keras-nightly 3.12.0.dev2025100503__py3-none-any.whl → 3.14.0.dev2026011604__py3-none-any.whl

keras/src/ops/image.py

@@ -565,6 +565,8 @@ class ExtractPatches(Operation):
         if isinstance(size, int):
             size = (size, size)
         self.size = size
+        if strides is None:
+            strides = size
         self.strides = strides
         self.dilation_rate = dilation_rate
         self.padding = padding
@@ -583,8 +585,6 @@ class ExtractPatches(Operation):
     def compute_output_spec(self, images):
         images_shape = list(images.shape)
         original_ndim = len(images_shape)
-        if not self.strides:
-            strides = (self.size[0], self.size[1])
         if self.data_format == "channels_last":
             channels_in = images_shape[-1]
         else:
@@ -597,7 +597,7 @@ class ExtractPatches(Operation):
             images_shape,
             filters,
             kernel_size,
-            strides=strides,
+            strides=self.strides,
             padding=self.padding,
             data_format=self.data_format,
             dilation_rate=self.dilation_rate,
@@ -616,42 +616,98 @@ def extract_patches(
     padding="valid",
     data_format=None,
 ):
-    """Extracts patches from the image(s).
+    """Extracts patches from the image(s) or volume(s).
+
+    This function supports both 2D and 3D patch extraction based on the
+    `size` argument length, similar to how `keras.ops.conv` handles
+    different dimensions.
 
     Args:
-        images: Input image or batch of images. Must be 3D or 4D.
-        size: Patch size int or tuple (patch_height, patch_width)
-        strides: strides along height and width. If not specified, or
-            if `None`, it defaults to the same value as `size`.
-        dilation_rate: This is the input stride, specifying how far two
-            consecutive patch samples are in the input. For value other than 1,
-            strides must be 1. NOTE: `strides > 1` is not supported in
-            conjunction with `dilation_rate > 1`
+        images: Input image/volume or batch of images/volumes.
+            For 2D patches: 3D `(H, W, C)` or 4D `(N, H, W, C)`.
+            For 3D patches: 4D `(D, H, W, C)` or 5D `(N, D, H, W, C)`.
+        size: Patch size as int or tuple.
+            Length 2 tuple `(patch_height, patch_width)` or int for 2D patches.
+            Length 3 tuple `(patch_depth, patch_height, patch_width)` for
+            3D patches.
+        strides: Strides for patch extraction. If not specified, defaults
+            to `size` (non-overlapping patches).
+        dilation_rate: Dilation rate for patch extraction. Note that
+            `dilation_rate > 1` is not supported with `strides > 1`.
         padding: The type of padding algorithm to use: `"same"` or `"valid"`.
         data_format: A string specifying the data format of the input tensor.
             It can be either `"channels_last"` or `"channels_first"`.
-            `"channels_last"` corresponds to inputs with shape
-            `(batch, height, width, channels)`, while `"channels_first"`
-            corresponds to inputs with shape `(batch, channels, height, width)`.
-            If not specified, the value will default to
-            `keras.config.image_data_format`.
+            If not specified, defaults to `keras.config.image_data_format`.
 
     Returns:
-        Extracted patches 3D (if not batched) or 4D (if batched)
+        Extracted patches with shape depending on input and `size`:
+        - 2D patches: 3D (unbatched) or 4D (batched)
+        - 3D patches: 4D (unbatched) or 5D (batched)
 
     Examples:
 
