LeNet5.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "feb7a061",
   "metadata": {},
   "source": [
    "# Step 04: Have a look of image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "0cae92f1",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import cv2 as cv\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1f825b2c",
   "metadata": {},
   "outputs": [],
   "source": [
    "label_map_table = {\n",
    "    0: \"normal\",\n",
    "    1: \"void\",\n",
    "    2: \"Horizontal Defect\",\n",
    "    3: \"Vertical Defect\",\n",
    "    4: \"Edge Defect\",\n",
    "    5: \"Partical\"\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "01d10129",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Label] => 3; [Label Actually Means] => Vertical Defect\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "root_train = \"./train_images/\"\n",
    "root_test = \"./test_images/\"\n",
    "train_csv = \"./train.csv\"\n",
    "test_csv = \"./test.csv\"\n",
    "df_train = pd.read_csv(train_csv) # df stands for dataframe\n",
    "\n",
    "id = 2525 # change this number to see other outcome\n",
    "png_img = cv.imread(root_train + df_train.ID[id])\n",
    "label = df_train.Label[id]\n",
    "print(f\"[Label] => {df_train.Label[id]}; [Label Actually Means] => {label_map_table[label]}\")\n",
    "plt.imshow(png_img)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "001894d9",
   "metadata": {},
   "source": [
    "# Step 05: Define Custom Dataset Class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "ae2b5d1a",
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.utils.data.dataset import Dataset\n",
    "from torchvision import transforms\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import cv2 as cv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "3dc081d1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# [Input Args]\n",
    "# 1. target_csv <string>: It's the metadata file describe the name of image and its label.\n",
    "# 2. root_path  <string>: It's the path to the image folder. Combination of this and name is the full path to the image.\n",
    "# 3. height <int>: Use this for elastically resize image to desired shape.\n",
    "# 4. width <int>: Use this for elastically resize image to desired shape.\n",
    "class AOI_Dataset(Dataset):\n",
    "    \n",
    "    # perform logic operation: think what kind of info I need when loading data\n",
    "    def __init__(self, target_csv, root_path, height, width, transform = None):\n",
    "        \n",
    "        # height, width\n",
    "        self.height = height\n",
    "        self.width = width\n",
    "        \n",
    "        # register self\n",
    "        self.target_csv = target_csv\n",
    "        self.root_path = root_path\n",
    "        \n",
    "        # 1. Read CSV file through root_path\n",
    "        self.df = pd.read_csv(self.target_csv)\n",
    "        \n",
    "        # 2. Remember the length\n",
    "        self.count = len(self.df)\n",
    "        \n",
    "        # 3. transform\n",
    "        self.transforms = transform\n",
    "    \n",
    "    # input: index\n",
    "    # output: pair of (image, lable)\n",
    "    def __getitem__(self, index):\n",
    "        # Read images\n",
    "        img = cv.imread(self.root_path + self.df.ID[index])\n",
    "        \n",
    "        # Use resize to a smaller shape when training takes so long.\n",
    "        img_resize = cv.resize(img, (self.height, self.width))\n",
    "\n",
    "        # To Tensor\n",
    "        img_tensor = self.transforms(np.uint8(img_resize))\n",
    "        \n",
    "        # Get label\n",
    "        label = self.df.Label[index]\n",
    "        \n",
    "        return (img_tensor, label)\n",
    "    \n",
    "    def __len__(self):\n",
    "        return self.count"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "82b2316c",
   "metadata": {},
   "source": [
    "# Step 06: Define DataLoader w.r.t Dataset Class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "18df19b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from torchvision import transforms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "261f9432",
   "metadata": {},
   "outputs": [],
   "source": [
    "height = 512\n",
    "width = 512\n",
    "\n",
    "transform_train = transforms.Compose([\n",
    "    transforms.ToTensor(),\n",
    "    # add mroe in the future\n",
    "])\n",
    "\n",
    "Train_Dataset = AOI_Dataset(target_csv = train_csv, root_path = root_train, height = height, width = width, transform = transform_train)\n",
    "\n",
    "batch_size = 8\n",
    "\n",
    "Train_DataLoader = torch.utils.data.DataLoader(Train_Dataset, batch_size = batch_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a2f599c",
   "metadata": {},
   "source": [
    "# Step 07: Define LeNet5 - like Structure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "36c41b72",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cpu\n",
      "LeNet(\n",
      "  (conv1): Conv2d(3, 6, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "  (conv2): Conv2d(6, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "  (conv3): Conv2d(16, 50, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "  (pool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  (fc1): Linear(in_features=204800, out_features=120, bias=True)\n",
      "  (fc2): Linear(in_features=120, out_features=84, bias=True)\n",
      "  (fc3): Linear(in_features=84, out_features=6, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "\n",
