You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
54 lines
1.5 KiB
54 lines
1.5 KiB
5 years ago
|
import torch
|
||
|
|
from torchvision import transforms
|
||
|
|
from PIL import Image
|
||
|
|
import math
|
||
|
|
import numpy as np
|
||
|
|
from data.random_erasing import RandomErasingNumpy
|
||
|
|
|
||
|
|
DEFAULT_CROP_PCT = 0.875
|
||
|
|
|
||
|
|
IMAGENET_DPN_MEAN = [124 / 255, 117 / 255, 104 / 255]
|
||
|
|
IMAGENET_DPN_STD = [1 / (.0167 * 255)] * 3
|
||
|
|
IMAGENET_INCEPTION_MEAN = [0.5, 0.5, 0.5]
|
||
|
|
IMAGENET_INCEPTION_STD = [0.5, 0.5, 0.5]
|
||
|
|
IMAGENET_DEFAULT_MEAN = [0.485, 0.456, 0.406]
|
||
|
|
IMAGENET_DEFAULT_STD = [0.229, 0.224, 0.225]
|
||
|
|
|
||
|
|
|
||
|
|
class AsNumpy:
|
||
|
|
|
||
|
|
def __call__(self, pil_img):
|
||
|
|
np_img = np.array(pil_img, dtype=np.uint8)
|
||
|
|
if np_img.ndim < 3:
|
||
|
|
np_img = np.expand_dims(np_img, axis=-1)
|
||
|
|
np_img = np.rollaxis(np_img, 2) # HWC to CHW
|
||
|
|
return np_img
|
||
|
|
|
||
|
|
|
||
|
|
def transforms_imagenet_train(
|
||
|
|
img_size=224,
|
||
|
|
scale=(0.1, 1.0),
|
||
|
|
color_jitter=(0.4, 0.4, 0.4),
|
||
|
|
random_erasing=0.4):
|
||
|
|
|
||
|
|
tfl = [
|
||
|
|
transforms.RandomResizedCrop(img_size, scale=scale),
|
||
|
|
transforms.RandomHorizontalFlip(),
|
||
|
|
transforms.ColorJitter(*color_jitter),
|
||
|
|
AsNumpy(),
|
||
|
|
]
|
||
|
|
#if random_erasing > 0.:
|
||
|
|
# tfl.append(RandomErasingNumpy(random_erasing, per_pixel=True))
|
||
|
|
return transforms.Compose(tfl)
|
||
|
|
|
||
|
|
|
||
|
|
def transforms_imagenet_eval(img_size=224, crop_pct=None):
|
||
|
|
crop_pct = crop_pct or DEFAULT_CROP_PCT
|
||
|
|
scale_size = int(math.floor(img_size / crop_pct))
|
||
|
|
|
||
|
|
return transforms.Compose([
|
||
|
|
transforms.Resize(scale_size, Image.BICUBIC),
|
||
|
|
transforms.CenterCrop(img_size),
|
||
|
|
AsNumpy(),
|
||
|
|
])
|