Skip to content

Revert "Type cast before normalize clip" #40

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
amyeroberts merged 1 commit into from
Sep 21, 2022
Merged

Revert "Type cast before normalize clip" #40

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
23 changes: 7 additions & 16 deletions src/transformers/models/clip/feature_extraction_clip.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,15 +108,13 @@ def __call__(
tensor. In case of a NumPy array/PyTorch tensor, each image should be of shape (C, H, W), where C is a
number of channels, H and W are image height and width.

return_tensors (`str` or [`~utils.TensorType`], *optional*, defaults to `None`):
If set, will return a tensor of a particular framework.
return_tensors (`str` or [`~utils.TensorType`], *optional*, defaults to `'np'`):
If set, will return tensors of a particular framework. Acceptable values are:

Acceptable values are:
- `'tf'`: Return TensorFlow `tf.constant` object.
- `'pt'`: Return PyTorch `torch.Tensor` object.
- `'np'`: Return NumPy `np.ndarray` object.
- `'jax'`: Return JAX `jnp.ndarray` object.
- None: Return list of `np.ndarray` objects.
- `'tf'`: Return TensorFlow `tf.constant` objects.
- `'pt'`: Return PyTorch `torch.Tensor` objects.
- `'np'`: Return NumPy `np.ndarray` objects.
- `'jax'`: Return JAX `jnp.ndarray` objects.

Returns:
[`BatchFeature`]: A [`BatchFeature`] with the following fields:
Expand Down Expand Up @@ -157,15 +155,8 @@ def __call__(
]
if self.do_center_crop and self.crop_size is not None:
images = [self.center_crop(image, self.crop_size) for image in images]

# if do_normalize=False, the casting to a numpy array won't happen, so we need to do it here
make_channel_first = True if isinstance(images[0], Image.Image) else images[0].shape[-1] in (1, 3)
images = [self.to_numpy_array(image, rescale=False, channel_first=make_channel_first) for image in images]

if self.do_normalize:
images = [
self.normalize(image=image, mean=self.image_mean, std=self.image_std, rescale=True) for image in images
]
images = [self.normalize(image=image, mean=self.image_mean, std=self.image_std) for image in images]

# return as BatchFeature
data = {"pixel_values": images}
Expand Down
102 changes: 0 additions & 102 deletions tests/models/clip/test_feature_extraction_clip.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,6 @@
import unittest

import numpy as np
from parameterized import parameterized

from transformers.testing_utils import require_torch, require_vision
from transformers.utils import is_torch_available, is_vision_available
Expand Down Expand Up @@ -233,58 +232,6 @@ def test_call_pytorch(self):
),
)

@parameterized.expand(
[
("do_resize_True_do_center_crop_True_do_normalize_True", True, True, True),
("do_resize_True_do_center_crop_True_do_normalize_False", True, True, False),
("do_resize_True_do_center_crop_False_do_normalize_True", True, False, True),
("do_resize_True_do_center_crop_False_do_normalize_False", True, False, False),
("do_resize_False_do_center_crop_True_do_normalize_True", False, True, True),
("do_resize_False_do_center_crop_True_do_normalize_False", False, True, False),
("do_resize_False_do_center_crop_False_do_normalize_True", False, False, True),
("do_resize_False_do_center_crop_False_do_normalize_False", False, False, False),
]
)
def test_call_flags(self, _, do_resize, do_center_crop, do_normalize):
# Initialize feature_extractor
feature_extractor = self.feature_extraction_class(**self.feat_extract_dict)
feature_extractor.do_center_crop = do_center_crop
feature_extractor.do_resize = do_resize
feature_extractor.do_normalize = do_normalize
# create random PIL images
image_inputs = self.feature_extract_tester.prepare_inputs(equal_resolution=False, torchify=True)

expected_shapes = [x.shape for x in image_inputs]
if do_resize:
# Same size logic inside resized
resized_shapes = []
for shape in expected_shapes:
c, h, w = shape
short, long = (w, h) if w <= h else (h, w)
min_size = self.feature_extract_tester.size
if short == min_size:
resized_shapes.append((c, h, w))
else:
short, long = min_size, int(long * min_size / short)
resized_shape = (c, long, short) if w <= h else (c, short, long)
resized_shapes.append(resized_shape)
expected_shapes = resized_shapes
if do_center_crop:
expected_shapes = [
(
self.feature_extract_tester.num_channels,
self.feature_extract_tester.crop_size,
self.feature_extract_tester.crop_size,
)
for _ in range(self.feature_extract_tester.batch_size)
]

pixel_values = feature_extractor(image_inputs, return_tensors=None)["pixel_values"]
self.assertEqual(len(pixel_values), self.feature_extract_tester.batch_size)
for idx, image in enumerate(pixel_values):
self.assertEqual(image.shape, expected_shapes[idx])
self.assertIsInstance(image, np.ndarray)


@require_torch
@require_vision
Expand Down Expand Up @@ -345,52 +292,3 @@ def test_call_pil_four_channels(self):
self.feature_extract_tester.crop_size,
),
)

@parameterized.expand(
[
("do_resize_True_do_center_crop_True_do_normalize_True", True, True, True),
("do_resize_True_do_center_crop_True_do_normalize_False", True, True, False),
("do_resize_True_do_center_crop_False_do_normalize_True", True, False, True),
("do_resize_True_do_center_crop_False_do_normalize_False", True, False, False),
("do_resize_False_do_center_crop_True_do_normalize_True", False, True, True),
("do_resize_False_do_center_crop_True_do_normalize_False", False, True, False),
("do_resize_False_do_center_crop_False_do_normalize_True", False, False, True),
("do_resize_False_do_center_crop_False_do_normalize_False", False, False, False),
]
)
def test_call_flags_four_channels(self, _, do_resize, do_center_crop, do_normalize):
# Initialize feature_extractor
feature_extractor = self.feature_extraction_class(**self.feat_extract_dict)
feature_extractor.do_center_crop = do_center_crop
feature_extractor.do_resize = do_resize
feature_extractor.do_normalize = do_normalize
# create random PIL images
# We can't currently pass in 4 channel pytorch images
image_inputs = self.feature_extract_tester.prepare_inputs(equal_resolution=False)

output_channels = self.expected_encoded_image_num_channels
crop_size = self.feature_extract_tester.crop_size
batch_size = self.feature_extract_tester.batch_size
expected_shapes = [(output_channels, *x.size[::-1]) for x in image_inputs]
if do_resize:
# Same size logic inside resized
resized_shapes = []
for shape in expected_shapes:
c, h, w = shape
short, long = (w, h) if w <= h else (h, w)
min_size = self.feature_extract_tester.size
if short == min_size:
resized_shapes.append((c, h, w))
else:
short, long = min_size, int(long * min_size / short)
resized_shape = (c, long, short) if w <= h else (c, short, long)
resized_shapes.append(resized_shape)
expected_shapes = resized_shapes
if do_center_crop:
expected_shapes = [(output_channels, crop_size, crop_size) for _ in range(batch_size)]

pixel_values = feature_extractor(image_inputs, return_tensors=None)["pixel_values"]
self.assertEqual(len(pixel_values), self.feature_extract_tester.batch_size)
for idx, image in enumerate(pixel_values):
self.assertEqual(image.shape, expected_shapes[idx])
self.assertIsInstance(image, np.ndarray)