{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n\n# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Dependencies","metadata":{}},{"cell_type":"code","source":"import torch\nimport torch.nn as nn\nimport torchvision.transforms.functional as TF\n\nimport os\nfrom PIL import Image\nfrom torch.utils.data import Dataset\nimport numpy as np\n\nfrom glob import glob\nimport zipfile\nimport shutil\n\nfrom tqdm import tqdm\nimport albumentations as A\nfrom albumentations.pytorch import ToTensorV2\nimport torch.optim as optim\n\nfrom torch.utils.data import DataLoader\nimport torchvision","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:40:21.846711Z","iopub.execute_input":"2023-08-11T09:40:21.847078Z","iopub.status.idle":"2023-08-11T09:40:27.982954Z","shell.execute_reply.started":"2023-08-11T09:40:21.847048Z","shell.execute_reply":"2023-08-11T09:40:27.981953Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Data","metadata":{}},{"cell_type":"code","source":"path_to_zip_file = \"/kaggle/input/carvana-image-masking-challenge/train.zip\"\ndirectory_to_extract_to = \"/kaggle/working/\"\nwith zipfile.ZipFile(path_to_zip_file, 'r') as zip_ref:\n    zip_ref.extractall(directory_to_extract_to)\n    \npath_to_zip_file = \"/kaggle/input/carvana-image-masking-challenge/train_masks.zip\"\ndirectory_to_extract_to = \"/kaggle/working/\"\nwith zipfile.ZipFile(path_to_zip_file, 'r') as zip_ref:\n    zip_ref.extractall(directory_to_extract_to)","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:40:27.984941Z","iopub.execute_input":"2023-08-11T09:40:27.985314Z","iopub.status.idle":"2023-08-11T09:40:38.705718Z","shell.execute_reply.started":"2023-08-11T09:40:27.985280Z","shell.execute_reply":"2023-08-11T09:40:38.704603Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"os.mkdir('/kaggle/working/val')\nos.mkdir('/kaggle/working/val_masks')\n\nfor file in sorted(os.listdir('/kaggle/working/train'))[:1520]:\n  shutil.move('/kaggle/working/train/' + file, '/kaggle/working/val')\n\nfor file in sorted(os.listdir('/kaggle/working/train_masks'))[:1520]:\n  shutil.move('/kaggle/working/train_masks/' + file, '/kaggle/working/val_masks')","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:40:38.707083Z","iopub.execute_input":"2023-08-11T09:40:38.707476Z","iopub.status.idle":"2023-08-11T09:40:38.900217Z","shell.execute_reply.started":"2023-08-11T09:40:38.707441Z","shell.execute_reply":"2023-08-11T09:40:38.899192Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"os.mkdir('/kaggle/working/saved_images')","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:40:38.903025Z","iopub.execute_input":"2023-08-11T09:40:38.903423Z","iopub.status.idle":"2023-08-11T09:40:38.909532Z","shell.execute_reply.started":"2023-08-11T09:40:38.903387Z","shell.execute_reply":"2023-08-11T09:40:38.908546Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Model","metadata":{}},{"cell_type":"code","source":"class DoubleConv(nn.Module):\n  def __init__(self, in_channels, out_channels):\n    super(DoubleConv, self).__init__()\n    self.conv = nn.Sequential(\n        nn.Conv2d(in_channels, out_channels, 3, 1, 1, bias=False),\n        nn.BatchNorm2d(out_channels),\n        nn.ReLU(inplace=True),\n        nn.Conv2d(out_channels, out_channels, 3, 1, 1, bias=False),\n        nn.BatchNorm2d(out_channels),\n        nn.ReLU(inplace=True),\n    )\n\n  def forward(self, x):\n    return self.conv(x)\n\nclass UNET(nn.Module):\n  def __init__(\n      self, in_channels=3, out_channels=1, features=[64,128,256,512],\n  ):\n\n    super(UNET, self).