From da04ef2eaa056b8eaf24e72290cd8d3016cef684 Mon Sep 17 00:00:00 2001
From: tanzhenyu <tanzheny@google.com>
Date: Mon, 15 Jun 2020 11:10:38 -0700
Subject: [PATCH 1/4] [Object_Detection] Add MultiScaleAnchorGenerator and
 docstring for AnchorGenerator.

---
 kerascv/layers/anchor_generators/__init__.py  |   0
 .../anchor_generator.py}                      |  46 +++-
 .../multi_scale_anchor_generator.py           | 115 +++++++++
 .../layers/anchor_generators/__init__.py      |   0
 .../anchor_generator_test.py                  | 238 ++++++++++++++++++
 .../multi_scale_anchor_generator_test.py      | 215 ++++++++++++++++
 .../kerascv/layers/anchor_generators_test.py  | 121 ---------
 7 files changed, 611 insertions(+), 124 deletions(-)
 create mode 100644 kerascv/layers/anchor_generators/__init__.py
 rename kerascv/layers/{anchor_generators.py => anchor_generators/anchor_generator.py} (64%)
 create mode 100644 kerascv/layers/anchor_generators/multi_scale_anchor_generator.py
 create mode 100644 tests/kerascv/layers/anchor_generators/__init__.py
 create mode 100644 tests/kerascv/layers/anchor_generators/anchor_generator_test.py
 create mode 100644 tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py
 delete mode 100644 tests/kerascv/layers/anchor_generators_test.py

diff --git a/kerascv/layers/anchor_generators/__init__.py b/kerascv/layers/anchor_generators/__init__.py
new file mode 100644
index 0000000000..e69de29bb2
diff --git a/kerascv/layers/anchor_generators.py b/kerascv/layers/anchor_generators/anchor_generator.py
similarity index 64%
rename from kerascv/layers/anchor_generators.py
rename to kerascv/layers/anchor_generators/anchor_generator.py
index f490e1e91b..ae24ad649e 100644
--- a/kerascv/layers/anchor_generators.py
+++ b/kerascv/layers/anchor_generators/anchor_generator.py
@@ -1,7 +1,47 @@
+# Copyright 2020 The Keras CV Authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
 import tensorflow as tf
 
 
 class AnchorGenerator(tf.keras.layers.Layer):
+    """Defines a AnchorGenerator that generates anchor boxes for a single feature map.
+
+    # Attributes:
+        image_size: A list/tuple of 2 ints, the 1st represents the image height, the 2nd image width.
+        scales: A list/tuple of positive floats (usually less than 1.) as a fraction to shorter side of `image_size`.
+            It represents the base anchor size (when aspect ratio is 1.). For example, if `image_size` is (300, 200),
+            and `scales=[.1]`, then the base anchor size is 20.
+        aspect_ratios: a list/tuple of positive floats representing the ratio of anchor width to anchor height.
+            **Must** have the same length as `scales`. For example, if `image_size=(300, 200)`, `scales=[.1]`,
+            and `aspect_ratios=[.64]`, the base anchor size is 20, then anchor height is 25 and anchor width is 16.
+            The anchor aspect ratio is independent to the original aspect ratio of image size.
+        anchor_stride: A list/tuple of 2 ints or floats representing the distance between anchor points.
+            For example, `anchor_stride=(30, 40)` means each anchor is separated by 30 pixels in height, and
+            40 pixels in width. Defaults to `None`, where anchor stride would be calculated as
+            `min(image_height, image_width) / feature_map_height` and
+            `min(image_height, image_width) / feature_map_width`.
+        anchor_offset: A list/tuple of 2 floats between [0., 1.] representing the center of anchor points relative to
+            the upper-left border of each feature map cell. Defaults to `None`, which is the center of each
+            feature map cell when `anchor_stride=None`, or center of anchor stride otherwise.
+        clip_boxes: Boolean to represents whether the anchor coordinates should be clipped to the image size.
+            Defaults to `True`.
+        norm_coords: Boolean to represents whether the anchor coordinates should be normalized to [0., 1.]
+            with respect to the image size. Defaults to `True`.
+
+    """
+
     def __init__(
         self,
         image_size,
@@ -53,7 +93,7 @@ def call(self, feature_map_size):
             anchor_offset_height = tf.cast(self.anchor_offset[0], dtype=tf.float32)
             anchor_offset_width = tf.cast(self.anchor_offset[1], dtype=tf.float32)
 
