[1/3~] Begin cleaning up ops namespace (keras-team#1299)

* Begin cleaning up ops namespace

* API start updates

* Fix imports

* move util

Author: LukeWood

keras_cv/layers/__init__.py

@@ -21,6 +21,7 @@
 from keras_cv.layers.fusedmbconv import FusedMBConvBlock
 from keras_cv.layers.mbconv import MBConvBlock
 from keras_cv.layers.object_detection.anchor_generator import AnchorGenerator
+from keras_cv.layers.object_detection.box_matcher import BoxMatcher
 from keras_cv.layers.object_detection.multi_class_non_max_suppression import (
     MultiClassNonMaxSuppression,
 )

keras_cv/ops/box_matcher.py renamed to keras_cv/layers/object_detection/box_matcher.py

@@ -11,17 +11,14 @@
 # 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.
-
-"Argmax-based box matching"
-
 from typing import List
 from typing import Tuple
 
 import tensorflow as tf
 
 
 @tf.keras.utils.register_keras_serializable(package="keras_cv")
-class ArgmaxBoxMatcher(tf.keras.layers.Layer):
+class BoxMatcher(tf.keras.layers.Layer):
     """Box matching logic based on argmax of highest value (e.g., IOU).
 
     This class computes matches from a similarity matrix. Each row will be
@@ -69,8 +66,8 @@ class ArgmaxBoxMatcher(tf.keras.layers.Layer):
     Usage:
 
     ```python
-    box_matcher = keras_cv.ops.ArgmaxBoxMatcher([0.3, 0.7], [-1, 0, 1])
-    iou_metric = keras_cv.bounding_box.compute_iou(anchors, gt_boxes)
+    box_matcher = keras_cv.layers.BoxMatcher([0.3, 0.7], [-1, 0, 1])
+    iou_metric = keras_cv.bounding_box.compute_iou(anchors, boxes)
     matched_columns, matched_match_values = box_matcher(iou_metric)
     cls_mask = tf.less_equal(matched_match_values, 0)
     ```
@@ -135,7 +132,7 @@ def _match_when_cols_are_empty():
                   storing the match type indicator (e.g. positive or negative
                   or ignored match).
             """
-            with tf.name_scope("empty_gt_boxes"):
+            with tf.name_scope("empty_boxes"):
                 matched_columns = tf.zeros([batch_size, num_rows], dtype=tf.int32)
                 matched_values = -tf.ones([batch_size, num_rows], dtype=tf.int32)
                 return matched_columns, matched_values
@@ -149,7 +146,7 @@ def _match_when_cols_are_non_empty():
                   storing the match type indicator (e.g. positive or negative
                   or ignored match).
             """
-            with tf.name_scope("non_empty_gt_boxes"):
+            with tf.name_scope("non_empty_boxes"):
                 matched_columns = tf.argmax(
                     similarity_matrix, axis=-1, output_type=tf.int32
                 )
@@ -207,11 +204,11 @@ def _match_when_cols_are_non_empty():
 
                 return matched_columns, matched_values
 
-        num_gt_boxes = (
+        num_boxes = (
             similarity_matrix.shape.as_list()[-1] or tf.shape(similarity_matrix)[-1]
         )
         matched_columns, matched_values = tf.cond(
-            pred=tf.greater(num_gt_boxes, 0),
+            pred=tf.greater(num_boxes, 0),
             true_fn=_match_when_cols_are_non_empty,
             false_fn=_match_when_cols_are_empty,
         )

keras_cv/ops/box_matcher_test.py renamed to keras_cv/layers/object_detection/box_matcher_test.py

@@ -14,17 +14,17 @@
 
 import tensorflow as tf
 
-from keras_cv.ops.box_matcher import ArgmaxBoxMatcher
+from keras_cv.layers.object_detection.box_matcher import BoxMatcher
 
