Spaces:
Running on Zero
Running on Zero
File size: 4,709 Bytes
de9997b | 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 | # Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.
#
from logging import getLogger
from math import floor
from PIL import ImageFilter
import torch
import torchvision.transforms as transforms
import torchvision.transforms.functional as F
_GLOBAL_SEED = 0
logger = getLogger()
def make_transforms(
crop_size=224,
crop_scale=(0.3, 1.0),
color_jitter=1.0,
random_resize_crop=True,
horizontal_flip=False,
color_distortion=False,
gaussian_blur=False,
normalization=((0.485, 0.456, 0.406),
(0.229, 0.224, 0.225)),
preserve_aspect_ratio=True,
max_patches=14*14,
patch_size=16
):
logger.info('making data transforms')
def get_color_distortion(s=1.0):
# s is the strength of color distortion.
color_jitter = transforms.ColorJitter(0.8*s, 0.8*s, 0.8*s, 0.2*s)
rnd_color_jitter = transforms.RandomApply([color_jitter], p=0.8)
rnd_gray = transforms.RandomGrayscale(p=0.2)
color_distort = transforms.Compose([
rnd_color_jitter,
rnd_gray])
return color_distort
transform_list = []
if preserve_aspect_ratio:
if random_resize_crop:
transform_list.append(RandomResizedCropARP(scale=crop_scale))
transform_list.append(
ResizeToFixedPatches(max_patches=max_patches, patch_size=patch_size)
)
else:
if random_resize_crop:
transform_list.append(transforms.RandomResizedCrop(crop_size, scale=crop_scale))
else:
transform_list.append(transforms.Resize((crop_size, crop_size)))
if horizontal_flip:
transform_list.append(transforms.RandomHorizontalFlip())
if color_distortion:
transform_list.append(get_color_distortion(s=color_jitter))
if gaussian_blur:
transform_list.append(GaussianBlur(p=0.5))
mean, std = normalization
transform_list.extend([transforms.ToTensor(), transforms.Normalize(mean, std)])
transform = transforms.Compose(transform_list)
return transform
class GaussianBlur(object):
"""
Apply Gaussian blur with a given probability.
:param p: Probability of applying blur
:param radius_min: Minimum blur radius
:param radius_max: Maximum blur radius
"""
def __init__(self, p=0.5, radius_min=0.1, radius_max=2.):
self.prob = p
self.radius_min = radius_min
self.radius_max = radius_max
def __call__(self, img):
if torch.bernoulli(torch.tensor(self.prob)) == 0:
return img
radius = self.radius_min + torch.rand(1) * (self.radius_max - self.radius_min)
return img.filter(ImageFilter.GaussianBlur(radius=radius))
class ResizeToFixedPatches(object):
"""
Resize an image so that its patch grid fits within max_patches.
:param max_patches: Maximum number of patches (H*W)
:param patch_size: Patch size in pixels
"""
def __init__(self, max_patches=14*14, patch_size=16):
self.max_patches = int(max_patches)
self.patch_size = int(patch_size)
def __call__(self, img):
image_width, image_height = img.size
scale = (self.max_patches * (self.patch_size / image_height) * (self.patch_size / image_width)) ** 0.5
scale += 1e-6
num_feasible_rows = max(min(floor(scale * image_height / self.patch_size), self.max_patches), 1)
num_feasible_cols = max(min(floor(scale * image_width / self.patch_size), self.max_patches), 1)
resized_height = max(int(num_feasible_rows * self.patch_size), 1)
resized_width = max(int(num_feasible_cols * self.patch_size), 1)
return F.resize(
img,
(resized_height, resized_width),
interpolation=transforms.InterpolationMode.BICUBIC,
antialias=True
)
class RandomResizedCropARP(object):
"""
Random resized crop that preserves the original aspect ratio.
Uses torchvision's RandomResizedCrop sampling logic with ratio fixed
to the input image's aspect ratio.
:param scale: Range of area scaling for the crop
"""
def __init__(self, scale=(0.3, 1.0)):
self.scale = scale
def __call__(self, img):
image_width, image_height = img.size
if image_width <= 0 or image_height <= 0:
return img
ratio = image_width / image_height
i, j, h, w = transforms.RandomResizedCrop.get_params(
img,
scale=self.scale,
ratio=(ratio, ratio)
)
return F.crop(img, i, j, h, w)
|