-        K = len(self.aspect_ratios)
+        len_k = len(self.aspect_ratios)
         aspect_ratios_sqrt = tf.cast(tf.sqrt(self.aspect_ratios), tf.float32)
         scales = tf.cast(self.scales, dtype=tf.float32)
         # [1, 1, K]
@@ -76,8 +116,8 @@ def call(self, feature_map_size):
         cx_grid = tf.expand_dims(cx_grid, axis=-1)
         cy_grid = tf.expand_dims(cy_grid, axis=-1)
         # [H, W, K]
-        cx_grid = tf.tile(cx_grid, (1, 1, K))
-        cy_grid = tf.tile(cy_grid, (1, 1, K))
+        cx_grid = tf.tile(cx_grid, (1, 1, len_k))
+        cy_grid = tf.tile(cy_grid, (1, 1, len_k))
         # [H, W, K]
         anchor_heights = tf.tile(
             anchor_heights, (feature_map_height, feature_map_width, 1)
diff --git a/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py b/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py
new file mode 100644
index 0000000000..5989981248
--- /dev/null
+++ b/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py
@@ -0,0 +1,115 @@
+# Copyright 2020 The Keras CV Authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import tensorflow as tf
+from .anchor_generator import AnchorGenerator
+
+
+class MultiScaleAnchorGenerator(tf.keras.layers.Layer):
+    """Defines a MultiScaleAnchorGenerator that generates anchor boxes for multiple feature maps.
+
+        # Attributes:
+            image_size: A list/tuple of 2 ints, the 1st represents the image height, the 2nd image width.
+            scales: A list/tuple of list/tuple of positive floats (usually less than 1.) as a fraction to shorter
+                side of `image_size`. It represents the base anchor size (when aspect ratio is 1.).
+                For example, if `image_size=(300, 200)`, and `scales=[[.1]]`, then the base anchor size is 20.
+                If `image_size=(300, 200)` and `scales=[[.1], [.2]]`, then the base anchor sizes are 20 and 40.
+            aspect_ratios: a list/tuple of list/tuple of positive floats representing the ratio of anchor width
+                to anchor height. **Must** have the same length as `scales`.
+                For example, if `image_size=(300, 200)`, `scales=[[.1]]`, and `aspect_ratios=[[.64]]`, the base anchor
+                size is 20, then anchor height is 25 and anchor width is 16. If `image_size=(300, 200)`,
+                `scales=[[.1], [.2]]`, and `aspect_ratios=[[.64], [.1]]`, the base anchor size is 20 and 40, then
+                the anchor heights are 25 and 40, the anchor widths are 16 and 40.
+                The anchor aspect ratio is independent to the original aspect ratio of image size.
+            anchor_strides: A list/tuple of list/tuple of 2 ints or floats representing the distance between anchor
+                points. For example, `anchor_stride=[(30, 40)]` means each anchor is separated by 30 pixels in height,
+                and 40 pixels in width. Defaults to `None`, where anchor stride would be calculated as
+                `min(image_height, image_width) / feature_map_height` and
+                `min(image_height, image_width) / feature_map_width` for each feature map.
+            anchor_offsets: A list/tuple of list/tuple of 2 floats between [0., 1.] representing the center of anchor
+                points relative to the upper-left border of each feature map cell. Defaults to `None`, which is the
+                center of each feature map cell when `anchor_strides=None`, or center of each anchor stride otherwise.
+            clip_boxes: Boolean to represents whether the anchor coordinates should be clipped to the image size.
+                Defaults to `True`.
+            norm_coords: Boolean to represents whether the anchor coordinates should be normalized to [0., 1.]
+                with respect to the image size. Defaults to `True`.
+
+    """
+
+    def __init__(
+        self,
+        image_size,
+        scales,
+        aspect_ratios,
+        anchor_strides=None,
+        anchor_offsets=None,
+        clip_boxes=True,
+        norm_coord=True,
+        name=None,
+        **kwargs
+    ):
+        self.image_size = image_size
+        self.image_height = image_size[0]
+        self.image_width = image_size[1]
+        self.scales = scales
+        self.aspect_ratios = aspect_ratios
+        if anchor_strides is None:
+            anchor_strides = [None] * len(scales)
+        if anchor_offsets is None:
+            anchor_offsets = [None] * len(scales)
+        self.anchor_strides = anchor_strides
+        self.anchor_offsets = anchor_offsets
+        self.clip_boxes = clip_boxes
+        self.norm_coord = norm_coord
+        self.anchor_generators = []
+        for (
+            i,
+            (scale_list, aspect_ratio_list, anchor_stride, anchor_offset),
+        ) in enumerate(zip(scales, aspect_ratios, anchor_strides, anchor_offsets)):
+            self.anchor_generators.append(
+                AnchorGenerator(
+                    image_size,
+                    scales=scale_list,
+                    aspect_ratios=aspect_ratio_list,
+                    anchor_stride=anchor_stride,
+                    anchor_offset=anchor_offset,
+                    clip_boxes=clip_boxes,
+                    norm_coord=norm_coord,
+                    name="anchor_generator_" + str(i),
+                )
+            )
+        super(MultiScaleAnchorGenerator, self).__init__(name=name, **kwargs)
+
+    def call(self, feature_map_sizes):
+        result = []
+        for feature_map_size, anchor_generator in zip(
+            feature_map_sizes, self.anchor_generators
+        ):
+            anchors = anchor_generator(feature_map_size)
+            anchors = tf.reshape(anchors, (-1, 4))
+            result.append(anchors)
+        return tf.concat(result, axis=0)
+
+    def get_config(self):
+        config = {
+            "image_size": self.image_size,
+            "scales": self.scales,
+            "aspect_ratios": self.aspect_ratios,
+            "anchor_stride": self.anchor_stride,
+            "anchor_offset": self.anchor_offset,
+            "clip_boxes": self.clip_boxes,
+            "norm_coord": self.norm_coord,
+        }
+        base_config = super(MultiScaleAnchorGenerator, self).get_config()
+        return dict(list(base_config.items()) + list(config.items()))
diff --git a/tests/kerascv/layers/anchor_generators/__init__.py b/tests/kerascv/layers/anchor_generators/__init__.py
new file mode 100644
index 0000000000..e69de29bb2
diff --git a/tests/kerascv/layers/anchor_generators/anchor_generator_test.py b/tests/kerascv/layers/anchor_generators/anchor_generator_test.py
new file mode 100644
index 0000000000..5ab7cbb84a
--- /dev/null
+++ b/tests/kerascv/layers/anchor_generators/anchor_generator_test.py
@@ -0,0 +1,238 @@
+# Copyright 2020 The Keras CV Authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import numpy as np
+from kerascv.layers.anchor_generators import anchor_generator
+
+
+def test_single_scale_absolute_coordinate():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.2],
+        aspect_ratios=[1.0],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen((2, 2))
+    expected_out = np.asarray(
+        [
+            [45, 45, 105, 105],
+            [45, 195, 105, 255],
+            [195, 45, 255, 105],
+            [195, 195, 255, 255],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_single_scale_non_square_image():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 200),
+        scales=[0.2],
+        aspect_ratios=[1.0],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen((2, 2))
+    expected_out = np.asarray(
+        [[30, 30, 70, 70], [30, 130, 70, 170], [130, 30, 170, 70], [130, 130, 170, 170]]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_single_scale_normalized_coordinate():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.2],
+        aspect_ratios=[1.0],
+        clip_boxes=False,
+        norm_coord=True,
+    )
+    anchor_out = anchor_gen((2, 2))
+    expected_out = np.asarray(
+        [
+            [0.15, 0.15, 0.35, 0.35],
+            [0.15, 0.65, 0.35, 0.85],
+            [0.65, 0.15, 0.85, 0.35],
