Super Kai (Kazuya Ito)

Posted on Feb 22

RandomAffine in PyTorch (6)

#python #pytorch #randomaffine #v2

*Memos:

My post explains RandomAffine() about degrees, translate, fill and center argument.
My post explains RandomAffine() about scale argument.
My post explains RandomAffine() about shear argument (1).
My post explains RandomAffine() about shear argument (2).
My post explains RandomAffine() about shear argument (3).
My post explains OxfordIIITPet().

RandomAffine() can do random rotation or random affine transformation for an image as shown below. *It's about shear argument (4):

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

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

shear0_0_0_0origin_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[0, 0, 0, 0])
)

shearn10n10_10_10_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[-10, -10, 10, 10])
    # transform=RandomAffine(degrees=[0, 0], shear=[170, 170, -170, -170])
)

shearn20n20_20_20_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[-20, -20, 20, 20])
    # transform=RandomAffine(degrees=[0, 0], shear=[160, 160, -160, -160])
)

shearn30n30_30_30_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[-30, -30, 30, 30])
    # transform=RandomAffine(degrees=[0, 0], shear=[150, 150, -150, -150])
)

shearn40n40_40_40_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[-40, -40, 40, 40])
    # transform=RandomAffine(degrees=[0, 0], shear=[140, 140, -140, -140])
)

shearn50n50_50_50_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[-50, -50, 50, 50])
    # transform=RandomAffine(degrees=[0, 0], shear=[130, 130, -130, -130])
)

shearn60n60_60_60_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[-60, -60, 60, 60])
    # transform=RandomAffine(degrees=[0, 0], shear=[120, 120, -120, -120])
)

shearn70n70_70_70_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[-70, -70, 70, 70])
    # transform=RandomAffine(degrees=[0, 0], shear=[110, 110, -110, -110])
)

shearn80n80_80_80_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[-80, -80, 80, 80])
    # transform=RandomAffine(degrees=[0, 0], shear=[100, 100, -100, -100])
)

shearn90n90_90_90_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[-90, -90, 90, 90])
    # transform=RandomAffine(degrees=[0, 0], shear=[90, 90, -90, -90])
)

shear10_10n10n10_data = OxfordIIITPet( # `n` is negative.
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[10, 10, -10, -10])
    # transform=RandomAffine(degrees=[0, 0], shear=[-170, -170, 170, 170])
)

shear20_20n20n20_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[20, 20, -20, -20])
    # transform=RandomAffine(degrees=[0, 0], shear=[-160, -160, 160, 160])
)

shear30_30n30n30_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[30, 30, -30, -30])
    # transform=RandomAffine(degrees=[0, 0], shear=[-150, -150, 150, 150])
)

shear40_40n40n40_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[40, 40, -40, -40])
    # transform=RandomAffine(degrees=[0, 0], shear=[-140, -140, 140, 140])
)

shear50_50n50n50_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[50, 50, -50, -50])
    # transform=RandomAffine(degrees=[0, 0], shear=[-130, -130, 130, 130])
)

shear60_60n60n60_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[60, 60, -60, -60])
    # transform=RandomAffine(degrees=[0, 0], shear=[-120, -120, 120, 120])
)

shear70_70n70n70_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[70, 70, -70, -70])
    # transform=RandomAffine(degrees=[0, 0], shear=[-110, -110, 110, 110])
)

shear80_80n80n80_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[80, 80, -80, -80])
    # transform=RandomAffine(degrees=[0, 0], shear=[-100, -100, 100, 100])
)

shear90_90n90n90_data = OxfordIIITPet(
    root="data",
    transform=RandomAffine(degrees=[0, 0], shear=[90, 90, -90, -90])
    # transform=RandomAffine(degrees=[0, 0], shear=[-90, -90, 90, 90])
)

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.xticks(ticks=[])
        plt.yticks(ticks=[])
    plt.tight_layout()
    plt.show()

show_images1(data=origin_data, main_title="origin_data")
print()
show_images1(data=shear0_0_0_0origin_data, 
             main_title="shear0_0_0_0origin_data")
show_images1(data=shearn10n10_10_10_data, main_title="shearn10n10_10_10_data")
show_images1(data=shearn20n20_20_20_data, main_title="shearn20n20_20_20_data")
show_images1(data=shearn30n30_30_30_data, main_title="shearn30n30_30_30_data")
show_images1(data=shearn40n40_40_40_data, main_title="shearn40n40_40_40_data")
show_images1(data=shearn50n50_50_50_data, main_title="shearn50n50_50_50_data")
show_images1(data=shearn60n60_60_60_data, main_title="shearn60n60_60_60_data")
show_images1(data=shearn70n70_70_70_data, main_title="shearn70n70_70_70_data")
show_images1(data=shearn80n80_80_80_data, main_title="shearn80n80_80_80_data")
show_images1(data=shearn90n90_90_90_data, main_title="shearn90n90_90_90_data")
print()
show_images1(data=shear0_0_0_0origin_data, 
             main_title="shear0_0_0_0origin_data")
show_images1(data=shear10_10n10n10_data,
             main_title="shear10_10n10n10_data")
show_images1(data=shear20_20n20n20_data,
             main_title="shear20_20n20n20_data")
show_images1(data=shear30_30n30n30_data,
             main_title="shear30_30n30n30_data")
show_images1(data=shear40_40n40n40_data,
             main_title="shear40_40n40n40_data")
show_images1(data=shear50_50n50n50_data, 
             main_title="shear50_50n50n50_data")
show_images1(data=shear60_60n60n60_data, 
             main_title="shear60_60n60n60_data")
show_images1(data=shear70_70n70n70_data, 
             main_title="shear70_70n70n70_data")
show_images1(data=shear80_80n80n80_data, 
             main_title="shear80_80n80n80_data")
show_images1(data=shear90_90n90n90_data, 
             main_title="shear90_90n90n90_data")

