[Object_Detection] Add MultiScaleAnchorGenerator and docstring for An… #2

tanzhenyu · 2020-06-15T18:13:46Z

[Object_Detection] Add MultiScaleAnchorGenerator

…chorGenerator.

bhack · 2020-06-15T18:19:36Z

Wow +611 -124 😄

bhack · 2020-06-15T18:25:38Z

kerascv/layers/anchor_generators/anchor_generator.py

+# 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):


Was this base class designed over the experience of https://github.com/tensorflow/models/blob/master/research/object_detection/core/anchor_generator.py? Or is this a from scratch design?

Good question!
This is mostly a scratch design for proof of concept at this moment. Francois and I will send out RFC for public comments and feedbacks later :-)

Ok cause there are a few derived classes at https://github.com/tensorflow/models/tree/master/research/object_detection/anchor_generators

Yep and we intend to make them layers. Though the arguments are slightly different than this directory

bhack · 2020-06-15T18:26:56Z

kerascv/layers/anchor_generators/multi_scale_anchor_generator.py

+from .anchor_generator import AnchorGenerator
+
+
+class MultiScaleAnchorGenerator(tf.keras.layers.Layer):


Same here for https://github.com/tensorflow/models/blob/master/research/object_detection/anchor_generators/multiscale_grid_anchor_generator.py#L30

tanzhenyu · 2020-06-15T18:38:04Z

Wow +611 -124 😄

Just moving files and adding comprehensive tests :-)

fchollet · 2020-06-15T21:51:56Z

kerascv/layers/anchor_generators/anchor_generator.py

+            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.]


In general, prefer fully-spelled out argument names, e.g. normalize_coordinates

fchollet · 2020-06-15T21:52:38Z

kerascv/layers/anchor_generators/multi_scale_anchor_generator.py

+from .anchor_generator import AnchorGenerator
+
+
+class MultiScaleAnchorGenerator(tf.keras.layers.Layer):


Given the similarity with AnchorGenerator why not subclass it?

This is the class that accepts a list of things and then apply to each one of the anchor generator?

fchollet · 2020-06-15T21:54:30Z

kerascv/layers/anchor_generators/anchor_generator.py

+            **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.


Nit: why anchor_stride and anchor_offset rather than just stride and offset? It's an anchor generator, after all.

Good point. Done.

fchollet · 2020-06-16T03:25:19Z

LGTM!

* add fmix init commit * fix issues try #1 * fix issues try #2 * change random generator to normal from uniform to support negative values * add support for variable channels attempt 1 * update fftfreq description * add changes * add support for variable label shapes * add changes * fix errors * remove newline * change name to FourierMix attempt 1 * fixed lint issue * remove edge softening * Final pass Co-authored-by: Luke Wood <lukewoodcs@gmail.com>

* add fmix init commit * fix issues try keras-team#1 * fix issues try keras-team#2 * change random generator to normal from uniform to support negative values * add support for variable channels attempt 1 * update fftfreq description * add changes * add support for variable label shapes * add changes * fix errors * remove newline * change name to FourierMix attempt 1 * fixed lint issue * remove edge softening * Final pass Co-authored-by: Luke Wood <lukewoodcs@gmail.com>

…d update iou losses (#1296) * first attempt at introducing YoloX * formatted and fixed bugs * cast fix #1 * cast fix #2 * cast fix #3 * cast fix #4 * adding ensure shape for support * reverting and removing ensure_shape * fixed another bug * updated train.py * updated docs, tests and added support for loss strings * first attempt at introducing YoloX * formatted and fixed bugs * adding ensure shape for support * reverting and removing ensure_shape * reformatted by black * fixed a linting issue * finally rebased atop the recent changes * finally rebased atop the new changes * fixed linting issues * reverted rebasing issues with iou loss * fixing rebased errors part 2 * fixed more linting issues * TPU testing changes * linting fixes * updated with implementation details from paper * updated based on review comments and api changes * first attempt at introducing YoloX * updated docs, tests and added support for loss strings * fixed linting issues * reverted rebasing issues with iou loss * review comments * removed examples * linting fix * fixed rebasing error * updated no_reduction warning * review comments * revert version and linting fixes

…sifier Port `ResNetV2Backbone`, `EfficientNetV2Backbone`, and `ImageClassifier` to Keras Core