+            [0.65, 0.65, 0.85, 0.85],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_single_scale_customized_anchor_stride():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.2],
+        aspect_ratios=[1.0],
+        anchor_stride=[100, 100],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen((2, 2))
+    # The center of absolute anchor points would be [50, 50], [50, 150], [150, 50] and [150, 150]
+    expected_out = np.asarray(
+        [[20, 20, 80, 80], [20, 120, 80, 180], [120, 20, 180, 80], [120, 120, 180, 180]]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_single_scale_customized_anchor_offset():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.2],
+        aspect_ratios=[1.0],
+        anchor_offset=[0.3, 0.3],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen((2, 2))
+    # The first center of absolute anchor points would be 300 / 2 * 0.3 = 45, the second would be 45 + 150 = 195
+    expected_out = np.asarray(
+        [[15, 15, 75, 75], [15, 165, 75, 225], [165, 15, 225, 75], [165, 165, 225, 225]]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_over_scale_absolute_coordinate_no_clip():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.7],
+        aspect_ratios=[1.0],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen((2, 2))
+    expected_out = np.asarray(
+        [
+            [-30, -30, 180, 180],
+            [-30, 120, 180, 330],
+            [120, -30, 330, 180],
+            [120, 120, 330, 330],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_over_scale_absolute_coordinate_clip():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.7],
+        aspect_ratios=[1.0],
+        clip_boxes=True,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen((2, 2))
+    expected_out = np.asarray(
+        [[0, 0, 180, 180], [0, 120, 180, 300], [120, 0, 300, 180], [120, 120, 300, 300]]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_over_scale_normalized_coordinate_no_clip():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.7],
+        aspect_ratios=[1.0],
+        clip_boxes=False,
+        norm_coord=True,
+    )
+    anchor_out = anchor_gen((2, 2))
+    expected_out = np.asarray(
+        [
+            [-0.1, -0.1, 0.6, 0.6],
+            [-0.1, 0.4, 0.6, 1.1],
+            [0.4, -0.1, 1.1, 0.6],
+            [0.4, 0.4, 1.1, 1.1],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_over_scale_normalized_coordinate_clip():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.7],
+        aspect_ratios=[1.0],
+        clip_boxes=True,
+        norm_coord=True,
+    )
+    anchor_out = anchor_gen((2, 2))
+    expected_out = np.asarray(
+        [
+            [0.0, 0.0, 0.6, 0.6],
+            [0.0, 0.4, 0.6, 1.0],
+            [0.4, 0.0, 1.0, 0.6],
+            [0.4, 0.4, 1.0, 1.0],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_multi_aspect_ratios():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.2, 0.2],
+        aspect_ratios=[0.64, 1.0],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen((2, 2))
+    # height is 300 * 0.2 / 0.8 = 75; width is 300 * 0.2 * 0.8 = 48
+    expected_out = np.asarray(
+        [
+            [37.5, 51.0, 112.5, 99.0],
+            [45.0, 45.0, 105.0, 105.0],
+            [37.5, 201.0, 112.5, 249.0],
+            [45.0, 195.0, 105.0, 255.0],
+            [187.5, 51.0, 262.5, 99.0],
+            [195.0, 45.0, 255.0, 105.0],
+            [187.5, 201.0, 262.5, 249.0],
+            [195.0, 195.0, 255.0, 255.0],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_multi_scales():
+    anchor_gen = anchor_generator.AnchorGenerator(
+        image_size=(300, 300),
+        scales=[0.2, 0.5],
+        aspect_ratios=[1.0, 1.0],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen((2, 2))
+    expected_out = np.asarray(
+        [
+            [45.0, 45.0, 105.0, 105.0],
+            [0.0, 0.0, 150.0, 150.0],
+            [45.0, 195.0, 105.0, 255.0],
+            [0.0, 150.0, 150.0, 300.0],
+            [195.0, 45.0, 255.0, 105.0],
+            [150.0, 0.0, 300.0, 150.0],
+            [195.0, 195.0, 255.0, 255.0],
+            [150.0, 150.0, 300.0, 300.0],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_config_with_custom_name():
+    layer = anchor_generator.AnchorGenerator(
+        (300, 300), [1.0], [1.0], name="anchor_generator"
+    )
+    config = layer.get_config()
+    layer_1 = anchor_generator.AnchorGenerator.from_config(config)
+    np.testing.assert_equal(layer_1.name, layer.name)
diff --git a/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py b/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py
new file mode 100644
index 0000000000..db1ca00304
--- /dev/null
+++ b/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py
@@ -0,0 +1,215 @@
+# Copyright 2020 The Keras CV Authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import numpy as np
+from kerascv.layers.anchor_generators import multi_scale_anchor_generator
+
+
+def test_single_feature_map_absolute_coordinate():
+    anchor_gen = multi_scale_anchor_generator.MultiScaleAnchorGenerator(
+        image_size=(300, 300),
+        scales=[[0.2]],
+        aspect_ratios=[[1.0]],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen([(2, 2)])
+    expected_out = np.asarray(
+        [
+            [45, 45, 105, 105],
+            [45, 195, 105, 255],
+            [195, 45, 255, 105],
+            [195, 195, 255, 255],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_single_feature_map_multi_aspect_ratios():
+    anchor_gen = multi_scale_anchor_generator.MultiScaleAnchorGenerator(
+        image_size=(300, 300),
+        scales=[[0.2, 0.2]],
+        aspect_ratios=[[0.64, 1.0]],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen([(2, 2)])
+    # height is 300 * 0.2 / 0.8 = 75; width is 300 * 0.2 * 0.8 = 48
+    expected_out = np.asarray(
+        [
+            [37.5, 51.0, 112.5, 99.0],
+            [45.0, 45.0, 105.0, 105.0],
+            [37.5, 201.0, 112.5, 249.0],
+            [45.0, 195.0, 105.0, 255.0],
+            [187.5, 51.0, 262.5, 99.0],
+            [195.0, 45.0, 255.0, 105.0],
+            [187.5, 201.0, 262.5, 249.0],
+            [195.0, 195.0, 255.0, 255.0],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_multi_feature_maps_absolute_coordinate():
+    anchor_gen = multi_scale_anchor_generator.MultiScaleAnchorGenerator(
+        image_size=(300, 300),
+        scales=[[0.1], [0.2]],
+        aspect_ratios=[[1.0], [1.0]],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen([(3, 3), (2, 2)])
+    # The first height and width is 30, the second height and width is 60.
+    expected_out = np.asarray(
+        [
+            [35.0, 35.0, 65.0, 65.0],
+            [35.0, 135.0, 65.0, 165.0],
+            [35.0, 235.0, 65.0, 265.0],
+            [135.0, 35.0, 165.0, 65.0],
+            [135.0, 135.0, 165.0, 165.0],
+            [135.0, 235.0, 165.0, 265.0],
+            [235.0, 35.0, 265.0, 65.0],
+            [235.0, 135.0, 265.0, 165.0],
+            [235.0, 235.0, 265.0, 265.0],
+            [45.0, 45.0, 105.0, 105.0],
+            [45.0, 195.0, 105.0, 255.0],
+            [195.0, 45.0, 255.0, 105.0],
+            [195.0, 195.0, 255.0, 255.0],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_multi_feature_maps_customized_stride():
+    anchor_gen = multi_scale_anchor_generator.MultiScaleAnchorGenerator(
+        image_size=(300, 300),
+        scales=[[0.1], [0.2]],
+        aspect_ratios=[[1.0], [1.0]],
+        anchor_strides=[[120, 120], [160, 160]],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen([(3, 3), (2, 2)])
+    # The first center of anchor point for the first feature map is 120 * 0.5 = 60, then 180
+    # The first center of anchor point for the second feature map is 160 * 0.5 = 80, then 240
+    expected_out = np.asarray(
+        [
+            [45.0, 45.0, 75.0, 75.0],
+            [45.0, 165.0, 75.0, 195.0],
+            [45.0, 285.0, 75.0, 315.0],
+            [165.0, 45.0, 195.0, 75.0],
+            [165.0, 165.0, 195.0, 195.0],
+            [165.0, 285.0, 195.0, 315.0],
+            [285.0, 45.0, 315.0, 75.0],
+            [285.0, 165.0, 315.0, 195.0],
