Enforce line length of 80 chars

benchmarks/vectorization_strategy_benchmark.py

@@ -72,7 +72,9 @@ def single_rectangle_mask(corners, mask_shape):
     return masks
 
 
-def fill_single_rectangle(image, centers_x, centers_y, widths, heights, fill_values):
+def fill_single_rectangle(
+    image, centers_x, centers_y, widths, heights, fill_values
+):
     """Fill rectangles with fill value into images.
 
     Args:

benchmarks/vectorized_auto_contrast.py

@@ -79,7 +79,9 @@ def augment_bounding_boxes(self, bounding_boxes, **kwargs):
     def augment_label(self, label, transformation=None, **kwargs):
         return label
 
-    def augment_segmentation_mask(self, segmentation_mask, transformation, **kwargs):
+    def augment_segmentation_mask(
+        self, segmentation_mask, transformation, **kwargs
+    ):
         return segmentation_mask
 
     def get_config(self):

benchmarks/vectorized_grayscale.py

@@ -90,7 +90,9 @@ def augment_bounding_boxes(self, bounding_boxes, **kwargs):
     def augment_label(self, label, transformation=None, **kwargs):
         return label
 
-    def augment_segmentation_mask(self, segmentation_mask, transformation, **kwargs):
+    def augment_segmentation_mask(
+        self, segmentation_mask, transformation, **kwargs
+    ):
         return segmentation_mask
 
     def get_config(self):

benchmarks/vectorized_random_brightness.py

@@ -74,10 +74,14 @@ def augment_image(self, image, transformation, **kwargs):
     def augment_label(self, label, transformation, **kwargs):
         return label
 
-    def augment_segmentation_mask(self, segmentation_mask, transformation, **kwargs):
+    def augment_segmentation_mask(
+        self, segmentation_mask, transformation, **kwargs
+    ):
         return segmentation_mask
 
-    def augment_bounding_boxes(self, bounding_boxes, transformation=None, **kwargs):
+    def augment_bounding_boxes(
+        self, bounding_boxes, transformation=None, **kwargs
+    ):
         return bounding_boxes
 
     def get_random_transformation(self, **kwargs):

benchmarks/vectorized_random_color_jitter.py

@@ -140,7 +140,9 @@ def augment_bounding_boxes(self, bounding_boxes, **kwargs):
     def augment_label(self, label, transformation=None, **kwargs):
         return label
 
-    def augment_segmentation_mask(self, segmentation_mask, transformation, **kwargs):
+    def augment_segmentation_mask(
+        self, segmentation_mask, transformation, **kwargs
+    ):
         return segmentation_mask
 
     def get_config(self):

benchmarks/vectorized_random_contrast.py

@@ -76,18 +76,24 @@ def get_random_transformation(self, **kwargs):
 
     def augment_image(self, image, transformation, **kwargs):
         contrast_factor = transformation
-        output = tf.image.adjust_contrast(image, contrast_factor=contrast_factor)
+        output = tf.image.adjust_contrast(
+            image, contrast_factor=contrast_factor
+        )
         output = tf.clip_by_value(output, 0, 255)
         output.set_shape(image.shape)
         return output
 
     def augment_label(self, label, transformation, **kwargs):
         return label
 
-    def augment_segmentation_mask(self, segmentation_mask, transformation, **kwargs):
+    def augment_segmentation_mask(
+        self, segmentation_mask, transformation, **kwargs
+    ):
         return segmentation_mask
 
-    def augment_bounding_boxes(self, bounding_boxes, transformation=None, **kwargs):
+    def augment_bounding_boxes(
+        self, bounding_boxes, transformation=None, **kwargs
+    ):
         return bounding_boxes
 
     def get_config(self):

benchmarks/vectorized_random_hue.py

@@ -86,7 +86,9 @@ def augment_bounding_boxes(self, bounding_boxes, **kwargs):
     def augment_label(self, label, transformation=None, **kwargs):
         return label
 