Add RandomNormal and RandomUniform Initializers

…d update iou losses (keras-team#1296) * first attempt at introducing YoloX * formatted and fixed bugs * cast fix keras-team#1 * cast fix keras-team#2 * cast fix keras-team#3 * cast fix keras-team#4 * adding ensure shape for support * reverting and removing ensure_shape * fixed another bug * updated train.py * updated docs, tests and added support for loss strings * first attempt at introducing YoloX * formatted and fixed bugs * adding ensure shape for support * reverting and removing ensure_shape * reformatted by black * fixed a linting issue * finally rebased atop the recent changes * finally rebased atop the new changes * fixed linting issues * reverted rebasing issues with iou loss * fixing rebased errors part 2 * fixed more linting issues * TPU testing changes * linting fixes * updated with implementation details from paper * updated based on review comments and api changes * first attempt at introducing YoloX * updated docs, tests and added support for loss strings * fixed linting issues * reverted rebasing issues with iou loss * review comments * removed examples * linting fix * fixed rebasing error * updated no_reduction warning * review comments * revert version and linting fixes

[Object_Detection] Add MultiScaleAnchorGenerator and docstring for An…

da04ef2

…chorGenerator.

tanzhenyu requested a review from fchollet June 15, 2020 18:13

Fix config and add an additional test.

d49bd9f

bhack reviewed Jun 15, 2020

View reviewed changes

Fix config.

16ca93e

fchollet reviewed Jun 15, 2020

View reviewed changes

Address comments

fd6706b

tanzhenyu requested a review from fchollet June 15, 2020 23:02

tanzhenyu merged commit bf00322 into keras-team:master Jun 16, 2020

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Apr 13, 2022

fix issues try keras-team#2

80a8e48

martin-gorner mentioned this pull request Sep 28, 2022

RandomCrop bugs #860

Open

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Oct 30, 2022

cast fix keras-team#2

438ea20

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Nov 3, 2022

cast fix keras-team#2

f4c28bb

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Jan 20, 2023

cast fix keras-team#2

f4f1afe

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Jan 29, 2023

cast fix keras-team#2

c53f5e3

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Feb 8, 2023

cast fix keras-team#2

8f56ff1

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Mar 22, 2023

cast fix keras-team#2

5418280

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Mar 22, 2023

cast fix keras-team#2

73273e5

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Mar 27, 2023

cast fix keras-team#2

5ce3393

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Mar 29, 2023

cast fix keras-team#2

8c74d0e

quantumalaviya added a commit to quantumalaviya/keras-cv that referenced this pull request Apr 24, 2023

cast fix keras-team#2

58ff095

tirthasheshpatel pushed a commit to tirthasheshpatel/keras-cv that referenced this pull request Jun 5, 2023

Merge pull request keras-team#2 from tirthasheshpatel/core-image-clas…

9543e00

…sifier Port `ResNetV2Backbone`, `EfficientNetV2Backbone`, and `ImageClassifier` to Keras Core

freedomtan pushed a commit to freedomtan/keras-cv that referenced this pull request Jul 20, 2023

Merge pull request keras-team#2 from keras-team/add_initializers

80ab603

Add RandomNormal and RandomUniform Initializers

freedomtan pushed a commit to freedomtan/keras-cv that referenced this pull request Jul 20, 2023

Fix torch bug keras-team#2

b65de5b

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[Object_Detection] Add MultiScaleAnchorGenerator and docstring for An… #2

[Object_Detection] Add MultiScaleAnchorGenerator and docstring for An… #2

tanzhenyu commented Jun 15, 2020

bhack commented Jun 15, 2020

bhack Jun 15, 2020

tanzhenyu Jun 15, 2020

bhack Jun 15, 2020

tanzhenyu Jun 15, 2020

bhack Jun 15, 2020

tanzhenyu Jun 15, 2020

tanzhenyu commented Jun 15, 2020

fchollet Jun 15, 2020

tanzhenyu Jun 15, 2020

fchollet Jun 15, 2020

tanzhenyu Jun 15, 2020

fchollet Jun 15, 2020

tanzhenyu Jun 15, 2020

fchollet commented Jun 16, 2020

		from .anchor_generator import AnchorGenerator


		class MultiScaleAnchorGenerator(tf.keras.layers.Layer):

[Object_Detection] Add MultiScaleAnchorGenerator and docstring for An… #2

[Object_Detection] Add MultiScaleAnchorGenerator and docstring for An… #2

Conversation

tanzhenyu commented Jun 15, 2020

bhack commented Jun 15, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

tanzhenyu commented Jun 15, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

fchollet commented Jun 16, 2020