+            [285.0, 285.0, 315.0, 315.0],
+            [50.0, 50.0, 110.0, 110.0],
+            [50.0, 210.0, 110.0, 270.0],
+            [210.0, 50.0, 270.0, 110.0],
+            [210.0, 210.0, 270.0, 270.0],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_multi_feature_maps_customized_offset():
+    anchor_gen = multi_scale_anchor_generator.MultiScaleAnchorGenerator(
+        image_size=(300, 300),
+        scales=[[0.1], [0.2]],
+        aspect_ratios=[[1.0], [1.0]],
+        anchor_offsets=[[0.2, 0.2], [0.3, 0.3]],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen([(3, 3), (2, 2)])
+    # The first center of anchor point for the first feature map is 100 * 0.2 = 20, then 120
+    # The first center of anchor point for the second feature map is 150 * 0.3 = 45, then 195
+    expected_out = np.asarray(
+        [
+            [5.0, 5.0, 35.0, 35.0],
+            [5.0, 105.0, 35.0, 135.0],
+            [5.0, 205.0, 35.0, 235.0],
+            [105.0, 5.0, 135.0, 35.0],
+            [105.0, 105.0, 135.0, 135.0],
+            [105.0, 205.0, 135.0, 235.0],
+            [205.0, 5.0, 235.0, 35.0],
+            [205.0, 105.0, 235.0, 135.0],
+            [205.0, 205.0, 235.0, 235.0],
+            [15, 15, 75, 75],
+            [15, 165, 75, 225],
+            [165, 15, 225, 75],
+            [165, 165, 225, 225],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_multi_feature_maps_over_scale_absolute_coordinate_no_clip():
+    anchor_gen = multi_scale_anchor_generator.MultiScaleAnchorGenerator(
+        image_size=(300, 300),
+        scales=[[0.1], [0.7]],
+        aspect_ratios=[[1.0], [1.0]],
+        clip_boxes=False,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen([(3, 3), (2, 2)])
+    # The first height and width is 30, the second height and width is 60.
+    expected_out = np.asarray(
+        [
+            [35.0, 35.0, 65.0, 65.0],
+            [35.0, 135.0, 65.0, 165.0],
+            [35.0, 235.0, 65.0, 265.0],
+            [135.0, 35.0, 165.0, 65.0],
+            [135.0, 135.0, 165.0, 165.0],
+            [135.0, 235.0, 165.0, 265.0],
+            [235.0, 35.0, 265.0, 65.0],
+            [235.0, 135.0, 265.0, 165.0],
+            [235.0, 235.0, 265.0, 265.0],
+            [-30, -30, 180, 180],
+            [-30, 120, 180, 330],
+            [120, -30, 330, 180],
+            [120, 120, 330, 330],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_multi_feature_maps_over_scale_absolute_coordinate_clip():
+    anchor_gen = multi_scale_anchor_generator.MultiScaleAnchorGenerator(
+        image_size=(300, 300),
+        scales=[[0.1], [0.7]],
+        aspect_ratios=[[1.0], [1.0]],
+        clip_boxes=True,
+        norm_coord=False,
+    )
+    anchor_out = anchor_gen([(3, 3), (2, 2)])
+    # The first height and width is 30, the second height and width is 60.
+    expected_out = np.asarray(
+        [
+            [35.0, 35.0, 65.0, 65.0],
+            [35.0, 135.0, 65.0, 165.0],
+            [35.0, 235.0, 65.0, 265.0],
+            [135.0, 35.0, 165.0, 65.0],
+            [135.0, 135.0, 165.0, 165.0],
+            [135.0, 235.0, 165.0, 265.0],
+            [235.0, 35.0, 265.0, 65.0],
+            [235.0, 135.0, 265.0, 165.0],
+            [235.0, 235.0, 265.0, 265.0],
+            [0, 0, 180, 180],
+            [0, 120, 180, 300],
+            [120, 0, 300, 180],
+            [120, 120, 300, 300],
+        ]
+    ).astype(np.float32)
+    np.testing.assert_allclose(expected_out, anchor_out)
diff --git a/tests/kerascv/layers/anchor_generators_test.py b/tests/kerascv/layers/anchor_generators_test.py
deleted file mode 100644
index 9f9640bf90..0000000000
--- a/tests/kerascv/layers/anchor_generators_test.py
+++ /dev/null
@@ -1,121 +0,0 @@
-import numpy as np
-from kerascv.layers.anchor_generators import AnchorGenerator
-
-
-def test_single_scale_absolute_coordinate():
-    anchor_gen = AnchorGenerator(
-        image_size=(300, 300), scales=[.2], aspect_ratios=[1.], clip_boxes=False, norm_coord=False)
-    anchor_out = anchor_gen((2, 2))
-    expected_out = np.asarray(
-        [[45, 45, 105, 105],
-         [45, 195, 105, 255],
-         [195, 45, 255, 105],
-         [195, 195, 255, 255]]).astype(np.float32)
-    np.testing.assert_allclose(anchor_out, expected_out)
-
-
-def test_single_scale_normalized_coordinate():
-    anchor_gen = AnchorGenerator(
-        image_size=(300, 300), scales=[.2], aspect_ratios=[1.], clip_boxes=False, norm_coord=True
-    )
-    anchor_out = anchor_gen((2, 2))
-    expected_out = np.asarray(
-        [[.15, .15, .35, .35],
-         [.15, .65, .35, .85],
-         [.65, .15, .85, .35],
-         [.65, .65, .85, .85]]).astype(np.float32)
-    np.testing.assert_allclose(anchor_out, expected_out)
-
-
-def test_over_scale_absolute_coordinate_no_clip():
-    anchor_gen = AnchorGenerator(
-        image_size=(300, 300), scales=[.7], aspect_ratios=[1.], clip_boxes=False, norm_coord=False
-    )
-    anchor_out = anchor_gen((2, 2))
-    expected_out = np.asarray(
-        [[-30, -30, 180, 180],
-         [-30, 120, 180, 330],
-         [120, -30, 330, 180],
-         [120, 120, 330, 330]]).astype(np.float32)
-    np.testing.assert_allclose(anchor_out, expected_out)
-
-
-def test_over_scale_absolute_coordinate_clip():
-    anchor_gen = AnchorGenerator(
-        image_size=(300, 300), scales=[.7], aspect_ratios=[1.], clip_boxes=True, norm_coord=False
-    )
-    anchor_out = anchor_gen((2, 2))
-    expected_out = np.asarray(
-        [[0, 0, 180, 180],
-         [0, 120, 180, 300],
-         [120, 0, 300, 180],
-         [120, 120, 300, 300]]).astype(np.float32)
-    np.testing.assert_allclose(anchor_out, expected_out)
-
-
-def test_over_scale_normalized_coordinate_no_clip():
-    anchor_gen = AnchorGenerator(
-        image_size=(300, 300), scales=[.7], aspect_ratios=[1.], clip_boxes=False, norm_coord=True
-    )
-    anchor_out = anchor_gen((2, 2))
-    expected_out = np.asarray(
-        [[-0.1, -0.1, 0.6, 0.6],
-         [-0.1, 0.4, 0.6, 1.1],
-         [0.4, -0.1, 1.1, 0.6],
-         [0.4, 0.4, 1.1, 1.1]]).astype(np.float32)
-    np.testing.assert_allclose(anchor_out, expected_out)
-
-
-def test_over_scale_normalized_coordinate_clip():
-    anchor_gen = AnchorGenerator(
-        image_size=(300, 300), scales=[.7], aspect_ratios=[1.], clip_boxes=True, norm_coord=True
-    )
-    anchor_out = anchor_gen((2, 2))
-    expected_out = np.asarray(
-        [[0., 0., 0.6, 0.6],
-         [0., 0.4, 0.6, 1.0],
-         [0.4, 0., 1.0, 0.6],
-         [0.4, 0.4, 1.0, 1.0]]).astype(np.float32)
-    np.testing.assert_allclose(anchor_out, expected_out)
-
-
-def test_multi_aspect_ratios():
-    anchor_gen = AnchorGenerator(
-        image_size=(300, 300), scales=[.2, .2], aspect_ratios=[.64, 1.], clip_boxes=False, norm_coord=False
-    )
-    anchor_out = anchor_gen((2, 2))
-    # height is 300 * 0.2 / 0.8 = 75; width is 300 * 0.2 * 0.8 = 48
-    expected_out = np.asarray(
-        [[37.5, 51., 112.5,  99.],
-         [45., 45., 105., 105.],
-         [37.5, 201., 112.5, 249.],
-         [45., 195., 105., 255.],
-         [187.5, 51., 262.5, 99.],
-         [195., 45., 255., 105.],
-         [187.5, 201., 262.5, 249.],
-         [195., 195., 255., 255.]]).astype(np.float32)
-    np.testing.assert_allclose(anchor_out, expected_out)
-
-
-def test_multi_scales():
-    anchor_gen = AnchorGenerator(
-        image_size=(300, 300), scales=[.2, .5], aspect_ratios=[1., 1.], clip_boxes=False, norm_coord=False
-    )
-    anchor_out = anchor_gen((2, 2))
-    expected_out = np.asarray(
-        [[45., 45., 105., 105.],
-         [0., 0., 150., 150.],
-         [45., 195., 105., 255.],
-         [0., 150., 150., 300.],
-         [195., 45., 255., 105.],
-         [150., 0., 300., 150.],
-         [195., 195., 255., 255.],
-         [150., 150., 300., 300.]]).astype(np.float32)
-    np.testing.assert_allclose(anchor_out, expected_out)
-
-
-def test_config_with_custom_name():
-    layer = AnchorGenerator((300, 300), [1.], [1.], name='anchor_generator')
-    config = layer.get_config()
-    layer_1 = AnchorGenerator.from_config(config)
-    np.testing.assert_equal(layer_1.name, layer.name)