-    def augment_segmentation_mask(self, segmentation_mask, transformation, **kwargs):
+    def augment_segmentation_mask(
+        self, segmentation_mask, transformation, **kwargs
+    ):
         return segmentation_mask
 
     def get_config(self):

benchmarks/vectorized_random_saturation.py

@@ -74,15 +74,21 @@ def augment_image(self, image, transformation=None, **kwargs):
         # it will be handled correctly when it is a one tensor.
         transformation = tf.convert_to_tensor(transformation)
         adjust_factor = transformation / (1 - transformation)
-        return tf.image.adjust_saturation(image, saturation_factor=adjust_factor)
+        return tf.image.adjust_saturation(
+            image, saturation_factor=adjust_factor
+        )
 
-    def augment_bounding_boxes(self, bounding_boxes, transformation=None, **kwargs):
+    def augment_bounding_boxes(
+        self, bounding_boxes, transformation=None, **kwargs
+    ):
         return bounding_boxes
 
     def augment_label(self, label, transformation=None, **kwargs):
         return label
 
-    def augment_segmentation_mask(self, segmentation_mask, transformation, **kwargs):
+    def augment_segmentation_mask(
+        self, segmentation_mask, transformation, **kwargs
+    ):
         return segmentation_mask
 
     def get_config(self):
@@ -96,7 +102,9 @@ def get_config(self):
     @classmethod
     def from_config(cls, config):
         if isinstance(config["factor"], dict):
-            config["factor"] = tf.keras.utils.deserialize_keras_object(config["factor"])
+            config["factor"] = tf.keras.utils.deserialize_keras_object(
+                config["factor"]
+            )
         return cls(**config)
 
 

build_deps/configure.py

@@ -158,7 +158,11 @@ def create_build_configuration():
         write("build:windows --host_cxxopt=/std:" + get_cpp_version())
 
     if is_macos() or is_linux():
-        if not is_linux_ppc64le() and not is_linux_arm() and not is_linux_aarch64():
+        if (
+            not is_linux_ppc64le()
+            and not is_linux_arm()
+            and not is_linux_aarch64()
+        ):
             write("build --copt=-mavx")
         write("build --cxxopt=-std=" + get_cpp_version())
         write("build --host_cxxopt=-std=" + get_cpp_version())

examples/layers/object_detection/anchor_generator_configuration.py

@@ -48,5 +48,7 @@ def pair_with_anchor_boxes(inputs):
 
 if __name__ == "__main__":
     dataset = demo_utils.load_voc_dataset(bounding_box_format="xywh")
-    result = dataset.map(pair_with_anchor_boxes, num_parallel_calls=tf.data.AUTOTUNE)
+    result = dataset.map(
+        pair_with_anchor_boxes, num_parallel_calls=tf.data.AUTOTUNE
+    )
     demo_utils.visualize_data(result, bounding_box_format="xywh")

examples/layers/object_detection/demo_utils.py

@@ -29,7 +29,10 @@ def preprocess_voc(inputs, format, image_size):
         source="rel_yxyx",
         target=format,
     )
-    return {"images": inputs["image"], "bounding_boxes": inputs["objects"]["bbox"]}
+    return {
+        "images": inputs["image"],
+        "bounding_boxes": inputs["objects"]["bbox"],
+    }
 
 
 def load_voc_dataset(
@@ -40,7 +43,9 @@ def load_voc_dataset(
 ):
     dataset = tfds.load(name, split=tfds.Split.TRAIN, shuffle_files=True)
     dataset = dataset.map(
-        lambda x: preprocess_voc(x, format=bounding_box_format, image_size=image_size),
+        lambda x: preprocess_voc(
+            x, format=bounding_box_format, image_size=image_size
+        ),
         num_parallel_calls=tf.data.AUTOTUNE,
     )
     dataset = dataset.padded_batch(

examples/layers/preprocessing/bounding_box/demo_utils.py

@@ -45,7 +45,9 @@ def load_voc_dataset(
         lambda x: preprocess_voc(x, format=bounding_box_format),
         num_parallel_calls=tf.data.AUTOTUNE,
     )
-    dataset = dataset.apply(tf.data.experimental.dense_to_ragged_batch(batch_size))
+    dataset = dataset.apply(
+        tf.data.experimental.dense_to_ragged_batch(batch_size)
+    )
     return dataset
 
