FiveCrop in PyTorch

Buy Me a Coffee☕

*Memos:

• My post explains OxfordIIITPet().

FiveCrop() can crop an image into 5 parts(Top-left, Top-right, Bottom-left, Bottom-right and Center) as shown below:

*Memos:

• The 1st argument for initialization is size(Required-Type:int or tuple/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(int or tuple/list(int)) means [size, size].
• The 1st argument is img(Required-Type:PIL Image or tensor(int)): *Memos:
  • A tensor must be 2D or 3D.
  • Don't use img=.
• v2 is recommended to use according to V1 or V2? Which one should I use?.

from torchvision.datasets import OxfordIIITPet
from torchvision.transforms.v2 import FiveCrop

fc = FiveCrop(size=100)

fc
# FiveCrop(size=(100, 100))

fc.size
# (100, 100)

origin_data = OxfordIIITPet(
    root="data",
    transform=None
)

s500_394origin_data = OxfordIIITPet( # `s` is size.
    root="data",
    transform=FiveCrop(size=[500, 394])
)

s300_data = OxfordIIITPet(
    root="data",
    transform=FiveCrop(size=300)
)

s200_data = OxfordIIITPet(
    root="data",
    transform=FiveCrop(size=200)
)

s100_data = OxfordIIITPet(
    root="data",
    transform=FiveCrop(size=100)
)

s50_data = OxfordIIITPet(
    root="data",
    transform=FiveCrop(size=50)
)

s10_data = OxfordIIITPet(
    root="data",
    transform=FiveCrop(size=10)
)

s1_data = OxfordIIITPet(
    root="data",
    transform=FiveCrop(size=1)
)

s200_300_data = OxfordIIITPet(
    root="data",
    transform=FiveCrop(size=[200, 300])
)

s300_200_data = OxfordIIITPet(
    root="data",
    transform=FiveCrop(size=[300, 200])
)

import matplotlib.pyplot as plt

def show_images1(fcims, main_title=None):
    plt.figure(figsize=[10, 5])
    plt.suptitle(t=main_title, y=0.8, fontsize=14)
    titles = ['Top-left', 'Top-right', 'Bottom-left',
              'Bottom-right', 'Center']
    for i, fcim in zip(range(1, 6), fcims):
        plt.subplot(1, 5, i)
        plt.title(label=titles[i-1], fontsize=14)
        plt.imshow(X=fcim)
    plt.tight_layout()
    plt.show()

plt.figure(figsize=[7, 9])
plt.title(label="s500_394origin_data", fontsize=14)
plt.imshow(X=origin_data[0][0])
show_images1(fcims=s500_394origin_data[0][0], main_title="s500_394origin_data")
show_images1(fcims=s300_data[0][0], main_title="s300_data")
show_images1(fcims=s200_data[0][0], main_title="s200_data")
show_images1(fcims=s100_data[0][0], main_title="s100_data")
show_images1(fcims=s50_data[0][0], main_title="s50_data")
show_images1(fcims=s10_data[0][0], main_title="s10_data")
show_images1(fcims=s1_data[0][0], main_title="s1_data")
print()
show_images1(fcims=s200_300_data[0][0], main_title="s200_300_data")
show_images1(fcims=s300_200_data[0][0], main_title="s300_200_data")

# ↓ ↓ ↓ ↓ ↓ ↓ The code below is identical to the code above. ↓ ↓ ↓ ↓ ↓ ↓
def show_images2(im, main_title=None, s=None):
    plt.figure(figsize=[10, 5])
    plt.suptitle(t=main_title, y=0.8, fontsize=14)
    titles = ['Top-left', 'Top-right', 'Bottom-left',
              'Bottom-right', 'Center']
    if not s:
        s = [im.size[1], im.size[0]] 
    fc = FiveCrop(size=s)
    for i, fcim in zip(range(1, 6), fc(im)):
        plt.subplot(1, 5, i)
        plt.title(label=titles[i-1], fontsize=14)
        plt.imshow(X=fcim)
    plt.tight_layout()
    plt.show()

plt.figure(figsize=[7, 9])
plt.title(label="s500_394origin_data", fontsize=14)
plt.imshow(X=origin_data[0][0])
show_images2(im=origin_data[0][0], main_title="s500_394origin_data")
# show_images2(im=origin_data[0][0], main_title="s500_394origin_data",
#              s=[500, 394])
show_images2(im=origin_data[0][0], main_title="s300_data", s=300)
show_images2(im=origin_data[0][0], main_title="s200_data", s=200)
show_images2(im=origin_data[0][0], main_title="s100_data", s=100)
show_images2(im=origin_data[0][0], main_title="s50_data", s=50)
show_images2(im=origin_data[0][0], main_title="s10_data", s=10)
show_images2(im=origin_data[0][0], main_title="s1_data", s=1)
print()
show_images2(im=origin_data[0][0], main_title="s200_300_data", s=[200, 300])
show_images2(im=origin_data[0][0], main_title="s300_200_data", s=[300, 200])

---