 
-class ArgmaxBoxMatcherTest(tf.test.TestCase):
+class BoxMatcherTest(tf.test.TestCase):
     def test_box_matcher_invalid_length(self):
         fg_threshold = 0.5
         bg_thresh_hi = 0.2
         bg_thresh_lo = 0.0
 
         with self.assertRaisesRegex(ValueError, "must be len"):
-            _ = ArgmaxBoxMatcher(
+            _ = BoxMatcher(
                 thresholds=[bg_thresh_lo, bg_thresh_hi, fg_threshold],
                 match_values=[-3, -2, -1],
             )
@@ -35,7 +35,7 @@ def test_box_matcher_unsorted_thresholds(self):
         bg_thresh_lo = 0.0
 
         with self.assertRaisesRegex(ValueError, "must be sorted"):
-            _ = ArgmaxBoxMatcher(
+            _ = BoxMatcher(
                 thresholds=[bg_thresh_hi, bg_thresh_lo, fg_threshold],
                 match_values=[-3, -2, -1, 1],
             )
@@ -47,7 +47,7 @@ def test_box_matcher_unbatched(self):
         bg_thresh_hi = 0.2
         bg_thresh_lo = 0.0
 
-        matcher = ArgmaxBoxMatcher(
+        matcher = BoxMatcher(
             thresholds=[bg_thresh_lo, bg_thresh_hi, fg_threshold],
             match_values=[-3, -2, -1, 1],
         )
@@ -67,7 +67,7 @@ def test_box_matcher_batched(self):
         bg_thresh_hi = 0.2
         bg_thresh_lo = 0.0
 
-        matcher = ArgmaxBoxMatcher(
+        matcher = BoxMatcher(
             thresholds=[bg_thresh_lo, bg_thresh_hi, fg_threshold],
             match_values=[-3, -2, -1, 1],
         )
@@ -90,7 +90,7 @@ def test_box_matcher_force_match(self):
         bg_thresh_hi = 0.2
         bg_thresh_lo = 0.0
 
-        matcher = ArgmaxBoxMatcher(
+        matcher = BoxMatcher(
             thresholds=[bg_thresh_lo, bg_thresh_hi, fg_threshold],
             match_values=[-3, -2, -1, 1],
             force_match_for_each_col=True,
@@ -113,7 +113,7 @@ def test_box_matcher_empty_gt_boxes(self):
         bg_thresh_hi = 0.2
         bg_thresh_lo = 0.0
 
-        matcher = ArgmaxBoxMatcher(
+        matcher = BoxMatcher(
             thresholds=[bg_thresh_lo, bg_thresh_hi, fg_threshold],
             match_values=[-3, -2, -1, 1],
         )

keras_cv/layers/object_detection/retina_net_label_encoder.py

@@ -16,8 +16,8 @@
 from tensorflow.keras import layers
 
 from keras_cv import bounding_box
-from keras_cv.ops import box_matcher
-from keras_cv.ops import target_gather
+from keras_cv.layers.object_detection import box_matcher
+from keras_cv.utils import target_gather
 
 
 class RetinaNetLabelEncoder(layers.Layer):
@@ -66,7 +66,7 @@ def __init__(
         )
         self.positive_threshold = positive_threshold
         self.negative_threshold = negative_threshold
-        self.box_matcher = box_matcher.ArgmaxBoxMatcher(
+        self.box_matcher = box_matcher.BoxMatcher(
             thresholds=[negative_threshold, positive_threshold],
             match_values=[-1, -2, 1],
             force_match_for_each_col=False,

keras_cv/layers/object_detection/roi_sampler.py

@@ -16,9 +16,9 @@
 
 from keras_cv import bounding_box
 from keras_cv.bounding_box import iou
-from keras_cv.ops import box_matcher
-from keras_cv.ops import sampling
-from keras_cv.ops import target_gather
+from keras_cv.layers.object_detection import box_matcher
+from keras_cv.layers.object_detection import sampling
+from keras_cv.layers.object_detection import target_gather
 
