[CODE] MixUp 분할해서 구현하기

MixUp 구현. 

import torch
import numpy as np
import matplotlib.pyplot as plt
import torchvision
import torchvision.transforms as transforms

transform = transforms.Compose([
        transforms.Resize((224, 224)),
        transforms.ToTensor(),
        # transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
    ])

batch_size = 4

trainset = torchvision.datasets.STL10(root='./temp', split = "train",
                                         download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size,
                                          shuffle=True, num_workers=2)
tf = transforms.ToPILImage()

def mixup(images, alpha = 1.0):
    indices = torch.randperm(len(images))
    shuffled_images = images[indices]
    print(shuffled_images.size(), images.size())
    print("indexing : ", indices)
    
    lam = np.clip(np.random.beta(alpha, alpha), 0.45, 0.45)
    print(f"lambda : {lam}")

    mixedup_images = lam*images + (1 - lam)*shuffled_images

    return mixedup_images, images, shuffled_images
    
image, label = next(iter(trainloader))

aa = tf(mixed_images[0])
aa.show()

aa = tf(order_images[0])
aa.show()

aa = tf(shuffling[0])
aa.show()

결과물