UNet_training.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "08c53233",
   "metadata": {},
   "source": [
    "# Sources\n",
    "* [usuyama/pytorch-unet](https://github.com/usuyama/pytorch-unet) from this repo we used training loop, simulated dataset, and other helpers for visualization.\n",
    "* [milesial/PyTorch-Unet](https://github.com/milesial/Pytorch-UNet/tree/master/unet) from this repo we used the `unet` module for neural network architecture.\n",
    "* [Original UNet paper](https://arxiv.org/pdf/1505.04597.pdf)\n",
    "* For more information on Dice loss and Binary Cross Entropy loss see the notebook `dice_loss_bce_loss.ipynb`.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f969d2e7",
   "metadata": {},
   "source": [
    "![alt text](UNet.png \"UNet\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17fa195a",
   "metadata": {},
   "source": [
    "# Visualize input dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "af3c06d7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3, 192, 192, 3)\n",
      "0 255\n",
      "(3, 6, 192, 192)\n",
      "0.0 1.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x1200 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import helper\n",
    "import simulation\n",
    "\n",
    "# Generate some random images\n",
    "input_images, target_masks = simulation.generate_random_data(192, 192, count=3)\n",
    "\n",
    "for x in [input_images, target_masks]:\n",
    "    print(x.shape)\n",
    "    print(x.min(), x.max())\n",
    "\n",
    "# Change channel-order and make 3 channels for matplot\n",
    "input_images_rgb = [x.astype(np.uint8) for x in input_images]\n",
    "\n",
    "# Map each channel (i.e. class) to each color\n",
    "target_masks_rgb = [helper.masks_to_colorimg(x) for x in target_masks]\n",
    "\n",
    "# Left: Input image (black and white), Right: Target mask (6ch)\n",
    "helper.plot_side_by_side([input_images_rgb, target_masks_rgb])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7095db63",
   "metadata": {},
   "source": [
    "# Prepare Dataset and Dataloader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "359aadb7",
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.utils.data import Dataset, DataLoader\n",
    "from torchvision import transforms, datasets, models\n",
    "\n",
    "class SimDataset(Dataset):\n",
    "    def __init__(self, count, transform=None):\n",
    "        self.input_images, self.target_masks = simulation.generate_random_data(192, 192, count=count)\n",
    "        self.transform = transform\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.input_images)\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        image = self.input_images[idx]\n",
    "        mask = self.target_masks[idx]\n",
    "        if self.transform:\n",
    "            image = self.transform(image)\n",
    "\n",
    "        return [image, mask]\n",
    "\n",
    "# use the same transformations for train/val in this example\n",
    "trans = transforms.Compose([\n",
    "    transforms.ToTensor(),\n",
    "    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]) # imagenet\n",
    "])\n",
    "\n",
    "train_set = SimDataset(2000, transform = trans)\n",
    "val_set = SimDataset(200, transform = trans)\n",
    "\n",
    "image_datasets = {\n",
    "    'train': train_set, 'val': val_set\n",
    "}\n",
    "\n",
    "batch_size = 25\n",
    "\n",
    "dataloaders = {\n",
    "    'train': DataLoader(train_set, batch_size=batch_size, shuffle=True, num_workers=0),\n",
    "    'val': DataLoader(val_set, batch_size=batch_size, shuffle=True, num_workers=0)\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c3cb59f3",
   "metadata": {},
   "source": [
    "# Check the output from Dataloader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "d5f60a16",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([25, 3, 192, 192]) torch.Size([25, 6, 192, 192])\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f75d8dd0f70>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import torchvision.utils\n",
    "\n",
    "def reverse_transform(inp):\n",
    "    inp = inp.numpy().transpose((1, 2, 0))\n",
    "    mean = np.array([0.485, 0.456, 0.406])\n",
    "    std = np.array([0.229, 0.224, 0.225])\n",
    "    inp = std * inp + mean\n",
    "    inp = np.clip(inp, 0, 1)\n",
    "    inp = (inp * 255).astype(np.uint8)\n",
    "\n",
    "    return inp\n",
    "\n",
    "# Get a batch of training data\n",
    "inputs, masks = next(iter(dataloaders['train']))\n",
    "\n",
    "print(inputs.shape, masks.shape)\n",
    "\n",
