def decode_single_level(anchor, box_delta):
        encoded_anchor = convert_format(
            anchor,
            source=anchor_format,
            target=encoded_format,
            height=height,
            width=width,
        )
        if variance is not None:

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Set encoded_format to exactly the format used at encode time: 'center_xywh' or 'center_yxhw'.
2. Store the encoding format alongside the deltas/anchors in checkpoints or configs so both ends agree.

Example fix

# before
boxes = decode_deltas_to_boxes(deltas, anchors, encoded_format="xywh")
# after
boxes = decode_deltas_to_boxes(deltas, anchors, encoded_format="center_xywh")

assert encoded_format in ("center_xywh", "center_yxhw"), f"bad encoded_format {encoded_format!r}"

def is_valid_encoding_format(f):
    return f in {"center_xywh", "center_yxhw"}

When it happens

Trigger: Passing encoded_format='xywh', 'corner', or forgetting the argument when deltas were produced with a custom encoder.

Common situations: Encode side used 'center_yxhw' but decode side copies 'center_xywh' or vice versa; renaming format constants mid-project.

Related errors

• `encoding_format` should be one of 'center_xywh' or 'center_
• `height` and `width` must be set if `format='xyxy'`.
• `variance` must be length 4, got {variance}
• compute_iou() expects boxes1 to be batched, or to be unbatch
• compute_iou() expects boxes2 to be batched, or to be unbatch

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/d2d55fa665478f9d. Report an issue: GitHub.