From d49bd9ff4e59a31987cfbbd16c37db700315c133 Mon Sep 17 00:00:00 2001
From: tanzhenyu <tanzheny@google.com>
Date: Mon, 15 Jun 2020 11:20:20 -0700
Subject: [PATCH 2/4] Fix config and add an additional test.

---
 .../multi_scale_anchor_generator_test.py                 | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py b/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py
index db1ca00304..40a17d94d0 100644
--- a/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py
+++ b/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py
@@ -213,3 +213,12 @@ def test_multi_feature_maps_over_scale_absolute_coordinate_clip():
         ]
     ).astype(np.float32)
     np.testing.assert_allclose(expected_out, anchor_out)
+
+
+def test_config_with_custom_name():
+    layer = multi_scale_anchor_generator.MultiScaleAnchorGenerator(
+        (300, 300), [[1.0]], [[1.0]], name="multi_anchor_generator"
+    )
+    config = layer.get_config()
+    layer_1 = multi_scale_anchor_generator.MultiScaleAnchorGenerator.from_config(config)
+    np.testing.assert_equal(layer_1.name, layer.name)

From 16ca93ef6adee82852d509e7787dc71bb335d232 Mon Sep 17 00:00:00 2001
From: tanzhenyu <tanzheny@google.com>
Date: Mon, 15 Jun 2020 11:34:27 -0700
Subject: [PATCH 3/4] Fix config.