+    >>> # 2D patches from batch of images
     >>> image = np.random.random(
     ...     (2, 20, 20, 3)
-    ... ).astype("float32") # batch of 2 RGB images
+    ... ).astype("float32")
     >>> patches = keras.ops.image.extract_patches(image, (5, 5))
     >>> patches.shape
     (2, 4, 4, 75)
-    >>> image = np.random.random((20, 20, 3)).astype("float32") # 1 RGB image
+
+    >>> # 2D patches from single image
+    >>> image = np.random.random((20, 20, 3)).astype("float32")
     >>> patches = keras.ops.image.extract_patches(image, (3, 3), (1, 1))
     >>> patches.shape
     (18, 18, 27)
+
+    >>> # 3D patches from batch of volumes
+    >>> volumes = np.random.random(
+    ...     (2, 10, 10, 10, 3)
+    ... ).astype("float32")
+    >>> patches = keras.ops.image.extract_patches(volumes, (3, 3, 3))
+    >>> patches.shape
+    (2, 3, 3, 3, 81)
+
+    >>> # 3D patches from single volume
+    >>> volume = np.random.random((10, 10, 10, 3)).astype("float32")
+    >>> patches = keras.ops.image.extract_patches(volume, (3, 3, 3))
+    >>> patches.shape
+    (3, 3, 3, 81)
     """
+    # Validate size argument
+    if not isinstance(size, int):
+        if not isinstance(size, (tuple, list)):
+            raise TypeError(
+                "Invalid `size` argument. Expected an int or a tuple. "
+                f"Received: size={size} of type {type(size).__name__}"
+            )
+        if len(size) not in (2, 3):
+            raise ValueError(
+                "Invalid `size` argument. Expected a tuple of length 2 or 3. "
+                f"Received: size={size} with length {len(size)}"
+            )
+
+    # Determine 2D vs 3D based on size argument
+    if not isinstance(size, int) and len(size) == 3:
+        # 3D patch extraction
+        if any_symbolic_tensors((images,)):
+            return ExtractPatches3D(
+                size=size,
+                strides=strides,
+                dilation_rate=dilation_rate,
+                padding=padding,
+                data_format=data_format,
+            ).symbolic_call(images)
+        return _extract_patches_3d(
+            images,
+            size,
+            strides,
+            dilation_rate,
+            padding,
+            data_format=data_format,
+        )
+
+    # 2D patch extraction (default)
     if any_symbolic_tensors((images,)):
         return ExtractPatches(
             size=size,
@@ -712,6 +768,187 @@ def _extract_patches(
     return patches
 
