CelebDF_XceptionNet_Code.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "PtHpzm2dvuYq",
    "outputId": "c6463539-1924-4930-b60b-48f188803b46"
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torchvision\n",
    "import torchvision.transforms as transforms\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "from torchvision import datasets\n",
    "from torchvision.io import read_image\n",
    "import glob\n",
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "from PIL import Image\n",
    "import timm\n",
    "torch.manual_seed(42)\n",
    "device = 'cuda'\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "oMsY8gQydb8U"
   },
   "source": [
    "## Dataset extracted from Celeb-DF Videos. DataLoader Creation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "luUItpGederJ"
   },
   "outputs": [],
   "source": [
    "class CustomDataset(Dataset):\n",
    "\tdef __init__(self, transform=None):\n",
    "\t\tself.imgs_path = 'C:/792/Celeb-DF-v2_images/'\n",
    "\t\tself.transform = transform\n",
    "\t\tfile_list = glob.glob(self.imgs_path + \"*\")\n",
    "\t\tself.data = []\n",
    "\t\tfor class_path in file_list:\n",
    "\t\t\tclass_name = class_path.split(\"\\\\\")[-1]\n",
    "\t\t\tfor img_path in glob.glob(class_path + \"/*.jpg\"):\n",
    "\t\t\t\tself.data.append([img_path, class_name])\n",
    "\t\tself.class_map = {\"Celeb-real_images\": 0, \"Celeb-synthesis_images\": 1}\n",
    "\t\n",
    "\tdef __len__(self):\n",
    "\t\treturn len(self.data)\n",
    "\n",
    "\tdef __getitem__(self, idx):\n",
    "\t\timg_path, class_name = self.data[idx]\n",
    "\t\timg = read_image(img_path)\n",
    "\t\tclass_id = self.class_map[class_name]\n",
    "\t\tclass_id = torch.tensor([class_id])\n",
    "\n",
    "\t\tif self.transform:\n",
    "\t\t\t\timg_tensor = self.transform(img)\n",
    "\t\treturn img_tensor, class_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "307i1exx5vWk"
   },
   "outputs": [],
   "source": [
    "train_transforms = transforms.Compose([\n",
    "    transforms.ToPILImage(),\n",
    "    transforms.Resize([128, 128]),\n",
    "    transforms.ToTensor()\n",
    "    ])\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train Test Split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "i_hhUKPTdetg"
   },
   "outputs": [],
   "source": [
    "dataset = CustomDataset(transform=train_transforms)\n",
    "train_size = int(0.8 * len(dataset))\n",
    "val_size = len(dataset) - train_size\n",
    "train_dataset, val_dataset = torch.utils.data.random_split(dataset, [train_size, val_size])\t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "93_xKYhpl2PN",
    "outputId": "702e4895-02c8-4d71-e156-fb7cc08b0d98"
   },
   "outputs": [],
   "source": [
    "len(dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "otvC3pkjl2PN",
    "outputId": "a2eb4462-0456-4ada-ae0a-3de91aac2b89"
   },
   "outputs": [],
   "source": [
    "train_data_loader = DataLoader(train_dataset, batch_size=128, shuffle=True)\n",
    "val_data_loader = DataLoader(val_dataset, batch_size=128, shuffle=True)\n",
    "len(train_data_loader)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "DetHchcb43Gc"
   },
   "source": [
    "## XceptionNet Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = timm.create_model('xception', pretrained=True, num_classes=1)\n",
    "model = model.to(device)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "TlETlqPcdGcs"
   },
   "source": [
    "## Optimizer and Loss Function\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "CJ5KcJeEHB71"
   },
   "outputs": [],
   "source": [
    "optimizer = torch.optim.Adam(model.parameters(),lr = 0.0001)\n",
    "criterion = nn.BCEWithLogitsLoss()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "t4dTS9b0dKen"
   },
   "source": [
    "## Training code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_losses, train_accs, val_losses, val_accs = [], [], [], []\n",
