Add Fast Image Processor for Donut (#37081)
* add donut fast image processor support * run make style * Update src/transformers/models/donut/image_processing_donut_fast.py Co-authored-by: Parteek <parteekkamboj112@gmail.com> * update test, remove none default values * add do_align_axis = True test, fix bug in slow image processor * run make style * remove np usage * make style * Apply suggestions from code review * Update src/transformers/models/donut/image_processing_donut_fast.py Co-authored-by: Yoni Gozlan <74535834+yonigozlan@users.noreply.github.com> * add size revert in preprocess * make style * fix copies * add test for preprocess with kwargs * make style * handle None input_data_format in align_long_axis --------- Co-authored-by: Parteek <parteekkamboj112@gmail.com> Co-authored-by: Yoni Gozlan <74535834+yonigozlan@users.noreply.github.com>
This commit is contained in:
Vinh H. Pham
@@ -18,7 +18,7 @@ import unittest
|
||||
import numpy as np
|
||||
|
||||
from transformers.testing_utils import is_flaky, require_torch, require_vision
|
||||
from transformers.utils import is_torch_available, is_vision_available
|
||||
from transformers.utils import is_torch_available, is_torchvision_available, is_vision_available
|
||||
|
||||
from ...test_image_processing_common import ImageProcessingTestMixin, prepare_image_inputs
|
||||
|
||||
@@ -31,6 +31,9 @@ if is_vision_available():
|
||||
|
||||
from transformers import DonutImageProcessor
|
||||
|
||||
if is_torchvision_available():
|
||||
from transformers import DonutImageProcessorFast
|
||||
|
||||
|
||||
class DonutImageProcessingTester:
|
||||
def __init__(
|
||||
@@ -96,6 +99,7 @@ class DonutImageProcessingTester:
|
||||
@require_vision
|
||||
class DonutImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase):
|
||||
image_processing_class = DonutImageProcessor if is_vision_available() else None
|
||||
fast_image_processing_class = DonutImageProcessorFast if is_torchvision_available() else None
|
||||
|
||||
def setUp(self):
|
||||
super().setUp()
|
||||
@@ -106,122 +110,156 @@ class DonutImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase):
|
||||
return self.image_processor_tester.prepare_image_processor_dict()
|
||||
|
||||
def test_image_processor_properties(self):
|
||||
image_processing = self.image_processing_class(**self.image_processor_dict)
|
||||
self.assertTrue(hasattr(image_processing, "do_resize"))
|
||||
self.assertTrue(hasattr(image_processing, "size"))
|
||||
self.assertTrue(hasattr(image_processing, "do_thumbnail"))
|
||||
self.assertTrue(hasattr(image_processing, "do_align_long_axis"))
|
||||
self.assertTrue(hasattr(image_processing, "do_pad"))
|
||||
self.assertTrue(hasattr(image_processing, "do_normalize"))
|
||||
self.assertTrue(hasattr(image_processing, "image_mean"))
|
||||
self.assertTrue(hasattr(image_processing, "image_std"))
|
||||
for image_processing_class in self.image_processor_list:
|
||||
image_processing = image_processing_class(**self.image_processor_dict)
|
||||
self.assertTrue(hasattr(image_processing, "do_resize"))
|
||||
self.assertTrue(hasattr(image_processing, "size"))
|
||||
self.assertTrue(hasattr(image_processing, "do_thumbnail"))
|
||||
self.assertTrue(hasattr(image_processing, "do_align_long_axis"))
|
||||
self.assertTrue(hasattr(image_processing, "do_pad"))
|
||||
self.assertTrue(hasattr(image_processing, "do_normalize"))
|
||||
self.assertTrue(hasattr(image_processing, "image_mean"))
|
||||
self.assertTrue(hasattr(image_processing, "image_std"))
|
||||
|
||||
def test_image_processor_from_dict_with_kwargs(self):
|
||||
image_processor = self.image_processing_class.from_dict(self.image_processor_dict)
|
||||
self.assertEqual(image_processor.size, {"height": 18, "width": 20})
|
||||
for image_processing_class in self.image_processor_list:
|
||||
image_processor = image_processing_class.from_dict(self.image_processor_dict)
|
||||
self.assertEqual(image_processor.size, {"height": 18, "width": 20})
|
||||
|
||||
image_processor = self.image_processing_class.from_dict(self.image_processor_dict, size=42)
|
||||
self.assertEqual(image_processor.size, {"height": 42, "width": 42})
|
||||
image_processor = image_processing_class.from_dict(self.image_processor_dict, size=42)
|
||||
self.assertEqual(image_processor.size, {"height": 42, "width": 42})
|
||||
|
||||
# Previous config had dimensions in (width, height) order
|
||||
image_processor = self.image_processing_class.from_dict(self.image_processor_dict, size=(42, 84))
|
||||
self.assertEqual(image_processor.size, {"height": 84, "width": 42})
|
||||
# Previous config had dimensions in (width, height) order
|
||||
image_processor = image_processing_class.from_dict(self.image_processor_dict, size=(42, 84))
|
||||
self.assertEqual(image_processor.size, {"height": 84, "width": 42})
|
||||
|
||||
def test_image_processor_preprocess_with_kwargs(self):
|
||||
for image_processing_class in self.image_processor_list:
|
||||
# Initialize image_processing
|
||||
image_processing = image_processing_class(**self.image_processor_dict)
|
||||
# create random PyTorch tensors
|
||||