 

examples/layers/preprocessing/bounding_box/jittered_resize_demo.py

@@ -19,7 +19,9 @@
 
 def main():
     augment = keras_cv.layers.JitteredResize(
-        target_size=(640, 640), scale_factor=(0.75, 1.3), bounding_box_format="xywh"
+        target_size=(640, 640),
+        scale_factor=(0.75, 1.3),
+        bounding_box_format="xywh",
     )
     dataset = demo_utils.load_voc_dataset(bounding_box_format="xywh")
     dataset = dataset.map(
@@ -28,7 +30,8 @@ def main():
     demo_utils.visualize_data(dataset, bounding_box_format="xywh")
 
     dataset = dataset.map(
-        lambda x: augment(x, training=False), num_parallel_calls=tf.data.AUTOTUNE
+        lambda x: augment(x, training=False),
+        num_parallel_calls=tf.data.AUTOTUNE,
     )
     demo_utils.visualize_data(dataset, bounding_box_format="xywh")
 

examples/layers/preprocessing/bounding_box/resizing_demo.py

@@ -23,7 +23,10 @@
 def main():
     dataset = demo_utils.load_voc_dataset(bounding_box_format="xywh")
     resizing = layers.Resizing(
-        height=300, width=400, pad_to_aspect_ratio=True, bounding_box_format="xywh"
+        height=300,
+        width=400,
+        pad_to_aspect_ratio=True,
+        bounding_box_format="xywh",
     )
     dataset = dataset.map(resizing)
     demo_utils.visualize_data(dataset, bounding_box_format="xywh")

examples/layers/preprocessing/classification/rand_augment_demo.py

@@ -26,7 +26,10 @@
 def main():
     ds = demo_utils.load_oxford_dataset()
     rand_augment = preprocessing.RandAugment(
-        value_range=(0, 255), augmentations_per_image=3, magnitude=0.5, rate=0.875
+        value_range=(0, 255),
+        augmentations_per_image=3,
+        magnitude=0.5,
+        rate=0.875,
     )
     ds = ds.map(rand_augment, num_parallel_calls=tf.data.AUTOTUNE)
     demo_utils.visualize_dataset(ds)

examples/layers/preprocessing/classification/random_augmentation_pipeline_demo.py

@@ -27,7 +27,9 @@ def create_custom_pipeline():
     layers = preprocessing.RandAugment.get_standard_policy(
         value_range=(0, 255), magnitude=0.75, magnitude_stddev=0.3
     )
-    layers = layers[:4]  # slice out some layers you don't want for whatever reason
+    layers = layers[
+        :4
+    ]  # slice out some layers you don't want for whatever reason
     layers = layers + [preprocessing.GridMask()]
     return preprocessing.RandomAugmentationPipeline(
         layers=layers, augmentations_per_image=3

examples/layers/preprocessing/classification/random_channel_shift_demo.py

@@ -27,7 +27,9 @@
 
 def main():
     ds = demo_utils.load_oxford_dataset()
-    rgbshift = preprocessing.RandomChannelShift(value_range=(0, 255), factor=0.4)
+    rgbshift = preprocessing.RandomChannelShift(
+        value_range=(0, 255), factor=0.4
+    )
     ds = ds.map(rgbshift, num_parallel_calls=tf.data.AUTOTUNE)
     demo_utils.visualize_dataset(ds)
 

examples/layers/preprocessing/classification/random_color_degeneration_demo.py

@@ -25,7 +25,9 @@
 
 def main():
     ds = demo_utils.load_oxford_dataset()
-    random_color_degeneration = preprocessing.RandomColorDegeneration(factor=(0, 1.0))
+    random_color_degeneration = preprocessing.RandomColorDegeneration(
+        factor=(0, 1.0)
+    )
     ds = ds.map(random_color_degeneration, num_parallel_calls=tf.data.AUTOTUNE)
     demo_utils.visualize_dataset(ds)
 

examples/layers/preprocessing/classification/random_hue_demo.py

@@ -24,7 +24,9 @@
 
 def main():
     ds = demo_utils.load_oxford_dataset()
-    random_hue = preprocessing.RandomHue(factor=(0.0, 1.0), value_range=(0, 255))
+    random_hue = preprocessing.RandomHue(
+        factor=(0.0, 1.0), value_range=(0, 255)
+    )
     ds = ds.map(random_hue, num_parallel_calls=tf.data.AUTOTUNE)
     demo_utils.visualize_dataset(ds)
 