    "plt.imshow(reverse_transform(inputs[3]))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "06c6a937",
   "metadata": {},
   "source": [
    "# Prepare a `UNet` model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "5bbcb5f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "from unet import UNet\n",
    "import torch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "039bc4cd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------\n",
      "        Layer (type)               Output Shape         Param #\n",
      "================================================================\n",
      "            Conv2d-1         [-1, 64, 224, 224]           1,728\n",
      "       BatchNorm2d-2         [-1, 64, 224, 224]             128\n",
      "              ReLU-3         [-1, 64, 224, 224]               0\n",
      "            Conv2d-4         [-1, 64, 224, 224]          36,864\n",
      "       BatchNorm2d-5         [-1, 64, 224, 224]             128\n",
      "              ReLU-6         [-1, 64, 224, 224]               0\n",
      "        DoubleConv-7         [-1, 64, 224, 224]               0\n",
      "         MaxPool2d-8         [-1, 64, 112, 112]               0\n",
      "            Conv2d-9        [-1, 128, 112, 112]          73,728\n",
      "      BatchNorm2d-10        [-1, 128, 112, 112]             256\n",
      "             ReLU-11        [-1, 128, 112, 112]               0\n",
      "           Conv2d-12        [-1, 128, 112, 112]         147,456\n",
      "      BatchNorm2d-13        [-1, 128, 112, 112]             256\n",
      "             ReLU-14        [-1, 128, 112, 112]               0\n",
      "       DoubleConv-15        [-1, 128, 112, 112]               0\n",
      "             Down-16        [-1, 128, 112, 112]               0\n",
      "        MaxPool2d-17          [-1, 128, 56, 56]               0\n",
      "           Conv2d-18          [-1, 256, 56, 56]         294,912\n",
      "      BatchNorm2d-19          [-1, 256, 56, 56]             512\n",
      "             ReLU-20          [-1, 256, 56, 56]               0\n",
      "           Conv2d-21          [-1, 256, 56, 56]         589,824\n",
      "      BatchNorm2d-22          [-1, 256, 56, 56]             512\n",
      "             ReLU-23          [-1, 256, 56, 56]               0\n",
      "       DoubleConv-24          [-1, 256, 56, 56]               0\n",
      "             Down-25          [-1, 256, 56, 56]               0\n",
      "        MaxPool2d-26          [-1, 256, 28, 28]               0\n",
      "           Conv2d-27          [-1, 512, 28, 28]       1,179,648\n",
      "      BatchNorm2d-28          [-1, 512, 28, 28]           1,024\n",
      "             ReLU-29          [-1, 512, 28, 28]               0\n",
      "           Conv2d-30          [-1, 512, 28, 28]       2,359,296\n",
      "      BatchNorm2d-31          [-1, 512, 28, 28]           1,024\n",
      "             ReLU-32          [-1, 512, 28, 28]               0\n",
      "       DoubleConv-33          [-1, 512, 28, 28]               0\n",
      "             Down-34          [-1, 512, 28, 28]               0\n",
      "        MaxPool2d-35          [-1, 512, 14, 14]               0\n",
      "           Conv2d-36          [-1, 512, 14, 14]       2,359,296\n",
      "      BatchNorm2d-37          [-1, 512, 14, 14]           1,024\n",
      "             ReLU-38          [-1, 512, 14, 14]               0\n",
      "           Conv2d-39          [-1, 512, 14, 14]       2,359,296\n",
      "      BatchNorm2d-40          [-1, 512, 14, 14]           1,024\n",
      "             ReLU-41          [-1, 512, 14, 14]               0\n",
      "       DoubleConv-42          [-1, 512, 14, 14]               0\n",
      "             Down-43          [-1, 512, 14, 14]               0\n",
      "         Upsample-44          [-1, 512, 28, 28]               0\n",
      "           Conv2d-45          [-1, 512, 28, 28]       4,718,592\n",
      "      BatchNorm2d-46          [-1, 512, 28, 28]           1,024\n",
      "             ReLU-47          [-1, 512, 28, 28]               0\n",
      "           Conv2d-48          [-1, 256, 28, 28]       1,179,648\n",
      "      BatchNorm2d-49          [-1, 256, 28, 28]             512\n",
      "             ReLU-50          [-1, 256, 28, 28]               0\n",
      "       DoubleConv-51          [-1, 256, 28, 28]               0\n",
      "               Up-52          [-1, 256, 28, 28]               0\n",
      "         Upsample-53          [-1, 256, 56, 56]               0\n",
      "           Conv2d-54          [-1, 256, 56, 56]       1,179,648\n",
      "      BatchNorm2d-55          [-1, 256, 56, 56]             512\n",
