Vectorize CropAndResize #1439

soma2000-lang · 2023-02-23T06:10:39Z

Results

keras_cv/layers/preprocessing/random_crop_and_resize_test.py

LukeWood · 2023-02-24T05:35:42Z

keras_cv/layers/preprocessing/random_crop_and_resize.py

-    def get_random_transformation(
-        self, image=None, label=None, bounding_box=None, **kwargs
+    def get_random_transformation_batch(
+        self, batch_size, label=None, bounding_box=None, **kwargs


if not used, removed label and bounding_box arguments

LukeWood · 2023-02-24T05:36:15Z

keras_cv/layers/preprocessing/random_crop_and_resize_test.py

+        input_resized = tf.image.resize(image, self.target_size)
+        output = layer(image, training=True)
+
+        self.assertNotAllClose(output, input_resized)


lets instead do:

self.assertNotAllClose(output[0], output[1])

LukeWood · 2023-02-24T18:02:46Z

@soma2000-lang looks like the unit tests are failing!

soma2000-lang · 2023-02-24T18:03:39Z

@LukeWood looking into this.Actually keras-cv is not properly installed in my system because after some system update the pip package did not seem to work.I am trying to fix the problem with pip right now.

soma2000-lang · 2023-02-25T04:47:27Z

@LukeWood I am trying to fix the pip issue with my system.Till I fix this,

I just run the test using Google Colab and it seems to pass.

Link -https://colab.research.google.com/drive/1-TZpbQDMp1fe3mrfvouWAqH_PISf9SBh?usp=sharing

it will be great if you check it once.

LukeWood · 2023-02-25T21:46:19Z

LukeWood · 2023-02-25T21:46:32Z

@soma2000-lang left a comment including a merge conflict. Can you resolve?

soma2000-lang · 2023-02-25T22:02:20Z

@LukeWood Resolved

LukeWood

almost ready! Please address remaining comments!

LukeWood · 2023-02-27T01:26:37Z

benchmarks/vectorized_crop_resize.py

@@ -98,7 +98,7 @@ def __init__(
        self.force_output_dense_images = True

    def get_random_transformation(
-        self, image=None, label=None, bounding_box=None, **kwargs
+        self, image=None,**kwargs


looks like image is also unused!

LukeWood · 2023-02-27T01:26:54Z

benchmarks/vectorized_crop_resize.py

@@ -98,7 +98,7 @@ def __init__(
        self.force_output_dense_images = True

    def get_random_transformation(
-        self, image=None, label=None, bounding_box=None, **kwargs
+        self, image=None,**kwargs


be sure to run ./shell/format.sh before requesting pull request reviews

LukeWood · 2023-03-01T18:23:39Z

keras_cv/layers/preprocessing/random_crop_and_resize_test.py

+
+        input_resized = tf.image.resize(image, self.target_size)
+        output = layer(image, training=True)
+        output[1] = input_resized


does this pass without this line?

LukeWood

by adding:


        input_resized = tf.image.resize(image, self.target_size)
        output = layer(image, training=True)
        output[1] = input_resized

you're no longer testing that the images are independently augmented. Please remove the assignment - if it fails it means the layer is not independently augmenting images.

LukeWood · 2023-03-01T18:24:54Z

keras_cv/layers/preprocessing/random_crop_and_resize_test.py

+
+        self.assertAllClose(output, input_resized)
+
+    def random_crop_and_resize_on_batched_image_training(self):


add test_ prefix or the test wont run

LukeWood · 2023-03-06T22:57:36Z

@soma2000-lang the goal is to independently augment each image. I think getting this to work will require you to update the logic a bit.

soma2000-lang · 2023-03-07T00:51:58Z

@LukeWood I made required changes.But some tests are still failing!