 
+class ExtractPatches3D(Operation):
+    def __init__(
+        self,
+        size,
+        strides=None,
+        dilation_rate=1,
+        padding="valid",
+        data_format=None,
+        *,
+        name=None,
+    ):
+        super().__init__(name=name)
+        if isinstance(size, int):
+            size = (size, size, size)
+        elif len(size) != 3:
+            raise TypeError(
+                "Invalid `size` argument. Expected an "
+                f"int or a tuple of length 3. Received: size={size}"
+            )
+        self.size = size
+        if strides is not None:
+            if isinstance(strides, int):
+                strides = (strides, strides, strides)
+            elif len(strides) != 3:
+                raise ValueError(f"Invalid `strides` argument. Got: {strides}")
+        else:
+            strides = size
+        self.strides = strides
+        self.dilation_rate = dilation_rate
+        self.padding = padding
+        self.data_format = backend.standardize_data_format(data_format)
+
+    def call(self, volumes):
+        return _extract_patches_3d(
+            volumes,
+            self.size,
+            self.strides,
+            self.dilation_rate,
+            self.padding,
+            self.data_format,
+        )
+
+    def compute_output_spec(self, volumes):
+        volumes_shape = list(volumes.shape)
+        original_ndim = len(volumes_shape)
+        strides = self.strides
+        if self.data_format == "channels_last":
+            channels_in = volumes_shape[-1]
+        else:
+            channels_in = volumes_shape[-4]
+        if original_ndim == 4:
+            volumes_shape = [1] + volumes_shape
+        filters = self.size[0] * self.size[1] * self.size[2] * channels_in
+        kernel_size = (self.size[0], self.size[1], self.size[2])
+        out_shape = compute_conv_output_shape(
+            volumes_shape,
+            filters,
+            kernel_size,
+            strides=strides,
+            padding=self.padding,
+            data_format=self.data_format,
+            dilation_rate=self.dilation_rate,
+        )
+        if original_ndim == 4:
+            out_shape = out_shape[1:]
+        return KerasTensor(shape=out_shape, dtype=volumes.dtype)
+
+
+def _extract_patches_3d(
+    volumes,
+    size,
+    strides=None,
+    dilation_rate=1,
+    padding="valid",
+    data_format=None,
+):
+    if isinstance(size, int):
+        patch_d = patch_h = patch_w = size
+    elif len(size) == 3:
+        patch_d, patch_h, patch_w = size
+    else:
+        raise TypeError(
+            "Invalid `size` argument. Expected an "
+            f"int or a tuple of length 3. Received: size={size}"
+        )
+    if strides is None:
+        strides = size
+    if isinstance(strides, int):
+        strides = (strides, strides, strides)
+    if len(strides) != 3:
+        raise ValueError(f"Invalid `strides` argument. Got: {strides}")
+    data_format = backend.standardize_data_format(data_format)
+    if data_format == "channels_last":
+        channels_in = volumes.shape[-1]
+    elif data_format == "channels_first":
+        channels_in = volumes.shape[-4]
+    out_dim = patch_d * patch_w * patch_h * channels_in
+    kernel = backend.numpy.eye(out_dim, dtype=volumes.dtype)
+    kernel = backend.numpy.reshape(
+        kernel, (patch_d, patch_h, patch_w, channels_in, out_dim)
+    )
+    _unbatched = False
+    if len(volumes.shape) == 4:
+        _unbatched = True
+        volumes = backend.numpy.expand_dims(volumes, axis=0)
+    patches = backend.nn.conv(
+        inputs=volumes,
+        kernel=kernel,
+        strides=strides,
+        padding=padding,
+        data_format=data_format,
+        dilation_rate=dilation_rate,
+    )
+    if _unbatched:
+        patches = backend.numpy.squeeze(patches, axis=0)
+    return patches
+
+
+@keras_export("keras.ops.image.extract_patches_3d")
+def extract_patches_3d(
+    volumes,
+    size,
+    strides=None,
+    dilation_rate=1,
+    padding="valid",
+    data_format=None,
+):
+    """Extracts patches from the volume(s).
+
+    Args:
+        volumes: Input volume or batch of volumes. Must be 4D or 5D.
+        size: Patch size int or tuple (patch_depth, patch_height, patch_width)
+        strides: strides along depth, height, and width. If not specified, or
+            if `None`, it defaults to the same value as `size`.
+        dilation_rate: This is the input stride, specifying how far two
+            consecutive patch samples are in the input. Note that using
+            `dilation_rate > 1` is not supported in conjunction with
+            `strides > 1` on the TensorFlow backend.
+        padding: The type of padding algorithm to use: `"same"` or `"valid"`.
+        data_format: A string specifying the data format of the input tensor.
+            It can be either `"channels_last"` or `"channels_first"`.
+            `"channels_last"` corresponds to inputs with shape
+            `(batch, depth, height, width, channels)`, while `"channels_first"`
+            corresponds to inputs with shape
+            `(batch, channels, depth, height, width)`. If not specified,
+             the value will default to `keras.config.image_data_format()`.
+
+    Returns:
+        Extracted patches 4D (if not batched) or 5D (if batched)
+
+    Examples:
+
+    >>> import numpy as np
+    >>> import keras
+    >>> # Batched case
+    >>> volumes = np.random.random(
+    ...     (2, 10, 10, 10, 3)
+    ... ).astype("float32") # batch of 2 volumes
+    >>> patches = keras.ops.image.extract_patches_3d(volumes, (3, 3, 3))
+    >>> patches.shape
+    (2, 3, 3, 3, 81)
+    >>> # Unbatched case
+    >>> volume = np.random.random((10, 10, 10, 3)).astype("float32") # 1 volume
+    >>> patches = keras.ops.image.extract_patches_3d(volume, (3, 3, 3))
+    >>> patches.shape
+    (3, 3, 3, 81)
+    """
+    if any_symbolic_tensors((volumes,)):
+        return ExtractPatches3D(
+            size=size,
+            strides=strides,
+            dilation_rate=dilation_rate,
+            padding=padding,
+            data_format=data_format,
+        ).symbolic_call(volumes)
+
+    return _extract_patches_3d(
+        volumes, size, strides, dilation_rate, padding, data_format=data_format
+    )
+
+
 class MapCoordinates(Operation):
     def __init__(self, order, fill_mode="constant", fill_value=0, *, name=None):
         super().__init__(name=name)

