*Memos:
- My post explains RandomResizedCrop() about
sizeargument (2). - My post explains RandomResizedCrop() about
scaleargument (1). - My post explains RandomResizedCrop() about
scaleargument (2). - My post explains RandomResizedCrop() about
ratioargument (1). - My post explains RandomResizedCrop() about
ratioargument (2). - My post explains RandomCrop() about
sizeargument. - My post explains CenterCrop().
- My post explains FiveCrop().
- My post explains Resize().
- My post explains OxfordIIITPet().
RandomResizedCrop() can crop a random part of an image, then resize it to a given size as shown below. *It's about size argument (1):
*Memos:
- The 1st argument for initialization is
size(Required-Type:intortuple/list(int) or size()): *Memos:- It's
[height, width]. - It must be
1 <= x. - A tuple/list must be the 1D with 1 or 2 elements.
- A single value(
intortuple/list(int)) means[size, size].
- It's
- The 2nd argument for initialization is
scale(Optional-Default:(0.08, 1.0)-Type:tuple/list(intorfloat)): *Memos:- It's
[min, max]so it must bemin <= max. - It must be
0 <= x. - A tuple/list must be the 1D with 2 elements.
- A double of
0or1 <= xgets the same result.
- It's
- The 3rd argument for initialization is
ratio(Optional-Default:(0.75, 1.3333333333333333)-Type:tuple/list(intorfloat)): *Memos:- It's
[min, max]so it must bemin <= max. - It must be
0 < x. - A tuple/list must be the 1D with 2 elements.
- It's
- The 4th argument for initialization is
interpolation(Optional-Default:InterpolationMode.BILINEAR-Type:InterpolationMode): *Memos:-
NEAREST,NEAREST_EXACT,BILINEAR,BICUBIC,BOX,HAMMINGandLANCZOSmodes can be used. - My post explains InterpolationMode with and without anti-aliasing.
-
- The 5th argument for initialization is
antialias(Optional-Default:True-Type:bool): *Memos:- If it's
True(Default) andinterpolationisBILINEARorBICUBIC, anti-aliasing is applied for both a PIL image and tensor. - If it's
FalseorNoneandinterpolationisBILINEARorBICUBIC, anti-aliasing is still applied for a PIL image while anti-aliasing isn't applied for a tensor.
- If it's
- The 1st argument is
img(Required-Type:PIL Imageortensor(int/float/complex/bool)): *Memos:- A tensor must be 2D or more D.
- Don't use
img=.
-
v2is recommended to use according to V1 or V2? Which one should I use?.
from torchvision.datasets import OxfordIIITPet from torchvision.transforms.v2 import RandomResizedCrop from torchvision.transforms.functional import InterpolationMode rrc = RandomResizedCrop(size=100) rrc = RandomResizedCrop(size=100, scale=(0.08, 1), ratio=(0.75, 1.3333333333333333), interpolation=InterpolationMode.BILINEAR, antialias=True) rrc # RandomResizedCrop(size=(100, 100), scale=(0.08, 1.0), # ratio=(0.75, 1.3333333333333333), # interpolation=InterpolationMode.BILINEAR, # antialias=True) rrc.size # (100, 100) rrc.scale # (0.08, 1.0) rrc.ratio # (0.75, 1.3333333333333333) rrc.interpolationa # <InterpolationMode.BILINEAR: 'bilinear'> rrc.antialias # True origin_data = OxfordIIITPet( root="data", transform=None ) s1000_data = OxfordIIITPet( # `s` is size. root="data", transform=RandomResizedCrop(size=1000) # transform=RandomResizedCrop(size=[1000]) # transform=RandomResizedCrop(size=[1000, 1000]) ) s500_data = OxfordIIITPet( root="data", transform=RandomResizedCrop(size=500) ) s100_data = OxfordIIITPet( root="data", transform=RandomResizedCrop(size=100) ) s50_data = OxfordIIITPet( root="data", transform=RandomResizedCrop(size=50) ) s10_data = OxfordIIITPet( root="data", transform=RandomResizedCrop(size=10) ) s1_data = OxfordIIITPet( root="data", transform=RandomResizedCrop(size=1) ) s600_900_data = OxfordIIITPet( root="data", transform=RandomResizedCrop(size=[600, 900]) ) s900_600_data = OxfordIIITPet( root="data", transform=RandomResizedCrop(size=[900, 600]) ) s200_300_data = OxfordIIITPet( root="data", transform=RandomResizedCrop(size=[200, 300]) ) s300_200_data = OxfordIIITPet( root="data", transform=RandomResizedCrop(size=[300, 200]) ) import matplotlib.pyplot as plt def show_images1(data, main_title=None): plt.figure(figsize=[10, 5]) plt.suptitle(t=main_title, y=0.8, fontsize=14) for i, (im, _) in zip(range(1, 6), data): plt.subplot(1, 5, i) plt.imshow(X=im) plt.tight_layout() plt.show() show_images1(data=origin_data, main_title="origin_data") show_images1(data=s1000_data, main_title="s1000_data") show_images1(data=s500_data, main_title="s500_data") show_images1(data=s100_data, main_title="s100_data") show_images1(data=s50_data, main_title="s50_data") show_images1(data=s10_data, main_title="s10_data") show_images1(data=s1_data, main_title="s1_data") print() show_images1(data=origin_data, main_title="origin_data") show_images1(data=s600_900_data, main_title="s600_900_data") show_images1(data=s900_600_data, main_title="s900_600_data") show_images1(data=s200_300_data, main_title="s200_300_data") show_images1(data=s300_200_data, main_title="s300_200_data") # ↓ ↓ ↓ ↓ ↓ ↓ The code below is identical to the code above. ↓ ↓ ↓ ↓ ↓ ↓ def show_images2(data, main_title=None, s=None, sc=(0.08, 1), r=(0.75, 1.3333333333333333), ip=InterpolationMode.BILINEAR, a=True): plt.figure(figsize=[10, 5]) plt.suptitle(t=main_title, y=0.8, fontsize=14) if main_title != "origin_data": for i, (im, _) in zip(range(1, 6), data): plt.subplot(1, 5, i) rrc = RandomResizedCrop(size=s, scale=sc, ratio=r, interpolation=ip, antialias=a) plt.imshow(X=rrc(im)) else: for i, (im, _) in zip(range(1, 6), data): plt.subplot(1, 5, i) plt.imshow(X=im) plt.tight_layout() plt.show() show_images2(data=origin_data, main_title="origin_data") show_images2(data=origin_data, main_title="s1000_data", s=1000) show_images2(data=origin_data, main_title="s500_data", s=500) show_images2(data=origin_data, main_title="s100_data", s=100) show_images2(data=origin_data, main_title="s50_data", s=50) show_images2(data=origin_data, main_title="s10_data", s=10) show_images2(data=origin_data, main_title="s1_data", s=1) print() show_images2(data=origin_data, main_title="origin_data") show_images2(data=origin_data, main_title="s600_900_data", s=[600, 900]) show_images2(data=origin_data, main_title="s900_600_data", s=[900, 600]) show_images2(data=origin_data, main_title="s200_300_data", s=[200, 300]) show_images2(data=origin_data, main_title="s300_200_data", s=[300, 200]) 
Top comments (0)