keras-team/keras · error · NotImplementedError

Unknown `interpolation` {interpolation}. Expected of one {se

Error message

Unknown `interpolation` {interpolation}. Expected of one {self._SUPPORTED_INTERPOLATION}.

What it means

Raised by RandomZoom's constructor when `interpolation` is not "nearest" or "bilinear". These are the only resampling methods implemented for the zoom transform.

Source

Thrown at keras/src/layers/preprocessing/image_preprocessing/random_zoom.py:135

        **kwargs,
    ):
        super().__init__(**kwargs)
        self.height_factor = height_factor
        self.height_lower, self.height_upper = self._set_factor(
            height_factor, "height_factor"
        )
        self.width_factor = width_factor
        if width_factor is not None:
            self.width_lower, self.width_upper = self._set_factor(
                width_factor, "width_factor"
            )
        if fill_mode not in self._SUPPORTED_FILL_MODE:
            raise NotImplementedError(
                f"Unknown `fill_mode` {fill_mode}. Expected of one "
                f"{self._SUPPORTED_FILL_MODE}."
            )
        if interpolation not in self._SUPPORTED_INTERPOLATION:
            raise NotImplementedError(
                f"Unknown `interpolation` {interpolation}. Expected of one "
                f"{self._SUPPORTED_INTERPOLATION}."
            )

        self.fill_mode = fill_mode
        self.fill_value = fill_value
        self.interpolation = interpolation
        self.seed = seed
        self.generator = SeedGenerator(seed)
        self.data_format = backend.standardize_data_format(data_format)
        self.supports_jit = False

    def _set_factor(self, factor, factor_name):
        if isinstance(factor, (tuple, list)):
            if len(factor) != 2:
                raise ValueError(
                    self._FACTOR_VALIDATION_ERROR
                    + f"Received: {factor_name}={factor}"

View on GitHub (pinned to 7a34a03db6)

Solutions

Use "bilinear" (default, smooth) or "nearest" (fast)
Pre/post-resize with tf.image.resize if you need higher-order interpolation

Example fix

// before
layer = RandomZoom(0.2, interpolation="area")
// after
layer = RandomZoom(0.2, interpolation="bilinear")

Defensive patterns

Strategy: validation

Validate before calling

assert interpolation in {"nearest", "bilinear"}, 'RandomZoom supports only nearest/bilinear'

Type guard

def is_valid_interp(v):
    return v in {"nearest", "bilinear"}

Try / catch

try:
    layer = RandomZoom(0.2, interpolation=interp)
except NotImplementedError:
    layer = RandomZoom(0.2, interpolation="bilinear")

Prevention

Restrict config schemas to the two supported values
Do extra resampling outside the layer if higher-order interpolation is needed

When it happens

Trigger: Calling RandomZoom(0.2, interpolation="bicubic") or "area".

Common situations: Copying interpolation settings from tf.image.resize or another layer/version that supports more methods; assuming OpenCV-style flags work.

Related errors

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/b51d99e155937e8f. Report an issue: GitHub.