      "             ReLU-56          [-1, 256, 56, 56]               0\n",
      "           Conv2d-57          [-1, 128, 56, 56]         294,912\n",
      "      BatchNorm2d-58          [-1, 128, 56, 56]             256\n",
      "             ReLU-59          [-1, 128, 56, 56]               0\n",
      "       DoubleConv-60          [-1, 128, 56, 56]               0\n",
      "               Up-61          [-1, 128, 56, 56]               0\n",
      "         Upsample-62        [-1, 128, 112, 112]               0\n",
      "           Conv2d-63        [-1, 128, 112, 112]         294,912\n",
      "      BatchNorm2d-64        [-1, 128, 112, 112]             256\n",
      "             ReLU-65        [-1, 128, 112, 112]               0\n",
      "           Conv2d-66         [-1, 64, 112, 112]          73,728\n",
      "      BatchNorm2d-67         [-1, 64, 112, 112]             128\n",
      "             ReLU-68         [-1, 64, 112, 112]               0\n",
      "       DoubleConv-69         [-1, 64, 112, 112]               0\n",
      "               Up-70         [-1, 64, 112, 112]               0\n",
      "         Upsample-71         [-1, 64, 224, 224]               0\n",
      "           Conv2d-72         [-1, 64, 224, 224]          73,728\n",
      "      BatchNorm2d-73         [-1, 64, 224, 224]             128\n",
      "             ReLU-74         [-1, 64, 224, 224]               0\n",
      "           Conv2d-75         [-1, 64, 224, 224]          36,864\n",
      "      BatchNorm2d-76         [-1, 64, 224, 224]             128\n",
      "             ReLU-77         [-1, 64, 224, 224]               0\n",
      "       DoubleConv-78         [-1, 64, 224, 224]               0\n",
      "               Up-79         [-1, 64, 224, 224]               0\n",
      "           Conv2d-80          [-1, 6, 224, 224]             390\n",
      "          OutConv-81          [-1, 6, 224, 224]               0\n",
      "================================================================\n",
      "Total params: 17,263,302\n",
      "Trainable params: 17,263,302\n",
      "Non-trainable params: 0\n",
      "----------------------------------------------------------------\n",
      "Input size (MB): 0.57\n",
      "Forward/backward pass size (MB): 725.05\n",
      "Params size (MB): 65.85\n",
      "Estimated Total Size (MB): 791.48\n",
      "----------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "model =  net = UNet(n_channels=3, n_classes=6, bilinear=True)\n",
    "model = model.to(device)\n",
    "\n",
    "# check keras-like model summary using torchsummary\n",
    "from torchsummary import summary\n",
    "summary(model, input_size=(3, 224, 224))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6f24e124",
   "metadata": {},
   "source": [
    "# Define the main training loop"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "161180c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "from collections import defaultdict\n",
    "import torch.nn.functional as F\n",
    "from loss import dice_loss\n",
    "\n",
    "def calc_loss(pred, target, metrics, bce_weight=0.5):\n",
    "    bce = F.binary_cross_entropy_with_logits(pred, target)\n",
    "\n",
    "    pred = F.sigmoid(pred)\n",
    "    dice = dice_loss(pred, target)\n",
    "\n",
    "    loss = bce * bce_weight + dice * (1 - bce_weight)\n",
    "\n",
    "    metrics['bce'] += bce.data.cpu().numpy() * target.size(0)\n",
    "    metrics['dice'] += dice.data.cpu().numpy() * target.size(0)\n",
    "    metrics['loss'] += loss.data.cpu().numpy() * target.size(0)\n",
    "\n",
    "    return loss\n",
    "\n",
    "def print_metrics(metrics, epoch_samples, phase):\n",
    "    outputs = []\n",
    "    for k in metrics.keys():\n",
    "        outputs.append(\"{}: {:4f}\".format(k, metrics[k] / epoch_samples))\n",
    "\n",
    "    print(\"{}: {}\".format(phase, \", \".join(outputs)))\n",
    "\n",
    "def train_model(model, optimizer, scheduler, num_epochs=25):\n",
    "    best_model_wts = copy.deepcopy(model.state_dict())\n",
    "    best_loss = 1e10\n",
    "\n",
    "    for epoch in range(num_epochs):\n",
    "        print('Epoch {}/{}'.format(epoch, num_epochs - 1))\n",
    "        print('-' * 10)\n",
    "\n",
    "        since = time.time()\n",
    "\n",
    "        # Each epoch has a training and validation phase\n",
    "        for phase in ['train', 'val']:\n",
    "            if phase == 'train':\n",
    "                scheduler.step()\n",
    "                for param_group in optimizer.param_groups:\n",
    "                    print(\"LR\", param_group['lr'])\n",
    "\n",