keras_cv/layers/preprocessing/random_crop_and_resize_test.py

LukeWood · 2023-03-07T20:14:10Z

keras_cv/layers/preprocessing/random_crop_and_resize_test.py

@@ -188,12 +188,12 @@ def test_augment_one_hot_segmentation_mask(self):
        self.assertAllClose(output["segmentation_masks"], input_mask_resized)

    def test_augment_bounding_box_single(self):
-        image = tf.zeros([20, 20, 3])
+        images = tf.zeros([20, 20, 3])


this is actually a single image

Yeah sorry ,will do the necessary

LukeWood · 2023-03-07T20:15:15Z

It looks like this is a legitimate failure - can you fix the test failures? @soma2000-lang

soma2000-lang · 2023-03-08T04:09:58Z

@LukeWood I sent you an email 2 days back .The emIl id starts with seckroll16 .I would be really greatfull if you can check out once.😊

Regards

soma2000-lang · 2023-03-12T02:42:42Z

@james77777778 It would be great if youvcan help me with this.I am stuck on this for long.

james77777778 · 2023-03-13T02:49:43Z

Hi @soma2000-lang
I have roughly read your PR and I can share the idea about self.get_random_transformation_batch

https://colab.research.google.com/drive/15ed5Fqn5zrQClmmuvB1DWrDGwMrHA4Ka?usp=sharing

I think the vectorizing could be complete by modifying self._transform_bounding_boxes and self._crop_and_resize based on batched and independently sampled transformations

soma2000-lang · 2023-03-13T02:53:19Z

@james77777778 Thanks!

LukeWood · 2023-03-14T18:37:01Z

@soma2000-lang can you add a test like the one in https://github.com/keras-team/keras-cv/pull/1480/files#diff-1c771135874332a25b437da4a5706faea85ff719331dfe7beb255403a1a66477

soma2000-lang · 2023-03-14T18:38:38Z

@LukeWood sure!.Also sorry for the delay I will try to finish this pr soon.

divyashreepathihalli · 2023-12-07T23:07:30Z

@soma2000-lang can you please rebase with master?

soma2000-lang · 2023-12-08T01:39:04Z

Sure

soma2000-lang marked this pull request as draft February 23, 2023 06:15

soma2000-lang changed the title ~~Mlpmixer~~ Vectorize CropAndResize Feb 23, 2023

soma2000-lang force-pushed the mlpmixer branch 2 times, most recently from 839cd0e to 7adadc2 Compare February 23, 2023 14:52

soma2000-lang marked this pull request as ready for review February 23, 2023 14:56

LukeWood reviewed Feb 24, 2023

View reviewed changes

keras_cv/layers/preprocessing/random_crop_and_resize_test.py Outdated Show resolved Hide resolved

soma2000-lang force-pushed the mlpmixer branch from 97e1537 to 288ad79 Compare February 24, 2023 04:00

soma2000-lang requested a review from LukeWood February 24, 2023 04:08

LukeWood reviewed Feb 24, 2023

View reviewed changes

soma2000-lang requested a review from LukeWood February 24, 2023 09:46

soma2000-lang force-pushed the mlpmixer branch from 5ae8c86 to b62a339 Compare February 25, 2023 05:02

LukeWood suggested changes Feb 26, 2023

View reviewed changes

LukeWood reviewed Feb 27, 2023

View reviewed changes

soma2000-lang force-pushed the mlpmixer branch 2 times, most recently from 9c01ece to dc56bf5 Compare March 1, 2023 02:36

soma2000-lang requested a review from LukeWood March 1, 2023 04:28

LukeWood reviewed Mar 1, 2023

View reviewed changes

LukeWood suggested changes Mar 1, 2023

View reviewed changes

LukeWood reviewed Mar 1, 2023

View reviewed changes

LukeWood reviewed Mar 7, 2023

View reviewed changes

keras_cv/layers/preprocessing/random_crop_and_resize_test.py Outdated Show resolved Hide resolved

