\n Generative Adversarial Networks (GANs)\ub294 \uc0dd\uc131 \ubaa8\ub378\uc758 \uc77c\uc885\uc73c\ub85c, \ub450 \uac1c\uc758 \uc2e0\uacbd\ub9dd(\uc0dd\uc131\uc790\uc640 \ud310\ubcc4\uc790) \uac04\uc758 \uacbd\uc7c1\uc744 \ud1b5\ud574 \uc2e4\uc81c\uc640 \uc720\uc0ac\ud55c \ub370\uc774\ud130\ub97c \uc0dd\uc131\ud558\ub294 \uae30\uc220\uc785\ub2c8\ub2e4.
\n \uc774\ubc88 \uae00\uc5d0\uc11c\ub294 GAN\uc744 \uc774\uc6a9\ud558\uc5ec \uc0ac\uacfc\uc640 \uc624\ub80c\uc9c0 \uc774\ubbf8\uc9c0\ub97c \uc0dd\uc131\ud558\ub294 \ubc29\ubc95\uc744 \ud0d0\uad6c\ud558\uaca0\uc2b5\ub2c8\ub2e4. \uc774 \uacfc\uc815\uc5d0\uc11c PyTorch \ud504\ub808\uc784\uc6cc\ud06c\ub97c \uc0ac\uc6a9\ud558\uc5ec GAN\uc744 \uad6c\ud604\ud560 \uac83\uc774\uba70,
\n \uc2e4\uc2b5\uc744 \uc704\ud55c \ud30c\uc774\uc36c \ucf54\ub4dc\ub3c4 \uc81c\uacf5\ud558\uaca0\uc2b5\ub2c8\ub2e4.\n <\/p>\n
\n GAN\uc740 Ian Goodfellow\uc5d0 \uc758\ud574 2014\ub144\uc5d0 \uc81c\uc548\ub41c \ubaa8\ub378\ub85c, \ub450 \uac1c\uc758 \uc778\uacf5\uc2e0\uacbd\ub9dd \uad6c\uc870\uac00 \uc11c\ub85c \uacbd\uc7c1\ud558\uc5ec \ud559\uc2b5\ud569\ub2c8\ub2e4.
\n \uc774\ub7ec\ud55c \uad6c\uc870\ub294 \ub2e4\uc74c\uacfc \uac19\uc740 \ub450 \ubd80\ubd84\uc73c\ub85c \ub098\ub20c \uc218 \uc788\uc2b5\ub2c8\ub2e4:\n <\/p>\n
\n GAN\uc758 \ud6c8\ub828 \uacfc\uc815\uc740 \ub2e4\uc74c\uacfc \uac19\uc2b5\ub2c8\ub2e4:\n <\/p>\n
\n GAN\uc744 \ud6c8\ub828\ud558\uae30 \uc704\ud574\uc11c \uc0ac\uacfc\uc640 \uc624\ub80c\uc9c0 \uc774\ubbf8\uc9c0\ub97c \ud3ec\ud568\ud558\ub294 \ub370\uc774\ud130\uc14b\uc744 \uc900\ube44\ud574\uc57c \ud569\ub2c8\ub2e4. \uc774\ubc88 \uc608\uc81c\uc5d0\uc11c\ub294 Kaggle\uc774\ub098 \ub2e4\ub978 \uc624\ud508 \ub370\uc774\ud130\uc14b\uc5d0\uc11c \uc0ac\uacfc\uc640 \uc624\ub80c\uc9c0 \ub370\uc774\ud130\ub97c \uc218\uc9d1\ud560 \uc608\uc815\uc785\ub2c8\ub2e4.
\n \uc774\ubbf8\uc9c0 \ub370\uc774\ud130\ub294 \uac19\uc740 \ud06c\uae30\ub85c \ubcc0\uacbd \ubc0f \uc815\uaddc\ud654\ub418\uba70, \uadf8 \ud6c4 \ud150\uc11c\ub85c \ubcc0\ud658\ub418\uc5b4\uc57c \ud569\ub2c8\ub2e4. \uc77c\ubc18\uc801\uc73c\ub85c \uc774\ubbf8\uc9c0\ub97c (64, 64) \ud06c\uae30\ub85c \uc870\uc815\ud558\uace0,
\n [-1, 1] \ubc94\uc704\ub85c \uc815\uaddc\ud654\ud558\ub294 \uac83\uc774 \uc77c\ubc18\uc801\uc785\ub2c8\ub2e4.\n <\/p>\n
\ub2e4\uc74c\uc740 \uc774\ubbf8\uc9c0 \uc804\ucc98\ub9ac \uacfc\uc815\uc744 \uad6c\ud604\ud558\ub294 \ud30c\uc774\uc36c \ucf54\ub4dc\uc785\ub2c8\ub2e4:<\/p>\n
\nimport os\nimport numpy as np\nimport cv2\nfrom torchvision import transforms\nfrom PIL import Image\nimport torch\n\ndef load_images_from_folder(folder):\n images = []\n for filename in os.listdir(folder):\n img = cv2.imread(os.path.join(folder, filename))\n if img is not None:\n img = cv2.resize(img, (64, 64))\n img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n images.append(img)\n return np.array(images)\n\nfolder = 'path_to_your_dataset'\ndataset = load_images_from_folder(folder)\n\ntransform = transforms.Compose([\n transforms.ToPILImage(),\n transforms.ToTensor(),\n transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),\n])\ntensor_images = [transform(Image.fromarray(img)).unsqueeze(0) for img in dataset]\n\nimages_tensor = torch.cat(tensor_images)\n <\/code><\/pre>\n3. GAN \uad6c\uc870 \uad6c\ud604<\/h2>\n
\n GAN\uc744 \uad6c\ud604\ud558\uae30 \uc704\ud574 \uba3c\uc800 \uc0dd\uc131\uc790\uc640 \ud310\ubcc4\uc790\ub97c \uc815\uc758\ud574\uc57c \ud569\ub2c8\ub2e4.