__init__()\n    self.ups = nn.ModuleList()\n    self.downs = nn.ModuleList()\n    self.pool = nn.MaxPool2d(kernel_size=2, stride=2)\n\n    for feature in features:\n      self.downs.append(DoubleConv(in_channels, feature))\n      in_channels = feature\n\n    for feature in reversed(features):\n      self.ups.append(\n          nn.ConvTranspose2d(\n              feature*2, feature, kernel_size=2, stride=2,\n          )\n      )\n\n      self.ups.append(DoubleConv(feature*2, feature))\n\n      self.bottleneck = DoubleConv(features[-1], features[-1]*2)\n\n      self.final_conv = nn.Conv2d(features[0], out_channels, kernel_size=1)\n\n  def forward(self, x):\n    skip_connections = []\n\n    for down in self.downs:\n      x = down(x)\n      skip_connections.append(x)\n      x = self.pool(x)\n\n    x = self.bottleneck(x)\n    skip_connections = skip_connections[::-1]\n\n    for idx in range(0, len(self.ups),2):\n      x = self.ups[idx](x)\n      skip_connection = skip_connections[idx//2]\n\n      if x.shape != skip_connection.shape:\n        x = TF.resize(x, size=skip_connection.shape[2:])\n\n      concat_skip = torch.cat((skip_connection, x), dim=1)\n      x = self.ups[idx+1](concat_skip)\n\n    return self.final_conv(x)\n\ndef test():\n  x = torch.rand((3,1,161,161))\n  model = UNET(in_channels=1, out_channels=1)\n  preds = model(x)\n  print(preds.shape)\n  print(x.shape)\n  assert preds.shape == x.shape\n\n\nif __name__ == '__main__':\n  test()\n\n","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:40:38.911070Z","iopub.execute_input":"2023-08-11T09:40:38.911837Z","iopub.status.idle":"2023-08-11T09:40:41.590740Z","shell.execute_reply.started":"2023-08-11T09:40:38.911802Z","shell.execute_reply":"2023-08-11T09:40:41.589558Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Dataset","metadata":{}},{"cell_type":"code","source":"class CarvanaDataset(Dataset):\n  def __init__(self, image_dir, mask_dir, transform=None):\n    self.image_dir = image_dir\n    self.mask_dir  = mask_dir\n    self.transform = transform\n    self.images = os.listdir(image_dir)\n\n  def __len__(self):\n    return len(self.images)\n\n  def __getitem__(self, index):\n    img_path = os.path.join(self.image_dir, self.images[index])\n    mask_path = os.path.join(self.mask_dir, self.images[index]).replace(\".jpg\", \"_mask.gif\")\n    image = np.array(Image.open(img_path).convert(\"RGB\"))\n    mask = np.array(Image.open(mask_path).convert(\"L\"), dtype=np.float32)\n    mask[mask == 255.0] = 1.0\n\n    if self.transform is not None:\n      augmentations = self.transform(image=image, mask=mask)\n      image = augmentations[\"image\"]\n      mask = augmentations[\"mask\"]\n\n    return image, mask\n\n","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:40:41.592488Z","iopub.execute_input":"2023-08-11T09:40:41.592861Z","iopub.status.idle":"2023-08-11T09:40:41.602884Z","shell.execute_reply.started":"2023-08-11T09:40:41.592827Z","shell.execute_reply":"2023-08-11T09:40:41.601786Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Train","metadata":{}},{"cell_type":"code","source":"# Hyperparameters\nLEARNING_RATE = 1e-4\nDEVICE = \"cuda\" if torch.cuda.is_available() else \"cpu\"\nBATCH_SIZE = 16\nNUM_EPOCHS = 3\nNUM_WORKERS = 2\nIMAGE_HEIGHT = 320\nIMAGE_WIDTH = 480\nPIN_MEMORY = True\nLOAD_MODEL = False\nTRAIN_IMG_DIR = '/kaggle/working/train'\nTRAIN_MASK_DIR = '/kaggle/working/train_masks'\nVAL_IMG_DIR = '/kaggle/working/val'\nVAL_MASK_DIR = '/kaggle/working/val_masks'","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:40:41.604134Z","iopub.execute_input":"2023-08-11T09:40:41.604408Z","iopub.status.idle":"2023-08-11T09:40:41.684627Z","shell.execute_reply.started":"2023-08-11T09:40:41.604384Z","shell.execute_reply":"2023-08-11T09:40:41.683606Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def save_checkpoint(state, filename=\"my_checkpoint.pth.tar\"):\n  print(\"=> Saving Checkpoint\")\n  torch.save(state, filename)\n\ndef load_checkpoint(checkpoint, model):\n  print(\"=> Loading Checkpoint\")\n  model.load_state_dict(checkpoint[\"state_dict\"])\n\ndef get_loaders(\n    train_dir,\n    train_maskdir,\n    val_dir,\n    val_maskdir,\n    batch_size,\n    train_transform,\n    val_transform,\n    num_workers=4,\n    pin_memory=True\n):\n  train_ds = CarvanaDataset(\n      image_dir=train_dir,\n      mask_dir=train_maskdir,\n      transform=train_transform,\n  )\n\n  train_loader = DataLoader(\n      train_ds,\n      batch_size=batch_size,\n      num_workers=num_workers,\n      pin_memory=pin_memory,\n      shuffle=True,\n  )\n\n  val_ds = CarvanaDataset(\n      image_dir=val_dir,\n      mask_dir=val_maskdir,\n      transform=val_transform,\n\n  )\n\n  val_loader = DataLoader(\n      val_ds,\n      batch_size=batch_size,\n      num_workers=num_workers,\n      pin_memory=pin_memory,\n      shuffle=False,\n  )\n\n  return train_loader, val_loader\n\ndef check_accuracy(loader, model, device=\"cuda\"):\n  num_correct = 0\n  num_pixels = 0\n  model.eval()\n\n  with torch.no_grad():\n    for x, y in loader:\n      x = x.to(device)\n      y = y.to(device).unsqueeze(1)\n      preds = torch.sigmoid(model(x))\n      preds = (preds > 0.5).float()\n      num_correct += (preds == y).sum()\n      num_pixels += torch.numel(preds)\n      # dice_score += (2* (preds*y).sum()) / (\n      #     (preds + y).sum() + 1e-8\n      # )\n\n    print(\n        f\"Got {num_correct}/{num_pixels} with acc {num_correct/num_pixels*100:.2f}\"\n    )\n\n    # print(f\"Dice Score: {dice_score/len(loader)}\")\n\n    model.train()\n\n\ndef save_predictions_as_imgs(\n    loader, model, folder=\"saved_images/\", device=\"cuda\"\n):\n    model.eval()\n    for idx, (x, y) in enumerate(loader):\n        x = x.to(device=device)\n        with torch.no_grad():\n            preds = torch.sigmoid(model(x))\n            preds = (preds > 0.5).float()\n        torchvision.utils.save_image(\n            preds, f\"{folder}/pred_{idx}.png\"\n        )\n        torchvision.utils.save_image(y.unsqueeze(1), f\"{folder}{idx}.png\")\n\n\n    torchvision.utils.save_image(y.unsqueeze(1), f\"{folder}{idx}.png\")\n\n","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:40:41.686149Z","iopub.execute_input":"2023-08-11T09:40:41.686543Z","iopub.status.idle":"2023-08-11T09:40:41.703392Z","shell.execute_reply.started":"2023-08-11T09:40:41.686503Z","shell.execute_reply":"2023-08-11T09:40:41.702359Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def