add more layers

benchmarks/vectorized_jittered_resize.py

@@ -20,6 +20,7 @@
 from tensorflow import keras
 
 from keras_cv import bounding_box
+from keras_cv.backend import random
 from keras_cv.layers import JitteredResize
 from keras_cv.layers.preprocessing.base_image_augmentation_layer import (
     BaseImageAugmentationLayer,
@@ -258,8 +259,8 @@ def test_consistency_with_old_impl(self):
 
         # makes offsets fixed to (0.5, 0.5)
         with unittest.mock.patch.object(
-            layer._random_generator,
-            "random_uniform",
+            random,
+            "uniform",
             return_value=tf.convert_to_tensor([[0.5, 0.5]]),
         ):
             output = layer(image)

benchmarks/vectorized_random_flip.py

@@ -237,14 +237,14 @@ def test_consistency_with_old_impl(self):
         )
 
         with unittest.mock.patch.object(
-            layer._random_generator,
-            "random_uniform",
+            random,
+            "uniform",
             return_value=tf.convert_to_tensor([[0.6]]),
         ):
             output = layer(image)
         with unittest.mock.patch.object(
-            old_layer._random_generator,
-            "random_uniform",
+            random,
+            "uniform",
             return_value=tf.convert_to_tensor(0.6),
         ):
             old_output = old_layer(image)

keras_cv/layers/preprocessing/random_channel_shift.py

@@ -13,6 +13,7 @@
 # limitations under the License.
 
 import tensorflow as tf
+
 from keras_cv.api_export import keras_cv_export
 from keras_cv.layers.preprocessing.base_image_augmentation_layer import (
     BaseImageAugmentationLayer,

keras_cv/layers/preprocessing/random_cutout.py

@@ -13,12 +13,14 @@
 # limitations under the License.
 
 import tensorflow as tf
+
 from keras_cv.api_export import keras_cv_export
 from keras_cv.backend import random
 from keras_cv.layers.preprocessing.base_image_augmentation_layer import (
     BaseImageAugmentationLayer,
 )
-from keras_cv.utils import fill_utils, preprocessing
+from keras_cv.utils import fill_utils
+from keras_cv.utils import preprocessing
 
 
 @keras_cv_export("keras_cv.layers.RandomCutout")

keras_cv/layers/preprocessing/random_shear.py

@@ -15,11 +15,12 @@
 import warnings
 
 import tensorflow as tf
+
 from keras_cv import bounding_box
 from keras_cv.api_export import keras_cv_export
-from keras_cv.layers.preprocessing.vectorized_base_image_augmentation_layer import (
+from keras_cv.layers.preprocessing.vectorized_base_image_augmentation_layer import (  # noqa: E501
     VectorizedBaseImageAugmentationLayer,
-)  # noqa: E501
+)
 from keras_cv.utils import preprocessing
 
 

keras_cv/layers/preprocessing/random_translation.py

@@ -13,11 +13,13 @@
 # limitations under the License.
 
 import tensorflow as tf
-from keras_cv import bounding_box, random
+
+from keras_cv import bounding_box
+from keras_cv import random
 from keras_cv.api_export import keras_cv_export
-from keras_cv.layers.preprocessing.vectorized_base_image_augmentation_layer import (
+from keras_cv.layers.preprocessing.vectorized_base_image_augmentation_layer import (  # noqa: E501
     VectorizedBaseImageAugmentationLayer,
-)  # noqa: E501
+)
 from keras_cv.utils import preprocessing as preprocessing_utils
 
 H_AXIS = -3

keras_cv/layers/preprocessing/random_zoom.py

@@ -14,12 +14,14 @@
 
 
 import tensorflow as tf
+from tensorflow.keras import backend
+
 from keras_cv.api_export import keras_cv_export
 from keras_cv.backend import random
-from keras_cv.layers.preprocessing.vectorized_base_image_augmentation_layer import \
-    VectorizedBaseImageAugmentationLayer  # noqa: E501
+from keras_cv.layers.preprocessing.vectorized_base_image_augmentation_layer import (  # noqa: E501
+    VectorizedBaseImageAugmentationLayer,
+)
 from keras_cv.utils import preprocessing as preprocessing_utils
-from tensorflow.keras import backend
 