image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False, torchify=True)
|
||||
|
||||
height = 84
|
||||
width = 42
|
||||
# Previous config had dimensions in (width, height) order
|
||||
encoded_images = image_processing(image_inputs[0], size=(width, height), return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
1,
|
||||
self.image_processor_tester.num_channels,
|
||||
height,
|
||||
width,
|
||||
),
|
||||
)
|
||||
|
||||
@is_flaky()
|
||||
def test_call_pil(self):
|
||||
# Initialize image_processing
|
||||
image_processing = self.image_processing_class(**self.image_processor_dict)
|
||||
# create random PIL images
|
||||
image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False)
|
||||
for image in image_inputs:
|
||||
self.assertIsInstance(image, Image.Image)
|
||||
for image_processing_class in self.image_processor_list:
|
||||
# Initialize image_processing
|
||||
image_processing = image_processing_class(**self.image_processor_dict)
|
||||
# create random PIL images
|
||||
image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False)
|
||||
for image in image_inputs:
|
||||
self.assertIsInstance(image, Image.Image)
|
||||
|
||||
# Test not batched input
|
||||
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
1,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
# Test not batched input
|
||||
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
1,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
|
||||
# Test batched
|
||||
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
self.image_processor_tester.batch_size,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
# Test batched
|
||||
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
self.image_processor_tester.batch_size,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
|
||||
@is_flaky()
|
||||
def test_call_numpy(self):
|
||||
# Initialize image_processing
|
||||
image_processing = self.image_processing_class(**self.image_processor_dict)
|
||||
# create random numpy tensors
|
||||
image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False, numpify=True)
|
||||
for image in image_inputs:
|
||||
self.assertIsInstance(image, np.ndarray)
|
||||
for image_processing_class in self.image_processor_list:
|
||||
# Initialize image_processing
|
||||
image_processing = image_processing_class(**self.image_processor_dict)
|
||||
# create random numpy tensors
|
||||
image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False, numpify=True)
|
||||
for image in image_inputs:
|
||||
self.assertIsInstance(image, np.ndarray)
|
||||
|
||||
# Test not batched input
|
||||
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
1,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
# Test not batched input
|
||||
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
1,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
|
||||
# Test batched
|
||||
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
self.image_processor_tester.batch_size,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
# Test batched
|
||||
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
self.image_processor_tester.batch_size,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
|
||||
@is_flaky()
|
||||
def test_call_pytorch(self):
|
||||
# Initialize image_processing
|
||||
image_processing = self.image_processing_class(**self.image_processor_dict)
|
||||
# create random PyTorch tensors
|
||||
image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False, torchify=True)
|
||||
for image in image_inputs:
|
||||
self.assertIsInstance(image, torch.Tensor)
|
||||
for image_processing_class in self.image_processor_list:
|
||||
# Initialize image_processing
|
||||
image_processing = image_processing_class(**self.image_processor_dict)
|
||||
# create random PyTorch tensors
|
||||
image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False, torchify=True)
|
||||
for image in image_inputs:
|
||||
self.assertIsInstance(image, torch.Tensor)
|
||||
|
||||
# Test not batched input
|
||||
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
1,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
# Test not batched input
|
||||
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
1,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
|
||||
# Test batched
|
||||
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
self.image_processor_tester.batch_size,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
# Test batched
|
||||
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
|
||||
self.assertEqual(
|
||||
encoded_images.shape,
|
||||
(
|
||||
self.image_processor_tester.batch_size,
|
||||
self.image_processor_tester.num_channels,
|
||||
self.image_processor_tester.size["height"],
|
||||
self.image_processor_tester.size["width"],
|
||||
),
|
||||
)
|
||||
|
||||
|
||||
@require_torch
|
||||
@require_vision
|
||||
class DonutImageProcessingAlignAxisTest(DonutImageProcessingTest):
|
||||
def setUp(self):
|
||||
super().setUp()
|
||||
self.image_processor_tester = DonutImageProcessingTester(self, do_align_axis=True)
|
||||
|
||||
Reference in New Issue
Block a user