    "                model.train()  # Set model to training mode\n",
    "            else:\n",
    "                model.eval()   # Set model to evaluate mode\n",
    "\n",
    "            metrics = defaultdict(float)\n",
    "            epoch_samples = 0\n",
    "\n",
    "            for inputs, labels in dataloaders[phase]:\n",
    "                inputs = inputs.to(device)\n",
    "                labels = labels.to(device)\n",
    "\n",
    "                # zero the parameter gradients\n",
    "                optimizer.zero_grad()\n",
    "\n",
    "                # forward\n",
    "                # track history if only in train\n",
    "                with torch.set_grad_enabled(phase == 'train'):\n",
    "                    outputs = model(inputs)\n",
    "                    loss = calc_loss(outputs, labels, metrics)\n",
    "\n",
    "                    # backward + optimize only if in training phase\n",
    "                    if phase == 'train':\n",
    "                        loss.backward()\n",
    "                        optimizer.step()\n",
    "\n",
    "                # statistics\n",
    "                epoch_samples += inputs.size(0)\n",
    "\n",
    "            print_metrics(metrics, epoch_samples, phase)\n",
    "            epoch_loss = metrics['loss'] / epoch_samples\n",
    "\n",
    "            # deep copy the model\n",
    "            if phase == 'val' and epoch_loss < best_loss:\n",
    "                print(\"saving best model\")\n",
    "                best_loss = epoch_loss\n",
    "                best_model_wts = copy.deepcopy(model.state_dict())\n",
    "\n",
    "        time_elapsed = time.time() - since\n",
    "        print('{:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60))\n",
    "\n",
    "    print('Best val loss: {:4f}'.format(best_loss))\n",
    "\n",
    "    # load best model weights\n",
    "    model.load_state_dict(best_model_wts)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "df10db71",
   "metadata": {},
   "source": [
    "# Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "55800a9d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda:0\n",
      "Epoch 0/59\n",
      "----------\n",
      "LR 0.0001\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/ruthvik/pytorch_1121_cu113_unet/lib/python3.10/site-packages/torch/optim/lr_scheduler.py:138: UserWarning: Detected call of `lr_scheduler.step()` before `optimizer.step()`. In PyTorch 1.1.0 and later, you should call them in the opposite order: `optimizer.step()` before `lr_scheduler.step()`.  Failure to do this will result in PyTorch skipping the first value of the learning rate schedule. See more details at https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate\n",
      "  warnings.warn(\"Detected call of `lr_scheduler.step()` before `optimizer.step()`. \"\n",
      "/home/ruthvik/pytorch_1121_cu113_unet/lib/python3.10/site-packages/torch/nn/functional.py:1967: UserWarning: nn.functional.sigmoid is deprecated. Use torch.sigmoid instead.\n",
      "  warnings.warn(\"nn.functional.sigmoid is deprecated. Use torch.sigmoid instead.\")\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: bce: 0.514121, dice: 0.979471, loss: 0.746796\n",
      "val: bce: 0.459575, dice: 0.977623, loss: 0.718599\n",
      "saving best model\n",
      "0m 24s\n",
      "Epoch 1/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.414441, dice: 0.974459, loss: 0.694450\n",
      "val: bce: 0.379762, dice: 0.972408, loss: 0.676085\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 2/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.346147, dice: 0.970363, loss: 0.658255\n",
      "val: bce: 0.314018, dice: 0.968091, loss: 0.641054\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 3/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.286223, dice: 0.965362, loss: 0.625792\n",
      "val: bce: 0.260262, dice: 0.962879, loss: 0.611570\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 4/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.236052, dice: 0.958850, loss: 0.597451\n",
      "val: bce: 0.214615, dice: 0.955372, loss: 0.584994\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 5/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.193300, dice: 0.950856, loss: 0.572078\n",
      "val: bce: 0.174865, dice: 0.946818, loss: 0.560842\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 6/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.158289, dice: 0.941293, loss: 0.549791\n",
      "val: bce: 0.143509, dice: 0.936650, loss: 0.540080\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 7/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.129616, dice: 0.929892, loss: 0.529754\n",