examples/models/generative/stable_diffusion/text_to_image.py

@@ -11,6 +11,8 @@
 from keras_cv.models import StableDiffusion
 
 model = StableDiffusion(img_height=512, img_width=512, jit_compile=True)
-img = model.text_to_image("Photograph of a beautiful horse running through a field")
+img = model.text_to_image(
+    "Photograph of a beautiful horse running through a field"
+)
 Image.fromarray(img[0]).save("horse.png")
 print("Saved at horse.png")

examples/training/classification/imagenet/basic_training.py

@@ -52,15 +52,21 @@
 flags.DEFINE_string(
     "model_name", None, "The name of the model in KerasCV.models to use."
 )
-flags.DEFINE_string("imagenet_path", None, "Directory from which to load Imagenet.")
+flags.DEFINE_string(
+    "imagenet_path", None, "Directory from which to load Imagenet."
+)
 flags.DEFINE_string(
     "backup_path", None, "Directory which will be used for training backups."
 )
 flags.DEFINE_string(
-    "weights_path", None, "Directory which will be used to store weight checkpoints."
+    "weights_path",
+    None,
+    "Directory which will be used to store weight checkpoints.",
 )
 flags.DEFINE_string(
-    "tensorboard_path", None, "Directory which will be used to store tensorboard logs."
+    "tensorboard_path",
+    None,
+    "Directory which will be used to store tensorboard logs.",
 )
 flags.DEFINE_integer(
     "batch_size",
@@ -150,7 +156,9 @@
 print("Number of accelerators: ", strategy.num_replicas_in_sync)
 
 BATCH_SIZE = FLAGS.batch_size * strategy.num_replicas_in_sync
-INITIAL_LEARNING_RATE = FLAGS.initial_learning_rate * strategy.num_replicas_in_sync
+INITIAL_LEARNING_RATE = (
+    FLAGS.initial_learning_rate * strategy.num_replicas_in_sync
+)
 """TFRecord-based tf.data.Dataset loads lazily so we can't get the length of the dataset. Temporary."""
 NUM_IMAGES = 1281167
 
@@ -245,7 +253,12 @@ def augment(img, label):
 
 
 def lr_warmup_cosine_decay(
-    global_step, warmup_steps, hold=0, total_steps=0, start_lr=0.0, target_lr=1e-2
+    global_step,
+    warmup_steps,
+    hold=0,
+    total_steps=0,
+    start_lr=0.0,
+    target_lr=1e-2,
 ):
     # Cosine decay
     learning_rate = (
@@ -269,7 +282,9 @@ def lr_warmup_cosine_decay(
             global_step > warmup_steps + hold, learning_rate, target_lr
         )
 
-    learning_rate = tf.where(global_step < warmup_steps, warmup_lr, learning_rate)
+    learning_rate = tf.where(
+        global_step < warmup_steps, warmup_lr, learning_rate
+    )
     return learning_rate
 
 
@@ -292,7 +307,9 @@ def lr_warmup_cosine_decay(
 
 
 class WarmUpCosineDecay(keras.optimizers.schedules.LearningRateSchedule):
-    def __init__(self, warmup_steps, total_steps, hold, start_lr=0.0, target_lr=1e-2):
+    def __init__(
+        self, warmup_steps, total_steps, hold, start_lr=0.0, target_lr=1e-2
+    ):
         super().__init__()
         self.start_lr = start_lr
         self.target_lr = target_lr
@@ -371,13 +388,19 @@ def __call__(self, step):
     callbacks.ModelCheckpoint(
         FLAGS.weights_path, save_weights_only=True, save_best_only=True
     ),
-    callbacks.TensorBoard(log_dir=FLAGS.tensorboard_path, write_steps_per_second=True),
+    callbacks.TensorBoard(
+        log_dir=FLAGS.tensorboard_path, write_steps_per_second=True
+    ),
 ]
 
 if FLAGS.learning_rate_schedule == REDUCE_ON_PLATEAU:
     model_callbacks.append(
         callbacks.ReduceLROnPlateau(
-            monitor="val_loss", factor=0.1, patience=10, min_delta=0.001, min_lr=0.0001
+            monitor="val_loss",
+            factor=0.1,
+            patience=10,
+            min_delta=0.001,
+            min_lr=0.0001,
         )
     )