---
 .../layers/anchor_generators/multi_scale_anchor_generator.py  | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py b/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py
index 5989981248..a276f1ccf1 100644
--- a/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py
+++ b/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py
@@ -106,8 +106,8 @@ def get_config(self):
             "image_size": self.image_size,
             "scales": self.scales,
             "aspect_ratios": self.aspect_ratios,
-            "anchor_stride": self.anchor_stride,
-            "anchor_offset": self.anchor_offset,
+            "anchor_strides": self.anchor_strides,
+            "anchor_offsets": self.anchor_offsets,
             "clip_boxes": self.clip_boxes,
             "norm_coord": self.norm_coord,
         }

From fd6706b7ca6ed00ec2e1b8444d363d07b4e84f1e Mon Sep 17 00:00:00 2001
From: tanzhenyu <tanzheny@google.com>
Date: Mon, 15 Jun 2020 15:47:06 -0700
Subject: [PATCH 4/4] Address comments

---
 .../anchor_generators/anchor_generator.py     | 58 +++++++++----------
 .../multi_scale_anchor_generator.py           | 49 ++++++++--------
 .../anchor_generator_test.py                  | 30 +++++-----
 .../multi_scale_anchor_generator_test.py      | 18 +++---
 4 files changed, 75 insertions(+), 80 deletions(-)

diff --git a/kerascv/layers/anchor_generators/anchor_generator.py b/kerascv/layers/anchor_generators/anchor_generator.py
index ae24ad649e..abcc71cce1 100644
--- a/kerascv/layers/anchor_generators/anchor_generator.py
+++ b/kerascv/layers/anchor_generators/anchor_generator.py
@@ -27,17 +27,17 @@ class AnchorGenerator(tf.keras.layers.Layer):
             **Must** have the same length as `scales`. For example, if `image_size=(300, 200)`, `scales=[.1]`,
             and `aspect_ratios=[.64]`, the base anchor size is 20, then anchor height is 25 and anchor width is 16.
             The anchor aspect ratio is independent to the original aspect ratio of image size.
-        anchor_stride: A list/tuple of 2 ints or floats representing the distance between anchor points.
-            For example, `anchor_stride=(30, 40)` means each anchor is separated by 30 pixels in height, and
+        stride: A list/tuple of 2 ints or floats representing the distance between anchor points.
+            For example, `stride=(30, 40)` means each anchor is separated by 30 pixels in height, and
             40 pixels in width. Defaults to `None`, where anchor stride would be calculated as
             `min(image_height, image_width) / feature_map_height` and
             `min(image_height, image_width) / feature_map_width`.
-        anchor_offset: A list/tuple of 2 floats between [0., 1.] representing the center of anchor points relative to
+        offset: A list/tuple of 2 floats between [0., 1.] representing the center of anchor points relative to
             the upper-left border of each feature map cell. Defaults to `None`, which is the center of each
-            feature map cell when `anchor_stride=None`, or center of anchor stride otherwise.
+            feature map cell when `stride=None`, or center of anchor stride otherwise.
         clip_boxes: Boolean to represents whether the anchor coordinates should be clipped to the image size.
             Defaults to `True`.
-        norm_coords: Boolean to represents whether the anchor coordinates should be normalized to [0., 1.]
+        normalize_coordinates: Boolean to represents whether the anchor coordinates should be normalized to [0., 1.]
             with respect to the image size. Defaults to `True`.
 
     """
@@ -47,10 +47,10 @@ def __init__(
         image_size,
         scales,
         aspect_ratios,
-        anchor_stride=None,
-        anchor_offset=None,
+        stride=None,
+        offset=None,
         clip_boxes=True,
-        norm_coord=True,
+        normalize_coordinates=True,
         name=None,
         **kwargs
     ):
@@ -61,10 +61,10 @@ def __init__(
         self.image_width = image_size[1]
         self.scales = scales
         self.aspect_ratios = aspect_ratios
-        self.anchor_stride = anchor_stride
-        self.anchor_offset = anchor_offset
+        self.stride = stride
+        self.offset = offset
         self.clip_boxes = clip_boxes
-        self.norm_coord = norm_coord
+        self.normalize_coordinates = normalize_coordinates
         super(AnchorGenerator, self).__init__(name=name, **kwargs)
 
     def call(self, feature_map_size):
@@ -75,23 +75,21 @@ def call(self, feature_map_size):
 
         min_image_size = tf.minimum(image_width, image_height)
 
-        if self.anchor_stride is None:
-            anchor_stride_height = tf.cast(
+        if self.stride is None:
+            stride_height = tf.cast(
                 min_image_size / feature_map_height, dtype=tf.float32
             )
-            anchor_stride_width = tf.cast(
-                min_image_size / feature_map_width, dtype=tf.float32
-            )
+            stride_width = tf.cast(min_image_size / feature_map_width, dtype=tf.float32)
         else:
-            anchor_stride_height = tf.cast(self.anchor_stride[0], dtype=tf.float32)
-            anchor_stride_width = tf.cast(self.anchor_stride[1], dtype=tf.float32)
+            stride_height = tf.cast(self.stride[0], dtype=tf.float32)
+            stride_width = tf.cast(self.stride[1], dtype=tf.float32)
 
-        if self.anchor_offset is None:
-            anchor_offset_height = tf.constant(0.5, dtype=tf.float32)
-            anchor_offset_width = tf.constant(0.5, dtype=tf.float32)
+        if self.offset is None:
+            offset_height = tf.constant(0.5, dtype=tf.float32)
+            offset_width = tf.constant(0.5, dtype=tf.float32)
         else:
-            anchor_offset_height = tf.cast(self.anchor_offset[0], dtype=tf.float32)
-            anchor_offset_width = tf.cast(self.anchor_offset[1], dtype=tf.float32)
+            offset_height = tf.cast(self.offset[0], dtype=tf.float32)
+            offset_width = tf.cast(self.offset[1], dtype=tf.float32)
 
         len_k = len(self.aspect_ratios)
         aspect_ratios_sqrt = tf.cast(tf.sqrt(self.aspect_ratios), tf.float32)
@@ -105,11 +103,9 @@ def call(self, feature_map_size):
         )
 