LukeWood reviewed Mar 7, 2023

View reviewed changes

james77777778 mentioned this pull request Mar 10, 2023

vectorize KPLS: tf.vectorized_map fallback cause #291

Closed

soma2000-lang force-pushed the mlpmixer branch from 9bf474d to 8a8d452 Compare March 12, 2023 02:13

james77777778 mentioned this pull request Mar 15, 2023

Bug: VectorizedBaseImageAugmentationLayer fails to process unbatched bounding_boxes #1512

Closed

soma2000-lang force-pushed the mlpmixer branch 2 times, most recently from aff34ed to de598c8 Compare March 31, 2023 03:54

soma2000-lang added 7 commits March 31, 2023 23:42

Add

f9b1e0c

Adding

9fbda5c

Adding

66414c2

fix

649ea89

Adding

4cb2533

Adding

576f92c

Adding

6a596a1

soma2000-lang force-pushed the mlpmixer branch from c21ff55 to 6a596a1 Compare April 1, 2023 00:12

soma2000-lang added 2 commits April 1, 2023 00:33

Ading

7a88bce

Adding

ed7f012

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Vectorize CropAndResize #1439

Vectorize CropAndResize #1439

soma2000-lang commented Feb 23, 2023 •

edited

Loading

LukeWood Feb 24, 2023

LukeWood Feb 24, 2023

LukeWood commented Feb 24, 2023

soma2000-lang commented Feb 24, 2023 •

edited

Loading

soma2000-lang commented Feb 25, 2023 •

edited

Loading

LukeWood commented Feb 25, 2023

LukeWood commented Feb 25, 2023

soma2000-lang commented Feb 25, 2023

LukeWood left a comment

LukeWood Feb 27, 2023

LukeWood Feb 27, 2023

LukeWood Mar 1, 2023

LukeWood left a comment

LukeWood Mar 1, 2023

LukeWood commented Mar 6, 2023

soma2000-lang commented Mar 7, 2023 •

edited

Loading

LukeWood Mar 7, 2023

soma2000-lang Mar 8, 2023

LukeWood commented Mar 7, 2023

soma2000-lang commented Mar 8, 2023 •

edited

Loading

soma2000-lang commented Mar 12, 2023

james77777778 commented Mar 13, 2023

soma2000-lang commented Mar 13, 2023

LukeWood commented Mar 14, 2023

soma2000-lang commented Mar 14, 2023

divyashreepathihalli commented Dec 7, 2023

soma2000-lang commented Dec 8, 2023


		self.assertAllClose(output, input_resized)

		def random_crop_and_resize_on_batched_image_training(self):

Vectorize CropAndResize #1439

Are you sure you want to change the base?

Vectorize CropAndResize #1439

Conversation

soma2000-lang commented Feb 23, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

LukeWood commented Feb 24, 2023

soma2000-lang commented Feb 24, 2023 • edited Loading

soma2000-lang commented Feb 25, 2023 • edited Loading

LukeWood commented Feb 25, 2023

LukeWood commented Feb 25, 2023

soma2000-lang commented Feb 25, 2023

LukeWood left a 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

LukeWood left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

LukeWood commented Mar 6, 2023

soma2000-lang commented Mar 7, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

LukeWood commented Mar 7, 2023

soma2000-lang commented Mar 8, 2023 • edited Loading

soma2000-lang commented Mar 12, 2023

james77777778 commented Mar 13, 2023

soma2000-lang commented Mar 13, 2023

LukeWood commented Mar 14, 2023

soma2000-lang commented Mar 14, 2023

divyashreepathihalli commented Dec 7, 2023

soma2000-lang commented Dec 8, 2023

soma2000-lang commented Feb 23, 2023 •

edited

Loading

soma2000-lang commented Feb 24, 2023 •

edited

Loading

soma2000-lang commented Feb 25, 2023 •

edited

Loading

soma2000-lang commented Mar 7, 2023 •

edited

Loading

soma2000-lang commented Mar 8, 2023 •

edited

Loading