\n \uc0dd\uc131\uc790\ub294 \uc77c\ubc18\uc801\uc73c\ub85c Fully Connected Layer\uc640 Convolutional Layer\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc774\ubbf8\uc9c0\ub97c \uc0dd\uc131\ud569\ub2c8\ub2e4.
\n \ud310\ubcc4\uc790\ub294 Convolutional Layer\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc774\ubbf8\uc9c0\uc758 \uc9c4\uc9dc \uc5ec\ubd80\ub97c \ud310\ub2e8\ud569\ub2c8\ub2e4.
\n \uc544\ub798\ub294 \ud30c\uc774\ud1a0\uce58\ub85c \uc791\uc131\ub41c \uac04\ub2e8\ud55c GAN \ubaa8\ub378\uc785\ub2c8\ub2e4.\n <\/p>\n
3.1. \uc0dd\uc131\uc790 \ubaa8\ub378<\/h3>\n\nimport torch.nn as nn\n\nclass Generator(nn.Module):\n def __init__(self, input_dim, output_dim):\n super(Generator, self).__init__()\n self.model = nn.Sequential(\n nn.Linear(input_dim, 128),\n nn.ReLU(),\n nn.Linear(128, 256),\n nn.ReLU(),\n nn.Linear(256, output_dim),\n nn.Tanh() # Output range to [-1, 1]\n )\n \n def forward(self, z):\n img = self.model(z)\n return img.view(img.size(0), 3, 64, 64) # Reshape for image output\n <\/code><\/pre>\n3.2. \ud310\ubcc4\uc790 \ubaa8\ub378<\/h3>\n\nclass Discriminator(nn.Module):\n def __init__(self, input_dim):\n super(Discriminator, self).__init__()\n self.model = nn.Sequential(\n nn.Conv2d(3, 32, kernel_size=4, stride=2, padding=1),\n nn.LeakyReLU(0.2),\n nn.Conv2d(32, 64, kernel_size=4, stride=2, padding=1),\n nn.LeakyReLU(0.2),\n nn.Flatten(),\n nn.Linear(64 * 16 * 16, 1),\n nn.Sigmoid() # Output range to [0, 1]\n )\n\n def forward(self, img):\n return self.model(img)\n <\/code><\/pre>\n4. GAN \ud6c8\ub828<\/h2>\n
\n GAN\uc744 \ud6c8\ub828\uc2dc\ud0a4\uae30 \uc704\ud574 \ub2e4\uc74c\uc758 \uacfc\uc815\uc744 \ubc18\ubcf5\ud569\ub2c8\ub2e4.
\n \uc0dd\uc131\uc790\ub294 \ubb34\uc791\uc704 \ub178\uc774\uc988\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc774\ubbf8\uc9c0\ub97c \uc0dd\uc131\ud558\uace0,
\n \ud310\ubcc4\uc790\ub294 \uc0dd\uc131\ub41c \uc774\ubbf8\uc9c0\uc640 \uc2e4\uc81c \uc774\ubbf8\uc9c0\ub97c \uad6c\ubd84\ud569\ub2c8\ub2e4.