         # [W]
-        cx = (tf.range(feature_map_width) + anchor_offset_width) * anchor_stride_width
+        cx = (tf.range(feature_map_width) + offset_width) * stride_width
         # [H]
-        cy = (
-            tf.range(feature_map_height) + anchor_offset_height
-        ) * anchor_stride_height
+        cy = (tf.range(feature_map_height) + offset_height) * stride_height
         # [H, W]
         cx_grid, cy_grid = tf.meshgrid(cx, cy)
         # [H, W, 1]
@@ -152,7 +148,7 @@ def call(self, feature_map_size):
                 [y_min_clipped, x_min_clipped, y_max_clipped, x_max_clipped], axis=1
             )
 
-        if self.norm_coord:
+        if self.normalize_coordinates:
             box_tensor = box_tensor / tf.constant(
                 [
                     [
@@ -172,10 +168,10 @@ def get_config(self):
             "image_size": self.image_size,
             "scales": self.scales,
             "aspect_ratios": self.aspect_ratios,
-            "anchor_stride": self.anchor_stride,
-            "anchor_offset": self.anchor_offset,
+            "stride": self.stride,
+            "offset": self.offset,
             "clip_boxes": self.clip_boxes,
-            "norm_coord": self.norm_coord,
+            "normalize_coordinates": self.normalize_coordinates,
         }
         base_config = super(AnchorGenerator, self).get_config()
         return dict(list(base_config.items()) + list(config.items()))
diff --git a/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py b/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py
index a276f1ccf1..81be946de2 100644
--- a/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py
+++ b/kerascv/layers/anchor_generators/multi_scale_anchor_generator.py
@@ -32,17 +32,17 @@ class MultiScaleAnchorGenerator(tf.keras.layers.Layer):
                 `scales=[[.1], [.2]]`, and `aspect_ratios=[[.64], [.1]]`, the base anchor size is 20 and 40, then
                 the anchor heights are 25 and 40, the anchor widths are 16 and 40.
                 The anchor aspect ratio is independent to the original aspect ratio of image size.
-            anchor_strides: A list/tuple of list/tuple of 2 ints or floats representing the distance between anchor
-                points. For example, `anchor_stride=[(30, 40)]` means each anchor is separated by 30 pixels in height,
+            strides: A list/tuple of list/tuple of 2 ints or floats representing the distance between anchor
+                points. For example, `stride=[(30, 40)]` means each anchor is separated by 30 pixels in height,
                 and 40 pixels in width. Defaults to `None`, where anchor stride would be calculated as
                 `min(image_height, image_width) / feature_map_height` and
                 `min(image_height, image_width) / feature_map_width` for each feature map.
-            anchor_offsets: A list/tuple of list/tuple of 2 floats between [0., 1.] representing the center of anchor
+            offsets: A list/tuple of list/tuple of 2 floats between [0., 1.] representing the center of anchor
                 points relative to the upper-left border of each feature map cell. Defaults to `None`, which is the
-                center of each feature map cell when `anchor_strides=None`, or center of each anchor stride otherwise.
+                center of each feature map cell when `strides=None`, or center of each anchor stride otherwise.
             clip_boxes: Boolean to represents whether the anchor coordinates should be clipped to the image size.
                 Defaults to `True`.
-            norm_coords: Boolean to represents whether the anchor coordinates should be normalized to [0., 1.]
+            normalize_coordinates: Boolean to represents whether the anchor coordinates should be normalized to [0., 1.]
                 with respect to the image size. Defaults to `True`.
 
     """
@@ -52,10 +52,10 @@ def __init__(
         image_size,
         scales,
         aspect_ratios,
-        anchor_strides=None,
-        anchor_offsets=None,
+        strides=None,
+        offsets=None,
         clip_boxes=True,
-        norm_coord=True,
+        normalize_coordinates=True,
         name=None,
         **kwargs
     ):
@@ -64,28 +64,27 @@ def __init__(
         self.image_width = image_size[1]
         self.scales = scales
         self.aspect_ratios = aspect_ratios
-        if anchor_strides is None:
-            anchor_strides = [None] * len(scales)
-        if anchor_offsets is None:
-            anchor_offsets = [None] * len(scales)
-        self.anchor_strides = anchor_strides
-        self.anchor_offsets = anchor_offsets
+        if strides is None:
+            strides = [None] * len(scales)
+        if offsets is None:
+            offsets = [None] * len(scales)
+        self.strides = strides
+        self.offsets = offsets
         self.clip_boxes = clip_boxes
-        self.norm_coord = norm_coord
+        self.normalize_coordinates = normalize_coordinates
         self.anchor_generators = []
-        for (
-            i,
-            (scale_list, aspect_ratio_list, anchor_stride, anchor_offset),
-        ) in enumerate(zip(scales, aspect_ratios, anchor_strides, anchor_offsets)):
+        for (i, (scale_list, aspect_ratio_list, stride, offset)) in enumerate(
+            zip(scales, aspect_ratios, strides, offsets)
+        ):
             self.anchor_generators.append(
                 AnchorGenerator(
                     image_size,
                     scales=scale_list,
                     aspect_ratios=aspect_ratio_list,
-                    anchor_stride=anchor_stride,
-                    anchor_offset=anchor_offset,
+                    stride=stride,
+                    offset=offset,
                     clip_boxes=clip_boxes,
-                    norm_coord=norm_coord,
+                    normalize_coordinates=normalize_coordinates,
                     name="anchor_generator_" + str(i),
                 )
             )
@@ -106,10 +105,10 @@ def get_config(self):
             "image_size": self.image_size,
             "scales": self.scales,
             "aspect_ratios": self.aspect_ratios,
-            "anchor_strides": self.anchor_strides,
-            "anchor_offsets": self.anchor_offsets,
+            "strides": self.strides,
+            "offsets": self.offsets,
             "clip_boxes": self.clip_boxes,
-            "norm_coord": self.norm_coord,
+            "normalize_coordinates": self.normalize_coordinates,
         }
         base_config = super(MultiScaleAnchorGenerator, self).get_config()
         return dict(list(base_config.items()) + list(config.items()))
diff --git a/tests/kerascv/layers/anchor_generators/anchor_generator_test.py b/tests/kerascv/layers/anchor_generators/anchor_generator_test.py
index 5ab7cbb84a..cc94a7ceaf 100644
--- a/tests/kerascv/layers/anchor_generators/anchor_generator_test.py
+++ b/tests/kerascv/layers/anchor_generators/anchor_generator_test.py
@@ -22,7 +22,7 @@ def test_single_scale_absolute_coordinate():
         scales=[0.2],
         aspect_ratios=[1.0],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen((2, 2))
     expected_out = np.asarray(
@@ -42,7 +42,7 @@ def test_single_scale_non_square_image():
         scales=[0.2],
         aspect_ratios=[1.0],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen((2, 2))
     expected_out = np.asarray(
@@ -57,7 +57,7 @@ def test_single_scale_normalized_coordinate():
         scales=[0.2],
         aspect_ratios=[1.0],
         clip_boxes=False,
-        norm_coord=True,
+        normalize_coordinates=True,
     )
     anchor_out = anchor_gen((2, 2))
     expected_out = np.asarray(
@@ -71,14 +71,14 @@ def test_single_scale_normalized_coordinate():
     np.testing.assert_allclose(expected_out, anchor_out)
 