# ↓ ↓ ↓ ↓ ↓ ↓ The code below is identical to the code above. ↓ ↓ ↓ ↓ ↓ ↓ 
def show_images2(data, main_title=None, d=None, t=None, sc=None, sh=None,
                 ip=InterpolationMode.NEAREST, f=0, c=None):
    plt.figure(figsize=[10, 5])
    plt.suptitle(t=main_title, y=0.8, fontsize=14)
    if d != None:
        for i, (im, _) in zip(range(1, 6), data):
            plt.subplot(1, 5, i)
            ra = RandomAffine(degrees=d, translate=t, scale=sc, shear=sh,
                              interpolation=ip, center=c, fill=f)
            plt.imshow(X=ra(im))
            plt.xticks(ticks=[])
            plt.yticks(ticks=[])
    else:
        for i, (im, _) in zip(range(1, 6), data):
            plt.subplot(1, 5, i)
            plt.imshow(X=im)
            plt.xticks(ticks=[])
            plt.yticks(ticks=[])  
    plt.tight_layout()
    plt.show()

show_images2(data=origin_data, main_title="origin_data")
print()
show_images2(data=origin_data, main_title="shear0_0_0_0origin_data",
             d=[0, 0], sh=[0, 0, 0, 0])
show_images2(data=origin_data, main_title="shearn10n10_10_10_data",
             d=[0, 0], sh=[-10, -10, 10, 10])
show_images2(data=origin_data, main_title="shearn20n20_20_20_data",
             d=[0, 0], sh=[-20, -20, 20, 20])
show_images2(data=origin_data, main_title="shearn30n30_30_30_data",
             d=[0, 0], sh=[-30, -30, 30, 30])
show_images2(data=origin_data, main_title="shearn40n40_40_40_data",
             d=[0, 0], sh=[-40, -40, 40, 40])
show_images2(data=origin_data, main_title="shearn50n50_50_50_data",
             d=[0, 0], sh=[-50, -50, 50, 50])
show_images2(data=origin_data, main_title="shearn60n60_60_60_data",
             d=[0, 0], sh=[-60, -60, 60, 60])
show_images2(data=origin_data, main_title="shearn70n70_70_70_data",
             d=[0, 0], sh=[-70, -70, 70, 70])
show_images2(data=origin_data, main_title="shearn80n80_80_80_data",
             d=[0, 0], sh=[-80, -80, 80, 80])
show_images2(data=origin_data, main_title="shearn90n90_90_90_data",
             d=[0, 0], sh=[-90, -90, 90, 90])
print()
show_images2(data=origin_data, main_title="shear0_0_0_0origin_data",
             d=[0, 0], sh=[0, 0, 0, 0])
show_images2(data=origin_data, main_title="shear10_10n10n10_data",
             d=[0, 0], sh=[10, 10, -10, -10])
show_images2(data=origin_data, main_title="shear20_20n20n20_data",
             d=[0, 0], sh=[20, 20, -20, -20])
show_images2(data=origin_data, main_title="shear30_30n30n30_data",
             d=[0, 0], sh=[30, 30, -30, -30])
show_images2(data=origin_data, main_title="shear40_40n40n40_data",
             d=[0, 0], sh=[40, 40, -40, -40])
show_images2(data=origin_data, main_title="shear50_50n50n50_data",
             d=[0, 0], sh=[50, 50, -50, -50])
show_images2(data=origin_data, main_title="shear60_60n60n60_data",
             d=[0, 0], sh=[60, 60, -60, -60])
show_images2(data=origin_data, main_title="shear70_70n70n70_data",
             d=[0, 0], sh=[70, 70, -70, -70])
show_images2(data=origin_data, main_title="shear80_80n80n80_data",
             d=[0, 0], sh=[80, 80, -80, -80])
show_images2(data=origin_data, main_title="shear90_90n90n90_data",
             d=[0, 0], sh=[90, 90, -90, -90])

Build apps, not infrastructure.

Dealing with servers, hardware, and infrastructure can take up your valuable time. Discover the benefits of Heroku, the PaaS of choice for developers since 2007.

Visit Site

DEV Community

RandomAffine in PyTorch (6)

Build apps, not infrastructure.

Top comments (0)

Create up to 10 Postgres Databases on Neon's free plan.

Read next

How to Install Ollama with DeepSeek-r1 and Integrate it with Python on Windows

🚀 My Journey into Web & Software Development

Deep Dive into Server-Sent Events (SSE)

Top 7 Python Libraries Every Data Analyst Should Know in 2025

Okay