keras/src/ops/linalg.py

@@ -1,4 +1,5 @@
 from keras.src import backend
+from keras.src import tree
 from keras.src.api_export import keras_export
 from keras.src.backend import KerasTensor
 from keras.src.backend import any_symbolic_tensors
@@ -732,3 +733,95 @@ def _assert_a_b_compat(a, b):
                 "Expected `a.shape[-1] == b.shape[-1]`. "
                 f"Received: a.shape={a.shape}, b.shape={b.shape}"
             )
+
+
+class JVP(Operation):
+    def __init__(self, has_aux=False, *, name=None):
+        super().__init__(name=name)
+        self.has_aux = has_aux
+
+    def call(self, fun, primals, tangents):
+        """Computes the JVP of `fun` at `primals` along `tangents`.
+
+        Args:
+            fun: A callable that takes tensors (or nested structures) as input
+                 and returns a tensor (or nested structure) as output.
+            primals: Input tensors (or nested structures) at which the Jacobian
+                     of `fun` is evaluated.
+            tangents: Tensors (or nested structures) representing the direction
+                      vectors for the JVP. Must have the same structure as
+                      `primals`.
+
+        Returns:
+            If `has_aux` is False:
+                A tuple (primals_out, tangents_out) where:
+                - primals_out: Output of `fun(*primals)`
+                - tangents_out: JVP of `fun` at `primals` along `tangents`
+            If `has_aux` is True:
+                A tuple (primals_out, tangents_out, aux) where:
+                - aux: Auxiliary data returned by `fun`
+        """
+        return backend.linalg.jvp(fun, primals, tangents, has_aux=self.has_aux)
+
+    def compute_output_spec(self, fun, primals, tangents):
+        # Infer primal output spec
+        if self.has_aux:
+            primals_out_spec, aux_spec = backend.compute_output_spec(
+                fun, *primals
+            )
+        else:
+            primals_out_spec = backend.compute_output_spec(fun, *primals)
+
+        # Tangents output should match primals output in structure and shape
+        tangents_out_spec = tree.map_structure(
+            lambda x: KerasTensor(x.shape, x.dtype), primals_out_spec
+        )
+
+        if self.has_aux:
+            return primals_out_spec, tangents_out_spec, aux_spec
+        return primals_out_spec, tangents_out_spec
+
+
+@keras_export(["keras.ops.jvp", "keras.ops.linalg.jvp"])
+def jvp(fun, primals, tangents, has_aux=False):
+    """Computes a (forward-mode) Jacobian-vector product of `fun`.
+    Args:
+        fun: Function to be differentiated. Its arguments should be arrays,
+            scalars, or standard Python containers of arrays or scalars. It
+            should return an array, scalar, or standard Python container of
+            arrays or scalars.
+        primals: The primal values at which the Jacobian of `fun` should be
+                evaluated. Should be either a tuple or a list of arguments,
+                and its length should be equal to the number of positional
+                parameters of `fun`.
+        tangents: The tangent vector for which the Jacobian-vector product
+                should be evaluated. Should be either a tuple or a list of
+                tangents, with the same tree structure and array shapes as
+                `primals`.
+        has_aux: Optional, bool. Indicates whether `fun` returns a pair where
+                the first element is considered the output of the mathematical
+                function to be differentiated and the second element is
+                auxiliary data. Default is False.
+
+    Returns:
+        If `has_aux` is False, returns a (`primals_out`, `tangents_out`) pair,
+        where `primals_out` is `fun(*primals)`, and `tangents_out` is the
+        Jacobian-vector product of `fun` evaluated at `primals` with
+        `tangents`. The `tangents_out` value has the same Python tree
+        structure and shapes as `primals_out`.
+
+        If `has_aux` is True, returns a (`primals_out`, `tangents_out`, `aux`)
+        tuple where `aux` is the auxiliary data returned by `fun`.
+
+    Example:
+    >>> from keras import ops
+    >>> a1, a2 = ops.convert_to_tensor(0.1), ops.convert_to_tensor(0.2)
+    >>> primals, tangents = ops.jvp(ops.sin, (a1,), (a2,))
+    >>> primals
+    0.09983342
+    >>> tangents
+    0.19900084
+    """
+    if any_symbolic_tensors((primals, tangents)):
+        return JVP(has_aux=has_aux).symbolic_call(fun, primals, tangents)
+    return backend.linalg.jvp(fun, primals, tangents, has_aux=has_aux)