 # In order to support both unbatched and batched inputs, the horizontal
 # and vertical axis is reverse indexed

keras_cv/layers/preprocessing/vectorized_base_image_augmentation_layer.py

@@ -101,7 +101,7 @@ def __init__(self):
     Note that since the randomness is also a common functionality, this layer
     also includes a keras_backend.RandomGenerator, which can be used to
     produce the random numbers. The random number generator is stored in the
-    `self._random_generator` attribute.
+    `self._seed_generator` attribute.
     """
 
     def __init__(self, seed=None, **kwargs):

keras_cv/layers/preprocessing/vectorized_base_image_augmentation_layer_test.py

@@ -15,6 +15,7 @@
 import tensorflow as tf
 
 from keras_cv import bounding_box
+from keras_cv.backend import random
 from keras_cv.layers.preprocessing.vectorized_base_image_augmentation_layer import (  # noqa: E501
     VectorizedBaseImageAugmentationLayer,
 )
@@ -33,8 +34,11 @@ def augment_ragged_image(self, image, transformation, **kwargs):
     def get_random_transformation_batch(self, batch_size, **kwargs):
         if self.fixed_value:
             return tf.ones((batch_size,)) * self.fixed_value
-        return self._random_generator.random_uniform(
-            (batch_size,), minval=self.add_range[0], maxval=self.add_range[1]
+        return random.uniform(
+            (batch_size,),
+            minval=self.add_range[0],
+            maxval=self.add_range[1],
+            seed=self._seed_generator,
         )
 
     def augment_images(self, images, transformations, **kwargs):
@@ -100,7 +104,7 @@ def get_random_transformation_batch(
         assert isinstance(bounding_boxes["classes"], TF_ALL_TENSOR_TYPES)
         assert isinstance(keypoints, TF_ALL_TENSOR_TYPES)
         assert isinstance(segmentation_masks, TF_ALL_TENSOR_TYPES)
-        return self._random_generator.random_uniform((batch_size,))
+        return random.uniform((batch_size,), seed=self._seed_generator)
 
     def augment_images(
         self,

keras_cv/layers/preprocessing_3d/base_augmentation_layer_3d.py

@@ -100,12 +100,15 @@ def augment_pointclouds(self, point_clouds, transformation):
     Note that since the randomness is also a common functionality, this layer
     also includes a keras.backend.RandomGenerator, which can be used to
     produce the random numbers. The random number generator is stored in the
-    `self._random_generator` attribute.
+    `self._seed_generator` attribute.
     """
 
     def __init__(self, seed=None, **kwargs):
         # To-do: remove this once th elayer is ported to keras 3
         # https://github.com/keras-team/keras-cv/issues/2136
+        self._seed_generator = SeedGenerator(
+            seed=seed,
+        )
         if config.keras_3():
             raise ValueError(
                 "This layer is not yet compatible with Keras 3."

keras_cv/layers/preprocessing_3d/base_augmentation_layer_3d_test.py

@@ -14,6 +14,7 @@
 import numpy as np
 import tensorflow as tf
 
+from keras_cv.backend import random
 from keras_cv.layers.preprocessing_3d import base_augmentation_layer_3d
 from keras_cv.tests.test_case import TestCase
 
@@ -27,14 +28,23 @@ def __init__(self, translate_noise=(0.0, 0.0, 0.0), **kwargs):
         self._translate_noise = translate_noise
 
     def get_random_transformation(self, **kwargs):
-        random_x = self._random_generator.random_normal(
-            (), mean=0.0, stddev=self._translate_noise[0]
+        random_x = random.normal(
+            (),
+            mean=0.0,
+            stddev=self._translate_noise[0],
+            seed=self._seed_generator,
         )
-        random_y = self._random_generator.random_normal(
-            (), mean=0.0, stddev=self._translate_noise[1]
+        random_y = random.normal(
+            (),
+            mean=0.0,
+            stddev=self._translate_noise[1],
+            seed=self._seed_generator,
         )
-        random_z = self._random_generator.random_normal(
-            (), mean=0.0, stddev=self._translate_noise[2]
+        random_z = random.normal(
+            (),
+            mean=0.0,
+            stddev=self._translate_noise[2],
+            seed=self._seed_generator,
         )
 
         return {

keras_cv/layers/preprocessing_3d/waymo/frustum_random_dropping_points.py

@@ -6,6 +6,7 @@
 
 from keras_cv import point_cloud
 from keras_cv.api_export import keras_cv_export
+from keras_cv.backend import random
 from keras_cv.layers.preprocessing_3d import base_augmentation_layer_3d
 
 POINT_CLOUDS = base_augmentation_layer_3d.POINT_CLOUDS
@@ -122,8 +123,11 @@ def get_random_transformation(self, point_clouds, **kwargs):
         frustum_mask = tf.concat(frustum_mask, axis=0)
         # Generate mask along point dimension.
         random_point_mask = (
-            self._random_generator.random_uniform(
-                [1, num_points, 1], minval=0.0, maxval=1
+            random.uniform(
+                [1, num_points, 1],
+                minval=0.0,
+                maxval=1,
+                seed=self._seed_generator,
             )
             < self._keep_probability
         )