 
 @tf.keras.utils.register_keras_serializable(package="keras_cv")
@@ -44,7 +44,7 @@ class _ROISampler(tf.keras.layers.Layer):
       bounding_box_format: The format of bounding boxes to generate. Refer
         [to the keras.io docs](https://keras.io/api/keras_cv/bounding_box/formats/)
         for more details on supported bounding box formats.
-      roi_matcher: a `ArgmaxBoxMatcher` object that matches proposals
+      roi_matcher: a `BoxMatcher` object that matches proposals
         with ground truth boxes. the positive match must be 1 and negative match must be -1.
         Such assumption is not being validated here.
       positive_fraction: the positive ratio w.r.t `num_sampled_rois`. Defaults to 0.25.
@@ -59,7 +59,7 @@ class _ROISampler(tf.keras.layers.Layer):
     def __init__(
         self,
         bounding_box_format: str,
-        roi_matcher: box_matcher.ArgmaxBoxMatcher,
+        roi_matcher: box_matcher.BoxMatcher,
         positive_fraction: float = 0.25,
         background_class: int = 0,
         num_sampled_rois: int = 256,
@@ -205,5 +205,5 @@ def get_config(self):
     @classmethod
     def from_config(cls, config, custom_objects=None):
         roi_matcher_config = config.pop("roi_matcher")
-        roi_matcher = box_matcher.ArgmaxBoxMatcher(**roi_matcher_config)
+        roi_matcher = box_matcher.BoxMatcher(**roi_matcher_config)
         return cls(roi_matcher=roi_matcher, **config)

keras_cv/layers/object_detection/roi_sampler_test.py

@@ -14,13 +14,13 @@
 
 import tensorflow as tf
 
+from keras_cv.layers.object_detection.box_matcher import BoxMatcher
 from keras_cv.layers.object_detection.roi_sampler import _ROISampler
-from keras_cv.ops.box_matcher import ArgmaxBoxMatcher
 
 
 class ROISamplerTest(tf.test.TestCase):
     def test_roi_sampler(self):
-        box_matcher = ArgmaxBoxMatcher(thresholds=[0.3], match_values=[-1, 1])
+        box_matcher = BoxMatcher(thresholds=[0.3], match_values=[-1, 1])
         roi_sampler = _ROISampler(
             bounding_box_format="xyxy",
             roi_matcher=box_matcher,
@@ -57,7 +57,7 @@ def test_roi_sampler(self):
         )
 
     def test_roi_sampler_small_threshold(self):
-        box_matcher = ArgmaxBoxMatcher(thresholds=[0.1], match_values=[-1, 1])
+        box_matcher = BoxMatcher(thresholds=[0.1], match_values=[-1, 1])
         roi_sampler = _ROISampler(
             bounding_box_format="xyxy",
             roi_matcher=box_matcher,
@@ -106,7 +106,7 @@ def test_roi_sampler_small_threshold(self):
 
     def test_roi_sampler_large_threshold(self):
         # the 2nd roi and 2nd gt box has IOU of 0.923, setting positive_threshold to 0.95 to ignore it
-        box_matcher = ArgmaxBoxMatcher(thresholds=[0.95], match_values=[-1, 1])
+        box_matcher = BoxMatcher(thresholds=[0.95], match_values=[-1, 1])
         roi_sampler = _ROISampler(
             bounding_box_format="xyxy",
             roi_matcher=box_matcher,
@@ -139,7 +139,7 @@ def test_roi_sampler_large_threshold(self):
 