    "\n",
    "for epoch in range(20):\n",
    "    print(\"=================\")\n",
    "    print(f\"Epoch: {epoch+1}\")\n",
    "\n",
    "    model.train()\n",
    "    avg_loss_train = 0\n",
    "    correct = 0\n",
    "    total = 0\n",
    "  \n",
    "    for i, data in enumerate(train_data_loader, 0):\n",
    "    \n",
    "        inputs,labels = data\n",
    "        inputs = inputs.to(device)\n",
    "        labels = labels.to(device)\n",
    "        outputs = model(inputs.float())\n",
    "        outputs = torch.sigmoid(outputs) \n",
    "    \n",
    "        optimizer.zero_grad()\n",
    "        loss = criterion(outputs, labels.float())\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        avg_loss_train += loss.item()\n",
    "        predicted = (outputs > 0.5).float()  \n",
    "        correct += (predicted == labels).sum().item()\n",
    "        total += labels.size(0)\n",
    "    epoch_loss = avg_loss_train / len(train_data_loader)\n",
    "    epoch_acc = correct / total\n",
    "\n",
    "    train_losses.append(epoch_loss)\n",
    "    train_accs.append(epoch_acc)\n",
    "\n",
    "    print(f\"Train Loss: {epoch_loss} | Train Accuracy: {epoch_acc}\")    \n",
    "\n",
    "    model.eval()\n",
    "    running_loss = 0\n",
    "    correct = 0\n",
    "    total = 0\n",
    "    with torch.no_grad():\n",
    "        for data in val_data_loader:\n",
    "            inputs,labels = data\n",
    "            inputs = inputs.to(device)\n",
    "            labels = labels.to(device)\n",
    "            outputs = model(inputs.float())\n",
    "            outputs = torch.sigmoid(outputs)  \n",
    "        \n",
    "            optimizer.zero_grad()\n",
    "            loss = criterion(outputs, labels.float())\n",
    "            running_loss += loss.item()\n",
    "            \n",
    "            predicted = (outputs > 0.5).float()  \n",
    "            correct += (predicted == labels).sum().item()\n",
    "            total += labels.size(0)\n",
    "        epoch_loss = running_loss / len(val_data_loader)\n",
    "        epoch_acc = correct / total\n",
    "    val_losses.append(epoch_loss)\n",
    "    val_accs.append(epoch_acc)\n",
    "\n",
    "    print(f\"Val Loss: {epoch_loss} | Val Accuracy: {epoch_acc}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "def plot_curves(train_losses, train_accuracies, val_losses, val_accuracies):\n",
    "\n",
    "  fig, (ax1, ax2) = plt.subplots(nrows=2, ncols=1, figsize=(8, 8))\n",
    "\n",
    "  # Plot the train and test losses\n",
    "  ax1.plot(train_losses, label='Train')\n",
    "  ax1.plot(val_losses, label='Val')\n",
    "  ax1.set_xlabel('Epoch')\n",
    "  ax1.set_ylabel('Loss')\n",
    "  ax1.set_title('Losses vs Epochs')\n",
    "  ax1.legend()\n",
    "\n",
    "  # Plot the train and test accuracies\n",
    "  ax2.plot(train_accuracies, label='Train')\n",
    "  ax2.plot(val_accuracies, label='Val')\n",
    "  ax2.set_xlabel('Epoch')\n",
    "  ax2.set_ylabel('Accuracy')\n",
    "  ax2.set_title('Accuracy vs Epochs')\n",
    "  ax2.legend()\n",
    "\n",
    "  # Show the plot\n",
    "  plt.show()\n",
    "  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "rpd5KbkurR_y"
   },
   "outputs": [],
   "source": [
    "torch.save({\n",
    "            'model': model,\n",
    "            'model_state_dict': model.state_dict(),\n",
    "            'train_losses' : train_losses,\n",
    "            'train_accs' : train_accs,\n",
    "            'val_losses' : val_losses,\n",
    "            'val_accs' : val_accs,\n",
    "            }, 'CelebDF.pt')\n",
    "print('model saved')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.11"
  },
  "vscode": {
   "interpreter": {
    "hash": "703fc444efd57af823539a0730dab3b55a3cc9cc45981e7b1b33eb3f139bc8b4"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}