      "val: bce: 0.116612, dice: 0.923524, loss: 0.520068\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 8/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.105848, dice: 0.915891, loss: 0.510869\n",
      "val: bce: 0.094700, dice: 0.907846, loss: 0.501273\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 9/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.085135, dice: 0.898073, loss: 0.491604\n",
      "val: bce: 0.076789, dice: 0.889412, loss: 0.483100\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 10/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.067377, dice: 0.874498, loss: 0.470937\n",
      "val: bce: 0.061009, dice: 0.863352, loss: 0.462180\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 11/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.052155, dice: 0.842970, loss: 0.447562\n",
      "val: bce: 0.045668, dice: 0.826096, loss: 0.435882\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 12/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.039189, dice: 0.800599, loss: 0.419894\n",
      "val: bce: 0.034532, dice: 0.779665, loss: 0.407099\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 13/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.028312, dice: 0.743353, loss: 0.385832\n",
      "val: bce: 0.023841, dice: 0.711595, loss: 0.367718\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 14/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.019662, dice: 0.669998, loss: 0.344830\n",
      "val: bce: 0.016245, dice: 0.629735, loss: 0.322990\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 15/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.013081, dice: 0.581533, loss: 0.297307\n",
      "val: bce: 0.010849, dice: 0.541328, loss: 0.276089\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 16/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.008507, dice: 0.486202, loss: 0.247355\n",
      "val: bce: 0.006809, dice: 0.438908, loss: 0.222858\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 17/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.005554, dice: 0.393626, loss: 0.199590\n",
      "val: bce: 0.004592, dice: 0.353371, loss: 0.178982\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 18/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.003677, dice: 0.307726, loss: 0.155701\n",
      "val: bce: 0.003272, dice: 0.277189, loss: 0.140231\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 19/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.002583, dice: 0.237196, loss: 0.119889\n",
      "val: bce: 0.002221, dice: 0.207080, loss: 0.104651\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 20/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.001921, dice: 0.183517, loss: 0.092719\n",
      "val: bce: 0.001861, dice: 0.170696, loss: 0.086278\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 21/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.001504, dice: 0.144810, loss: 0.073157\n",
      "val: bce: 0.001587, dice: 0.145820, loss: 0.073704\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 22/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.001222, dice: 0.117245, loss: 0.059233\n",
      "val: bce: 0.001188, dice: 0.107152, loss: 0.054170\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 23/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.001010, dice: 0.096083, loss: 0.048546\n",
      "val: bce: 0.001089, dice: 0.096059, loss: 0.048574\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 24/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.000918, dice: 0.082594, loss: 0.041756\n",
      "val: bce: 0.001013, dice: 0.083938, loss: 0.042476\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 25/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.000800, dice: 0.070385, loss: 0.035593\n",
      "val: bce: 0.000915, dice: 0.073555, loss: 0.037235\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 26/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.000729, dice: 0.061233, loss: 0.030981\n",
      "val: bce: 0.000748, dice: 0.056674, loss: 0.028711\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 27/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.000653, dice: 0.054125, loss: 0.027389\n",
      "val: bce: 0.000727, dice: 0.054520, loss: 0.027624\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 28/59\n",
      "----------\n",
      "LR 0.0001\n",
      "train: bce: 0.000611, dice: 0.048399, loss: 0.024505\n",
      "val: bce: 0.000697, dice: 0.048698, loss: 0.024698\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 29/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000531, dice: 0.044442, loss: 0.022487\n",