     def test_roi_sampler_large_threshold_custom_bg_class(self):
         # the 2nd roi and 2nd gt box has IOU of 0.923, setting positive_threshold to 0.95 to ignore it
-        box_matcher = ArgmaxBoxMatcher(thresholds=[0.95], match_values=[-1, 1])
+        box_matcher = BoxMatcher(thresholds=[0.95], match_values=[-1, 1])
         roi_sampler = _ROISampler(
             bounding_box_format="xyxy",
             roi_matcher=box_matcher,
@@ -173,7 +173,7 @@ def test_roi_sampler_large_threshold_custom_bg_class(self):
 
     def test_roi_sampler_large_threshold_append_gt_boxes(self):
         # the 2nd roi and 2nd gt box has IOU of 0.923, setting positive_threshold to 0.95 to ignore it
-        box_matcher = ArgmaxBoxMatcher(thresholds=[0.95], match_values=[-1, 1])
+        box_matcher = BoxMatcher(thresholds=[0.95], match_values=[-1, 1])
         roi_sampler = _ROISampler(
             bounding_box_format="xyxy",
             roi_matcher=box_matcher,
@@ -204,7 +204,7 @@ def test_roi_sampler_large_threshold_append_gt_boxes(self):
         self.assertAllGreaterEqual(tf.reduce_min(sampled_gt_classes), 0)
 
     def test_roi_sampler_large_num_sampled_rois(self):
-        box_matcher = ArgmaxBoxMatcher(thresholds=[0.95], match_values=[-1, 1])
+        box_matcher = BoxMatcher(thresholds=[0.95], match_values=[-1, 1])
         roi_sampler = _ROISampler(
             bounding_box_format="xyxy",
             roi_matcher=box_matcher,
@@ -227,7 +227,7 @@ def test_roi_sampler_large_num_sampled_rois(self):
             _, _, _ = roi_sampler(rois, gt_boxes, gt_classes)
 
     def test_serialization(self):
-        box_matcher = ArgmaxBoxMatcher(thresholds=[0.95], match_values=[-1, 1])
+        box_matcher = BoxMatcher(thresholds=[0.95], match_values=[-1, 1])
         roi_sampler = _ROISampler(
             bounding_box_format="xyxy",
             roi_matcher=box_matcher,

keras_cv/layers/object_detection/rpn_label_encoder.py

@@ -18,9 +18,9 @@
 
 from keras_cv import bounding_box
 from keras_cv.bounding_box import iou
-from keras_cv.ops import box_matcher
-from keras_cv.ops import sampling
-from keras_cv.ops import target_gather
+from keras_cv.layers.object_detection import box_matcher
+from keras_cv.layers.object_detection import sampling
+from keras_cv.utils import target_gather
 
 
 @tf.keras.utils.register_keras_serializable(package="keras_cv")
@@ -74,7 +74,7 @@ def __init__(
         self.negative_threshold = negative_threshold
         self.samples_per_image = samples_per_image
         self.positive_fraction = positive_fraction
-        self.box_matcher = box_matcher.ArgmaxBoxMatcher(
+        self.box_matcher = box_matcher.BoxMatcher(
             thresholds=[negative_threshold, positive_threshold],
             match_values=[-1, -2, 1],
             force_match_for_each_col=False,

keras_cv/ops/sampling.py renamed to keras_cv/layers/object_detection/sampling.py

@@ -14,7 +14,7 @@
 
 import tensorflow as tf
 
-from keras_cv.ops.sampling import balanced_sample
+from keras_cv.layers.object_detection.sampling import balanced_sample
 
 
 class BalancedSamplingTest(tf.test.TestCase):

keras_cv/ops/sampling_test.py renamed to keras_cv/layers/object_detection/sampling_test.py

@@ -12,16 +12,14 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from typing import Optional
-
 import tensorflow as tf
 
 
 def _target_gather(
     targets: tf.Tensor,
     indices: tf.Tensor,
-    mask: Optional[tf.Tensor] = None,
-    mask_val: Optional[float] = 0.0,
+    mask=None,
+    mask_val=0.0,
 ):
     """A utility function wrapping tf.gather, which deals with:
      1) both batched and unbatched `targets`