データ不足がモデルトレーニングに及ぼす影響

from PIL import Image

def rotate_image(image, angle):
    rotated_image = image.rotate(angle)
    return rotated_image

image = Image.open('image.jpg')
rotated_image = rotate_image(image, 90)
rotated_image.save('rotated_image.jpg')

from keras.applications import VGG16
from keras.models import Model

base_model = VGG16(weights='imagenet', include_top=False, input_shape=(224, 224, 3))
x = base_model.output
x = GlobalAveragePooling2D()(x)
x = Dense(1024, activation='relu')(x)
predictions = Dense(num_classes, activation='softmax')(x)
model = Model(inputs=base_model.input, outputs=predictions)
model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])

import torch
import torchvision
import torch.nn as nn

source_model = torchvision.models.resnet50(pretrained=True)
target_model = torchvision.models.resnet50(pretrained=False)

for param in source_model.parameters():
    param.requires_grad = False

source_features = source_model.features(x)
target_features = target_model.features(x)

class DANNClassifier(nn.Module):
    def __init__(self, num_classes):
        super(DANNClassifier, self).__init__()
        self.fc = nn.Linear(2048, num_classes)
    def forward(self, x):
        x = self.fc(x)
        return x

source_classifier = DANNClassifier(num_classes)
target_classifier = DANNClassifier(num_classes)

source_outputs = source_classifier(source_features)
target_outputs = target_classifier(target_features)