      "val: bce: 0.000591, dice: 0.046361, loss: 0.023476\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 30/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000489, dice: 0.042939, loss: 0.021714\n",
      "val: bce: 0.000608, dice: 0.046497, loss: 0.023553\n",
      "0m 25s\n",
      "Epoch 31/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000471, dice: 0.041914, loss: 0.021192\n",
      "val: bce: 0.000659, dice: 0.045738, loss: 0.023198\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 32/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000454, dice: 0.040950, loss: 0.020702\n",
      "val: bce: 0.000665, dice: 0.045378, loss: 0.023022\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 33/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000435, dice: 0.039932, loss: 0.020183\n",
      "val: bce: 0.000642, dice: 0.044491, loss: 0.022566\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 34/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000418, dice: 0.039077, loss: 0.019747\n",
      "val: bce: 0.000669, dice: 0.044763, loss: 0.022716\n",
      "0m 25s\n",
      "Epoch 35/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000399, dice: 0.038226, loss: 0.019313\n",
      "val: bce: 0.000662, dice: 0.043932, loss: 0.022297\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 36/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000382, dice: 0.037338, loss: 0.018860\n",
      "val: bce: 0.000693, dice: 0.044076, loss: 0.022385\n",
      "0m 25s\n",
      "Epoch 37/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000368, dice: 0.036623, loss: 0.018495\n",
      "val: bce: 0.000674, dice: 0.043401, loss: 0.022037\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 38/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000356, dice: 0.035880, loss: 0.018118\n",
      "val: bce: 0.000670, dice: 0.043155, loss: 0.021912\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 39/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000347, dice: 0.035242, loss: 0.017794\n",
      "val: bce: 0.000673, dice: 0.042581, loss: 0.021627\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 40/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000338, dice: 0.034633, loss: 0.017485\n",
      "val: bce: 0.000670, dice: 0.041766, loss: 0.021218\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 41/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000331, dice: 0.034038, loss: 0.017184\n",
      "val: bce: 0.000660, dice: 0.041630, loss: 0.021145\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 42/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000323, dice: 0.033383, loss: 0.016853\n",
      "val: bce: 0.000660, dice: 0.040803, loss: 0.020731\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 43/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000316, dice: 0.032824, loss: 0.016570\n",
      "val: bce: 0.000657, dice: 0.040832, loss: 0.020745\n",
      "0m 25s\n",
      "Epoch 44/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000310, dice: 0.032285, loss: 0.016297\n",
      "val: bce: 0.000664, dice: 0.039596, loss: 0.020130\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 45/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000303, dice: 0.031664, loss: 0.015983\n",
      "val: bce: 0.000680, dice: 0.039890, loss: 0.020285\n",
      "0m 25s\n",
      "Epoch 46/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000296, dice: 0.031060, loss: 0.015678\n",
      "val: bce: 0.000664, dice: 0.039012, loss: 0.019838\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 47/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000290, dice: 0.030482, loss: 0.015386\n",
      "val: bce: 0.000667, dice: 0.038289, loss: 0.019478\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 48/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000285, dice: 0.029947, loss: 0.015116\n",
      "val: bce: 0.000656, dice: 0.038031, loss: 0.019343\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 49/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000281, dice: 0.029420, loss: 0.014850\n",
      "val: bce: 0.000660, dice: 0.037617, loss: 0.019138\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 50/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000275, dice: 0.028887, loss: 0.014581\n",
      "val: bce: 0.000662, dice: 0.037341, loss: 0.019002\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 51/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000270, dice: 0.028359, loss: 0.014314\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "val: bce: 0.000656, dice: 0.036249, loss: 0.018452\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 52/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000264, dice: 0.027797, loss: 0.014030\n",