train_fn(loader, model, optimizer, loss_fn, scaler):\n    loop = tqdm(loader)\n\n    for batch_idx, (data, targets) in enumerate(loop):\n        data = data.to(device=DEVICE)\n        targets = targets.float().unsqueeze(1).to(device=DEVICE)\n\n        # forward\n        with torch.cuda.amp.autocast():\n            predictions = model(data)\n            loss = loss_fn(predictions, targets)\n\n        # backward\n        optimizer.zero_grad()\n        scaler.scale(loss).backward()\n        scaler.step(optimizer)\n        scaler.update()\n\n        # update tqdm loop\n        loop.set_postfix(loss=loss.item())\n\n\ndef main():\n    train_transform = A.Compose(\n        [\n            A.Resize(height=IMAGE_HEIGHT, width=IMAGE_WIDTH),\n            A.Rotate(limit=35, p=1.0),\n            A.HorizontalFlip(p=0.5),\n            A.VerticalFlip(p=0.1),\n            A.Normalize(\n                mean=[0.0, 0.0, 0.0],\n                std=[1.0, 1.0, 1.0],\n                max_pixel_value=255.0,\n            ),\n            ToTensorV2(),\n        ],\n    )\n\n    val_transforms = A.Compose(\n        [\n            A.Resize(height=IMAGE_HEIGHT, width=IMAGE_WIDTH),\n            A.Normalize(\n                mean=[0.0, 0.0, 0.0],\n                std=[1.0, 1.0, 1.0],\n                max_pixel_value=255.0,\n            ),\n            ToTensorV2(),\n        ],\n    )\n\n    model = UNET(in_channels=3, out_channels=1).to(DEVICE)\n    loss_fn = nn.BCEWithLogitsLoss()\n    optimizer = optim.Adam(model.parameters(), lr=LEARNING_RATE)\n\n    train_loader, val_loader = get_loaders(\n        TRAIN_IMG_DIR,\n        TRAIN_MASK_DIR,\n        VAL_IMG_DIR,\n        VAL_MASK_DIR,\n        BATCH_SIZE,\n        train_transform,\n        val_transforms,\n        NUM_WORKERS,\n        PIN_MEMORY,\n    )\n\n    if LOAD_MODEL:\n        load_checkpoint(torch.load(\"my_checkpoint.pth.tar\"), model)\n\n\n    # check_accuracy(val_loader, model, device=DEVICE)\n    scaler = torch.cuda.amp.GradScaler()\n\n    for epoch in range(NUM_EPOCHS):\n        train_fn(train_loader, model, optimizer, loss_fn, scaler)\n\n        # save model\n        checkpoint = {\n            \"state_dict\": model.state_dict(),\n            \"optimizer\":optimizer.state_dict(),\n        }\n        save_checkpoint(checkpoint)\n\n        # check accuracy\n        check_accuracy(val_loader, model, device=DEVICE)\n\n        # print some examples to a folder\n        save_predictions_as_imgs(\n            val_loader, model, folder=\"saved_images/\", device=DEVICE\n        )\n\n\nif __name__ == \"__main__\":\n    main()","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:40:41.704937Z","iopub.execute_input":"2023-08-11T09:40:41.705332Z","iopub.status.idle":"2023-08-11T09:58:16.336696Z","shell.execute_reply.started":"2023-08-11T09:40:41.705298Z","shell.execute_reply":"2023-08-11T09:58:16.335514Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Hyperparameters\nLEARNING_RATE = 1e-4\nDEVICE = \"cuda\" if torch.cuda.is_available() else \"cpu\"\nBATCH_SIZE = 16\nNUM_EPOCHS = 10\nNUM_WORKERS = 1\nIMAGE_HEIGHT = 320\nIMAGE_WIDTH = 480\nPIN_MEMORY = True\nLOAD_MODEL = True\nTRAIN_IMG_DIR = '/kaggle/working/train'\nTRAIN_MASK_DIR = '/kaggle/working/train_masks'\nVAL_IMG_DIR = '/kaggle/working/val'\nVAL_MASK_DIR = '/kaggle/working/val_masks'","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:58:16.340793Z","iopub.execute_input":"2023-08-11T09:58:16.341207Z","iopub.status.idle":"2023-08-11T09:58:16.347940Z","shell.execute_reply.started":"2023-08-11T09:58:16.341171Z","shell.execute_reply":"2023-08-11T09:58:16.346706Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def train_fn(loader, model, optimizer, loss_fn, scaler):\n    loop = tqdm(loader)\n\n    for batch_idx, (data, targets) in enumerate(loop):\n        data = data.to(device=DEVICE)\n        targets = targets.float().unsqueeze(1).to(device=DEVICE)\n\n        # forward\n        with torch.cuda.amp.autocast():\n            predictions = model(data)\n            loss = loss_fn(predictions, targets)\n\n        # backward\n        optimizer.zero_grad()\n        scaler.scale(loss).backward()\n        scaler.step(optimizer)\n        scaler.update()\n\n        # update tqdm loop\n        loop.set_postfix(loss=loss.item())\n\n\ndef main():\n    train_transform = A.Compose(\n        [\n            A.Resize(height=IMAGE_HEIGHT, width=IMAGE_WIDTH),\n            A.Rotate(limit=35, p=1.0),\n            A.HorizontalFlip(p=0.5),\n            A.VerticalFlip(p=0.1),\n            A.Normalize(\n                mean=[0.0, 0.0, 0.0],\n                std=[1.0, 1.0, 1.0],\n                max_pixel_value=255.0,\n            ),\n            ToTensorV2(),\n        ],\n    )\n\n    val_transforms = A.Compose(\n        [\n            A.Resize(height=IMAGE_HEIGHT, width=IMAGE_WIDTH),\n            A.Normalize(\n                mean=[0.0, 0.0, 0.0],\n                std=[1.0, 1.0, 1.0],\n                max_pixel_value=255.0,\n            ),\n            ToTensorV2(),\n        ],\n    )\n\n    model = UNET(in_channels=3, out_channels=1).to(DEVICE)\n    loss_fn = nn.BCEWithLogitsLoss()\n    optimizer = optim.Adam(model.parameters(), lr=LEARNING_RATE)\n\n    train_loader, val_loader = get_loaders(\n        TRAIN_IMG_DIR,\n        TRAIN_MASK_DIR,\n        VAL_IMG_DIR,\n        VAL_MASK_DIR,\n        BATCH_SIZE,\n        train_transform,\n        val_transforms,\n        NUM_WORKERS,\n        PIN_MEMORY,\n    )\n\n    if LOAD_MODEL:\n        load_checkpoint(torch.load(\"my_checkpoint.pth.tar\"), model)\n\n\n    check_accuracy(val_loader, model, device=DEVICE)\n    scaler = torch.cuda.amp.GradScaler()\n\n    for epoch in range(NUM_EPOCHS):\n        train_fn(train_loader, model, optimizer, loss_fn, scaler)\n\n        # save model\n        checkpoint = {\n            \"state_dict\": model.state_dict(),\n            \"optimizer\":optimizer.state_dict(),\n        }\n        save_checkpoint(checkpoint)\n\n        # check accuracy\n        check_accuracy(val_loader, model, device=DEVICE)\n\n        # print some examples to a folder\n        save_predictions_as_imgs(\n            val_loader, model, folder=\"saved_images/\", device=DEVICE\n        )\n\n\nif __name__ == \"__main__\":\n    main()","metadata":{"execution":{"iopub.status.busy":"2023-08-11T09:58:16.349801Z","iopub.execute_input":"2023-08-11T09:58:16.350150Z","iopub.status.idle":"2023-08-11T10:58:21.340445Z","shell.execute_reply.started":"2023-08-11T09:58:16.350101Z","shell.execute_reply":"2023-08-11T10:58:21.339162Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}