jax-ml/jax · error · ValueError

shape must have length equal to the number of dimensions of

Error message

shape must have length equal to the number of dimensions of x;  {shape} vs {image.shape}

What it means

jax.image.scale.scale_and_translate requires the shape argument's length to equal image.ndim (it describes the full output shape, not just spatial dims). After canonicalizing, a length mismatch raises ValueError comparing len(shape) with image.shape.

Source

Thrown at jax/_src/image/scale.py:374

      passed scale and translation should be applied to.
    scale: A [K] array with the same number of dimensions as image, containing
      the scale to apply in each dimension.
    translation: A [K] array with the same number of dimensions as image,
      containing the translation to apply in each dimension.
    method: the resizing method to use; either a ``ResizeMethod`` instance or a
      string. Available methods are: ``LINEAR``, ``LANCZOS3``, ``LANCZOS5``,
      ``CUBIC``, ``CUBIC_PYTORCH``, ``AREA``.
    antialias: Should an antialiasing filter be used when downsampling? Defaults
      to ``True``. Has no effect when upsampling.

  Returns:
    The scale and translated image.
  """
  shape = core.canonicalize_shape(shape)
  if len(shape) != image.ndim:
    msg = ('shape must have length equal to the number of dimensions of x; '
           f' {shape} vs {image.shape}')
    raise ValueError(msg)
  if isinstance(method, str):
    method = ResizeMethod.from_string(method)
  if method == ResizeMethod.NEAREST:
    # Nearest neighbor is currently special-cased for straight resize, so skip
    # for now.
    raise ValueError('Nearest neighbor resampling is not currently supported '
                     'for scale_and_translate.')
  assert isinstance(method, ResizeMethod)

  if method == ResizeMethod.CUBIC_PYTORCH and antialias:
    method = ResizeMethod.CUBIC
  radius, kernel = _kernels[method]
  edge_padding = (method == ResizeMethod.CUBIC_PYTORCH and not antialias)
  image, = promote_dtypes_inexact(image)
  scale, translation = promote_dtypes_inexact(scale, translation)
  return _scale_and_translate(
     image, shape, spatial_dims, scale, translation, kernel, antialias,
     precision, edge_padding=edge_padding, radius=radius)

View on GitHub (pinned to 1e1c6a8fc0)

Solutions

1. Pass the full output shape matching image.ndim, e.g. (H2, W2, C) for CHW-unaware arrays (channels included)
2. Derive it programmatically: out_shape = list(img.shape); out_shape[-3:-1] = (h2, w2) for channel-last
3. Prefer the higher-level jax.image.resize which handles this mapping from spatial dims

Example fix

# before
img = jnp.zeros((224, 224, 3))
out = scale_and_translate(img, (112, 112), ...)  # ValueError
# after
out = scale_and_translate(img, (112, 112, 3), ...)
# or jax.image.resize(img, (112, 112), method='linear')

assert len(shape) == image.ndim, f'need full output shape of length {image.ndim}'
scale_and_translate(image, shape, scale, translation, method)

# no static type narrowing available; runtime check only

When it happens

Trigger: Passing a 2-element spatial shape for a 3D (H,W,C) image, or a 4-element shape for a batched (N,H,W,C) image, to scale_and_translate (and via _resize).

Common situations: Coming from cv2/torchvision where resize takes only (height, width); forgetting channel/batch dims; computing shape from a different tensor than the image passed in.

Related errors

• unexpected JAX type (e.g. shape/dtype) for argument to VJP f
• cotangent type does not match function output, expected {out
• Mismatched number of outputs from callback. Expected: {}, Ac
• Incorrect output shape for return value #{i}: Expected: {out
• Length of sharding.spec ({len(out_s.spec)}) must be equal to

AI-assisted analysis of jax-ml/jax@1e1c6a8fc0 (2026-08-27). Data as JSON: /api/errors/8318635ae72efba3. Report an issue: GitHub.