      "val: bce: 0.000658, dice: 0.036220, loss: 0.018439\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 53/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000259, dice: 0.027258, loss: 0.013758\n",
      "val: bce: 0.000652, dice: 0.035546, loss: 0.018099\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 54/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000253, dice: 0.026694, loss: 0.013473\n",
      "val: bce: 0.000642, dice: 0.034780, loss: 0.017711\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 55/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000248, dice: 0.026193, loss: 0.013220\n",
      "val: bce: 0.000647, dice: 0.034490, loss: 0.017568\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 56/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000243, dice: 0.025665, loss: 0.012954\n",
      "val: bce: 0.000636, dice: 0.034366, loss: 0.017501\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 57/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000238, dice: 0.025149, loss: 0.012694\n",
      "val: bce: 0.000623, dice: 0.033226, loss: 0.016924\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 58/59\n",
      "----------\n",
      "LR 1e-05\n",
      "train: bce: 0.000234, dice: 0.024693, loss: 0.012464\n",
      "val: bce: 0.000638, dice: 0.033115, loss: 0.016876\n",
      "saving best model\n",
      "0m 25s\n",
      "Epoch 59/59\n",
      "----------\n",
      "LR 1.0000000000000002e-06\n",
      "train: bce: 0.000229, dice: 0.024332, loss: 0.012281\n",
      "val: bce: 0.000630, dice: 0.032883, loss: 0.016756\n",
      "saving best model\n",
      "0m 25s\n",
      "Best val loss: 0.016756\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "import torch.optim as optim\n",
    "from torch.optim import lr_scheduler\n",
    "import time\n",
    "import copy\n",
    "\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "print(device)\n",
    "\n",
    "num_class = 6\n",
    "#model = model(num_class).to(device)\n",
    "\n",
    "# freeze backbone layers\n",
    "#for l in model.base_layers:\n",
    "#    for param in l.parameters():\n",
    "#        param.requires_grad = False\n",
    "\n",
    "optimizer_ft = optim.Adam(filter(lambda p: p.requires_grad, model.parameters()), lr=1e-4)\n",
    "\n",
    "exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft, step_size=30, gamma=0.1)\n",
    "\n",
    "model = train_model(model, optimizer_ft, exp_lr_scheduler, num_epochs=60)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79f8b616",
   "metadata": {},
   "source": [
    "# Using the trained model "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "3f473ddc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3, 6, 192, 192)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1200x1200 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import math\n",
    "\n",
    "model.eval()   # Set model to the evaluation mode\n",
    "\n",
    "# Create another simulation dataset for test\n",
    "test_dataset = SimDataset(3, transform = trans)\n",
    "test_loader = DataLoader(test_dataset, batch_size=3, shuffle=False, num_workers=0)\n",
    "\n",
    "# Get the first batch\n",
    "inputs, labels = next(iter(test_loader))\n",
    "inputs = inputs.to(device)\n",
    "labels = labels.to(device)\n",
    "\n",
    "# Predict\n",
    "pred = model(inputs)\n",
    "# The loss functions include the sigmoid function.\n",
    "pred = F.sigmoid(pred)\n",
    "pred = pred.data.cpu().numpy()\n",
    "print(pred.shape)\n",
    "\n",
    "# Change channel-order and make 3 channels for matplot\n",
    "input_images_rgb = [reverse_transform(x) for x in inputs.cpu()]\n",
    "\n",
    "# Map each channel (i.e. class) to each color\n",
    "target_masks_rgb = [helper.masks_to_colorimg(x) for x in labels.cpu().numpy()]\n",
    "pred_rgb = [helper.masks_to_colorimg(x) for x in pred]\n",
    "\n",
    "helper.plot_side_by_side([input_images_rgb, target_masks_rgb, pred_rgb])\n",
    "#Left: Input image, Middle: Correct mask (Ground-truth), Rigth: Predicted mask"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "30b994df",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "pytorch_1121_cu113_unet",
   "language": "python",
   "name": "pytorch_1121_cu113_unet"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.6"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": true
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}