 
-def test_single_scale_customized_anchor_stride():
+def test_single_scale_customized_stride():
     anchor_gen = anchor_generator.AnchorGenerator(
         image_size=(300, 300),
         scales=[0.2],
         aspect_ratios=[1.0],
-        anchor_stride=[100, 100],
+        stride=[100, 100],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen((2, 2))
     # The center of absolute anchor points would be [50, 50], [50, 150], [150, 50] and [150, 150]
@@ -88,14 +88,14 @@ def test_single_scale_customized_anchor_stride():
     np.testing.assert_allclose(expected_out, anchor_out)
 
 
-def test_single_scale_customized_anchor_offset():
+def test_single_scale_customized_offset():
     anchor_gen = anchor_generator.AnchorGenerator(
         image_size=(300, 300),
         scales=[0.2],
         aspect_ratios=[1.0],
-        anchor_offset=[0.3, 0.3],
+        offset=[0.3, 0.3],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen((2, 2))
     # The first center of absolute anchor points would be 300 / 2 * 0.3 = 45, the second would be 45 + 150 = 195
@@ -111,7 +111,7 @@ def test_over_scale_absolute_coordinate_no_clip():
         scales=[0.7],
         aspect_ratios=[1.0],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen((2, 2))
     expected_out = np.asarray(
@@ -131,7 +131,7 @@ def test_over_scale_absolute_coordinate_clip():
         scales=[0.7],
         aspect_ratios=[1.0],
         clip_boxes=True,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen((2, 2))
     expected_out = np.asarray(
@@ -146,7 +146,7 @@ def test_over_scale_normalized_coordinate_no_clip():
         scales=[0.7],
         aspect_ratios=[1.0],
         clip_boxes=False,
-        norm_coord=True,
+        normalize_coordinates=True,
     )
     anchor_out = anchor_gen((2, 2))
     expected_out = np.asarray(
@@ -166,7 +166,7 @@ def test_over_scale_normalized_coordinate_clip():
         scales=[0.7],
         aspect_ratios=[1.0],
         clip_boxes=True,
-        norm_coord=True,
+        normalize_coordinates=True,
     )
     anchor_out = anchor_gen((2, 2))
     expected_out = np.asarray(
@@ -186,7 +186,7 @@ def test_multi_aspect_ratios():
         scales=[0.2, 0.2],
         aspect_ratios=[0.64, 1.0],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen((2, 2))
     # height is 300 * 0.2 / 0.8 = 75; width is 300 * 0.2 * 0.8 = 48
@@ -211,7 +211,7 @@ def test_multi_scales():
         scales=[0.2, 0.5],
         aspect_ratios=[1.0, 1.0],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen((2, 2))
     expected_out = np.asarray(
diff --git a/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py b/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py
index 40a17d94d0..1f7ea0b979 100644
--- a/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py
+++ b/tests/kerascv/layers/anchor_generators/multi_scale_anchor_generator_test.py
@@ -22,7 +22,7 @@ def test_single_feature_map_absolute_coordinate():
         scales=[[0.2]],
         aspect_ratios=[[1.0]],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen([(2, 2)])
     expected_out = np.asarray(
@@ -42,7 +42,7 @@ def test_single_feature_map_multi_aspect_ratios():
         scales=[[0.2, 0.2]],
         aspect_ratios=[[0.64, 1.0]],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen([(2, 2)])
     # height is 300 * 0.2 / 0.8 = 75; width is 300 * 0.2 * 0.8 = 48
@@ -67,7 +67,7 @@ def test_multi_feature_maps_absolute_coordinate():
         scales=[[0.1], [0.2]],
         aspect_ratios=[[1.0], [1.0]],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen([(3, 3), (2, 2)])
     # The first height and width is 30, the second height and width is 60.
@@ -96,9 +96,9 @@ def test_multi_feature_maps_customized_stride():
         image_size=(300, 300),
         scales=[[0.1], [0.2]],
         aspect_ratios=[[1.0], [1.0]],
-        anchor_strides=[[120, 120], [160, 160]],
+        strides=[[120, 120], [160, 160]],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen([(3, 3), (2, 2)])
     # The first center of anchor point for the first feature map is 120 * 0.5 = 60, then 180
@@ -128,9 +128,9 @@ def test_multi_feature_maps_customized_offset():
         image_size=(300, 300),
         scales=[[0.1], [0.2]],
         aspect_ratios=[[1.0], [1.0]],
-        anchor_offsets=[[0.2, 0.2], [0.3, 0.3]],
+        offsets=[[0.2, 0.2], [0.3, 0.3]],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen([(3, 3), (2, 2)])
     # The first center of anchor point for the first feature map is 100 * 0.2 = 20, then 120
@@ -161,7 +161,7 @@ def test_multi_feature_maps_over_scale_absolute_coordinate_no_clip():
         scales=[[0.1], [0.7]],
         aspect_ratios=[[1.0], [1.0]],
         clip_boxes=False,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen([(3, 3), (2, 2)])
     # The first height and width is 30, the second height and width is 60.
@@ -191,7 +191,7 @@ def test_multi_feature_maps_over_scale_absolute_coordinate_clip():
         scales=[[0.1], [0.7]],
         aspect_ratios=[[1.0], [1.0]],
         clip_boxes=True,
-        norm_coord=False,
+        normalize_coordinates=False,
     )
     anchor_out = anchor_gen([(3, 3), (2, 2)])
     # The first height and width is 30, the second height and width is 60.