-
Notifications
You must be signed in to change notification settings - Fork 25.3k
/
feature_extraction_mobilevit.py
199 lines (162 loc) 路 8.54 KB
/
feature_extraction_mobilevit.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
# coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# 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.
"""Feature extractor class for MobileViT."""
from typing import List, Optional, Tuple, Union
import numpy as np
from PIL import Image
from ...feature_extraction_utils import BatchFeature, FeatureExtractionMixin
from ...image_utils import ImageFeatureExtractionMixin, ImageInput, is_torch_tensor
from ...utils import TensorType, is_torch_available, logging
if is_torch_available():
import torch
logger = logging.get_logger(__name__)
class MobileViTFeatureExtractor(FeatureExtractionMixin, ImageFeatureExtractionMixin):
r"""
Constructs a MobileViT feature extractor.
This feature extractor inherits from [`FeatureExtractionMixin`] which contains most of the main methods. Users
should refer to this superclass for more information regarding those methods.
Args:
do_resize (`bool`, *optional*, defaults to `True`):
Whether to resize the input to a certain `size`.
size (`int` or `Tuple(int)`, *optional*, defaults to 288):
Resize the input to the given size. If a tuple is provided, it should be (width, height). If only an
integer is provided, then the input will be resized to match the shorter side. Only has an effect if
`do_resize` is set to `True`.
resample (`int`, *optional*, defaults to `PIL.Image.BILINEAR`):
An optional resampling filter. This can be one of `PIL.Image.NEAREST`, `PIL.Image.BOX`,
`PIL.Image.BILINEAR`, `PIL.Image.HAMMING`, `PIL.Image.BICUBIC` or `PIL.Image.LANCZOS`. Only has an effect
if `do_resize` is set to `True`.
do_center_crop (`bool`, *optional*, defaults to `True`):
Whether to crop the input at the center. If the input size is smaller than `crop_size` along any edge, the
image is padded with 0's and then center cropped.
crop_size (`int`, *optional*, defaults to 256):
Desired output size when applying center-cropping. Only has an effect if `do_center_crop` is set to `True`.
do_flip_channel_order (`bool`, *optional*, defaults to `True`):
Whether to flip the color channels from RGB to BGR.
"""
model_input_names = ["pixel_values"]
def __init__(
self,
do_resize=True,
size=288,
resample=Image.BILINEAR,
do_center_crop=True,
crop_size=256,
do_flip_channel_order=True,
**kwargs
):
super().__init__(**kwargs)
self.do_resize = do_resize
self.size = size
self.resample = resample
self.do_center_crop = do_center_crop
self.crop_size = crop_size
self.do_flip_channel_order = do_flip_channel_order
def __call__(
self, images: ImageInput, return_tensors: Optional[Union[str, TensorType]] = None, **kwargs
) -> BatchFeature:
"""
Main method to prepare for the model one or several image(s).
<Tip warning={true}>
NumPy arrays and PyTorch tensors are converted to PIL images when resizing, so the most efficient is to pass
PIL images.
</Tip>
Args:
images (`PIL.Image.Image`, `np.ndarray`, `torch.Tensor`, `List[PIL.Image.Image]`, `List[np.ndarray]`, `List[torch.Tensor]`):
The image or batch of images to be prepared. Each image can be a PIL image, NumPy array or PyTorch
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 `'np'`):
If set, will return tensors of a particular framework. Acceptable values are:
- `'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:
- **pixel_values** -- Pixel values to be fed to a model, of shape (batch_size, num_channels, height,
width).
"""
# Input type checking for clearer error
valid_images = False
# Check that images has a valid type
if isinstance(images, (Image.Image, np.ndarray)) or is_torch_tensor(images):
valid_images = True
elif isinstance(images, (list, tuple)):
if len(images) == 0 or isinstance(images[0], (Image.Image, np.ndarray)) or is_torch_tensor(images[0]):
valid_images = True
if not valid_images:
raise ValueError(
"Images must of type `PIL.Image.Image`, `np.ndarray` or `torch.Tensor` (single example), "
"`List[PIL.Image.Image]`, `List[np.ndarray]` or `List[torch.Tensor]` (batch of examples)."
)
is_batched = bool(
isinstance(images, (list, tuple))
and (isinstance(images[0], (Image.Image, np.ndarray)) or is_torch_tensor(images[0]))
)
if not is_batched:
images = [images]
# transformations (resizing + normalization)
if self.do_resize and self.size is not None:
images = [
self.resize(image=image, size=self.size, resample=self.resample, default_to_square=False)
for image in images
]
if self.do_center_crop and self.crop_size is not None:
images = [self.center_crop(image, self.crop_size) for image in images]
images = [self.to_numpy_array(image) for image in images]
# the pretrained checkpoints assume images are BGR, not RGB
if self.do_flip_channel_order:
images = [self.flip_channel_order(image) for image in images]
# return as BatchFeature
data = {"pixel_values": images}
encoded_inputs = BatchFeature(data=data, tensor_type=return_tensors)
return encoded_inputs
def post_process_semantic_segmentation(self, outputs, target_sizes: List[Tuple] = None):
"""
Converts the output of [`MobileViTForSemanticSegmentation`] into semantic segmentation maps. Only supports
PyTorch.
Args:
outputs ([`MobileViTForSemanticSegmentation`]):
Raw outputs of the model.
target_sizes (`List[Tuple]`, *optional*):
A list of length `batch_size`, where each item is a `Tuple[int, int]` corresponding to the requested
final size (height, width) of each prediction. If left to None, predictions will not be resized.
Returns:
`List[torch.Tensor]`:
A list of length `batch_size`, where each item is a semantic segmentation map of shape (height, width)
corresponding to the target_sizes entry (if `target_sizes` is specified). Each entry of each
`torch.Tensor` correspond to a semantic class id.
"""
logits = outputs.logits
# Resize logits and compute semantic segmentation maps
if target_sizes is not None:
if len(logits) != len(target_sizes):
raise ValueError(
"Make sure that you pass in as many target sizes as the batch dimension of the logits"
)
if is_torch_tensor(target_sizes):
target_sizes = target_sizes.numpy()
semantic_segmentation = []
for idx in range(len(logits)):
resized_logits = torch.nn.functional.interpolate(
logits[idx].unsqueeze(dim=0), size=target_sizes[idx], mode="bilinear", align_corners=False
)
semantic_map = resized_logits[0].argmax(dim=0)
semantic_segmentation.append(semantic_map)
else:
semantic_segmentation = logits.argmax(dim=1)
semantic_segmentation = [semantic_segmentation[i] for i in range(semantic_segmentation.shape[0])]
return semantic_segmentation