\n \uc774\ud6c4 \uc190\uc2e4 \ud568\uc218\ub97c \ubc14\ud0d5\uc73c\ub85c \ub450 \ubaa8\ub378\uc744 \uac01\uac01 \uc5c5\ub370\uc774\ud2b8\ud569\ub2c8\ub2e4.\n <\/p>\n
\nimport torch.optim as optim\n\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n\n# Hyperparameters\ninput_dim = 100\noutput_dim = 3 * 64 * 64\nlr = 0.0002\nnum_epochs = 200\n\n# Models and optimizers\ngenerator = Generator(input_dim, output_dim).to(device)\ndiscriminator = Discriminator(output_dim).to(device)\ncriterion = nn.BCELoss()\noptimizer_G = optim.Adam(generator.parameters(), lr=lr)\noptimizer_D = optim.Adam(discriminator.parameters(), lr=lr)\n\n# Label for real and fake images\nreal_labels = torch.ones(batch_size, 1).to(device)\nfake_labels = torch.zeros(batch_size, 1).to(device)\n\nfor epoch in range(num_epochs):\n for i, imgs in enumerate(dataloader):\n # Train Discriminator\n optimizer_D.zero_grad()\n real_imgs = imgs.to(device)\n real_loss = criterion(discriminator(real_imgs), real_labels)\n \n z = torch.randn(batch_size, input_dim).to(device)\n fake_imgs = generator(z)\n fake_loss = criterion(discriminator(fake_imgs.detach()), fake_labels)\n \n d_loss = real_loss + fake_loss\n d_loss.backward()\n optimizer_D.step()\n \n # Train Generator\n optimizer_G.zero_grad()\n g_loss = criterion(discriminator(fake_imgs), real_labels)\n g_loss.backward()\n optimizer_G.step()\n\n if (i + 1) % 100 == 0:\n print(f'Epoch [{epoch + 1}\/{num_epochs}], Step [{i + 1}\/{len(dataloader)}], '\n f'D Loss: {d_loss.item():.4f}, G Loss: {g_loss.item():.4f}')\n <\/code><\/pre>\n5. \uacb0\uacfc \ubc0f \uc2dc\uac01\ud654<\/h2>\n
\n \ud6c8\ub828\uc774 \uc644\ub8cc\ub418\uba74 \uc0dd\uc131\ub41c \uc774\ubbf8\uc9c0\ub97c \uc2dc\uac01\ud654\ud558\uc5ec \uc131\ub2a5\uc744 \ud3c9\uac00\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.
\n \uc544\ub798\ub294 \uc0dd\uc131\ub41c \uc774\ubbf8\uc9c0\ub97c \uadf8\ub9ac\ub4dc \ud615\ud0dc\ub85c \ud45c\uc2dc\ud558\ub294 \ud30c\uc774\uc36c \ucf54\ub4dc\uc785\ub2c8\ub2e4.\n <\/p>\n
\nimport matplotlib.pyplot as plt\n\ndef show_generated_images(generator, num_images):\n z = torch.randn(num_images, input_dim).to(device)\n generated_images = generator(z)\n\n grid = torchvision.utils.make_grid(generated_images.cpu().detach(), nrow=5, normalize=True)\n \n plt.imshow(grid.permute(1, 2, 0))\n plt.axis('off')\n plt.show()\n\nshow_generated_images(generator, 25)\n <\/code><\/pre>\n6. \uacb0\ub860<\/h2>\n
\n GAN\uc744 \ud65c\uc6a9\ud55c \uc0ac\uacfc\uc640 \uc624\ub80c\uc9c0 \uc0dd\uc131 \ubaa8\ub378\uc744 \uad6c\ucd95\ud558\uace0 \ud6c8\ub828\uc2dc\ud0a4\ub294 \uacfc\uc815\uc5d0 \ub300\ud574 \uc0b4\ud3b4\ubcf4\uc558\uc2b5\ub2c8\ub2e4. \uc2e4\uc9c8\uc801\uc778 \ub370\uc774\ud130\uc14b\uc744 \ud65c\uc6a9\ud558\uc5ec
\n PyTorch \ud504\ub808\uc784\uc6cc\ud06c\uc758 \uac15\ub825\ud55c \uae30\ub2a5\uc744 \uc774\uc6a9\ud574 \ubaa8\ub378\uc744 \uad6c\ud604\ud558\ub294 \ubc29\ubc95\uc744 \ubc30\uc6e0\uc2b5\ub2c8\ub2e4. \ub2e4\uc591\ud55c \ubd84\uc57c\uc5d0 \uc751\uc6a9 \uac00\ub2a5\uc131\uc774 \uc788\ub294 GAN\uc758 \ud798\uc744
\n \uacbd\ud5d8\ud55c \ub9cc\ud07c, \uc774\ub97c \ud1b5\ud574 \uc55e\uc73c\ub85c \ub354 \ubc1c\uc804\ub41c \ubaa8\ub378\uc744 \ub9cc\ub4e4\uc5b4\ubcf4\uc2dc\uae38 \ubc14\ub78d\ub2c8\ub2e4.\n <\/p>\n
\n \ub354 \ub9ce\uc740 \uac83\ub4e4\uc744 \ubc30\uc6b0\uace0 \uc2f6\ub2e4\uba74, GAN\uc758 \ub2e4\uc591\ud55c \ubcc0\uc885\uc778 CycleGAN\uc774\ub098 StyleGAN\uc744 \ud559\uc2b5\ud574\ubcf4\ub294 \uac83\ub3c4 \uc88b\uc740 \ubc29\ubc95\uc785\ub2c8\ub2e4.
\n \uc774\ub7ec\ud55c \uace0\uae09 \ub0b4\uc6a9\ub4e4\uc744 \ud1b5\ud574 \ub525\ub7ec\ub2dd \uae30\uc220\uc744 \ud55c\uce35 \ub354 \ub113\ud600\uac00\uc2dc\uae38 \ubc14\ub78d\ub2c8\ub2e4.\n <\/p>\n
<\/body><\/p>\n","protected":false},"excerpt":{"rendered":"