    "# Use GPU\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "print(device)\n",
    "\n",
    "class LeNet(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.conv1 = nn.Conv2d( 3,  6, 3, padding = 1) # 加深 channel same\n",
    "        self.conv2 = nn.Conv2d( 6, 16, 3, padding = 1) # 加深 channel\n",
    "        self.conv3 = nn.Conv2d(16, 50, 3, padding = 1) # 加深 channel\n",
    "        self.pool = nn.MaxPool2d(2, 2)\n",
    "        self.fc1 = nn.Linear(50 * 64 * 64, 120) # Why 64 Why 50\n",
    "        self.fc2 = nn.Linear(120, 84)\n",
    "        self.fc3 = nn.Linear(84, 6)\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = self.pool(F.relu(self.conv1(x))) # 512 x 512 x  3 -> 256 x 256 x  6\n",
    "        x = self.pool(F.relu(self.conv2(x))) # 256 x 256 x  6 -> 128 x 128 x 16\n",
    "        x = self.pool(F.relu(self.conv3(x))) # 128 x 128 x 16 ->  64 x  64 x 50\n",
    "        x = x.view(-1, 50 * 64 * 64)\n",
    "        x = F.relu(self.fc1(x))\n",
    "        x = F.relu(self.fc2(x))\n",
    "        x = self.fc3(x)\n",
    "        #x = F.softmax(x)\n",
    "        return x\n",
    "    \n",
    "lenet = LeNet()\n",
    "print(lenet.to(device))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "331cc0e1",
   "metadata": {},
   "source": [
    "# Step 08: Define Loss Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "fc3035b5",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.optim as optim\n",
    "\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = optim.RMSprop(lenet.parameters(), lr=1e-4) # learning rate is also adjustable\n",
    "epoch = 10 # change this to whatever number you'd like"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a17b0c45",
   "metadata": {},
   "source": [
    "# Step 09: Training Phase"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e9d2c967",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1,    10] loss: 1.750\n",
      "[1,    20] loss: 1.684\n",
      "[1,    30] loss: 1.725\n",
      "[1,    40] loss: 1.726\n",
      "[1,    50] loss: 1.684\n",
      "[1,    60] loss: 1.694\n",
      "[1,    70] loss: 1.730\n",
      "[1,    80] loss: 1.614\n",
      "[1,    90] loss: 1.629\n",
      "[1,   100] loss: 1.558\n",
      "[1,   110] loss: 1.551\n",
      "[1,   120] loss: 1.513\n",
      "[1,   130] loss: 1.681\n",
      "[1,   140] loss: 1.592\n",
      "[1,   150] loss: 1.534\n",
      "[1,   160] loss: 1.450\n",
      "[1,   170] loss: 1.545\n",
      "[1,   180] loss: 1.432\n",
      "[1,   190] loss: 1.563\n",
      "[1,   200] loss: 1.463\n",
      "[1,   210] loss: 1.246\n",
      "[1,   220] loss: 1.509\n",
      "[1,   230] loss: 1.369\n",
      "[1,   240] loss: 1.396\n",
      "[1,   250] loss: 1.270\n",
      "[1,   260] loss: 1.291\n",
      "[1,   270] loss: 1.395\n",
      "[1,   280] loss: 1.329\n",
      "[1,   290] loss: 1.374\n",
      "[1,   300] loss: 1.393\n",
      "[1,   310] loss: 1.303\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 316/316 [01:31<00:00,  3.46it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of the network on the train images: 51 %\n",
      "[2,    10] loss: 1.146\n",
      "[2,    20] loss: 1.292\n",
      "[2,    30] loss: 1.331\n",
      "[2,    40] loss: 1.330\n",
      "[2,    50] loss: 1.233\n",
      "[2,    60] loss: 1.315\n",
      "[2,    70] loss: 1.306\n",
      "[2,    80] loss: 1.112\n",
      "[2,    90] loss: 1.231\n",
      "[2,   100] loss: 1.085\n",
      "[2,   110] loss: 1.113\n",
      "[2,   120] loss: 1.176\n",
      "[2,   130] loss: 1.382\n",
      "[2,   140] loss: 1.222\n",
      "[2,   150] loss: 1.203\n",
      "[2,   160] loss: 1.043\n",
      "[2,   170] loss: 1.143\n",
      "[2,   180] loss: 1.110\n",
      "[2,   190] loss: 1.197\n",
      "[2,   200] loss: 1.101\n",
      "[2,   210] loss: 0.915\n",
      "[2,   220] loss: 1.193\n",
      "[2,   230] loss: 1.096\n",
      "[2,   240] loss: 1.036\n",
      "[2,   250] loss: 0.949\n",
      "[2,   260] loss: 0.989\n",
      "[2,   270] loss: 1.113\n",
      "[2,   280] loss: 1.036\n",
      "[2,   290] loss: 1.110\n",
      "[2,   300] loss: 1.116\n",
      "[2,   310] loss: 0.983\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 316/316 [01:34<00:00,  3.36it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of the network on the train images: 63 %\n",
      "[3,    10] loss: 0.907\n",
      "[3,    20] loss: 1.060\n",
      "[3,    30] loss: 1.055\n",
      "[3,    40] loss: 1.007\n",
      "[3,    50] loss: 1.049\n",
      "[3,    60] loss: 1.003\n",
      "[3,    70] loss: 0.993\n",
      "[3,    80] loss: 0.828\n",
      "[3,    90] loss: 1.088\n",
      "[3,   100] loss: 0.886\n",
      "[3,   110] loss: 0.875\n",
      "[3,   120] loss: 1.049\n",
      "[3,   130] loss: 1.149\n",
      "[3,   140] loss: 1.053\n",
      "[3,   150] loss: 0.950\n",
      "[3,   160] loss: 0.754\n",
      "[3,   170] loss: 0.934\n",
      "[3,   180] loss: 0.929\n",
      "[3,   190] loss: 0.947\n",
      "[3,   200] loss: 0.874\n",
      "[3,   210] loss: 0.718\n",
      "[3,   220] loss: 1.002\n",
      "[3,   230] loss: 0.876\n",
      "[3,   240] loss: 0.870\n",
      "[3,   250] loss: 0.731\n",
      "[3,   260] loss: 0.820\n",
      "[3,   270] loss: 0.924\n",
      "[3,   280] loss: 0.867\n",
      "[3,   290] loss: 0.900\n",
      "[3,   300] loss: 0.903\n",
      "[3,   310] loss: 0.778\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 316/316 [01:39<00:00,  3.17it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of the network on the train images: 70 %\n",
      "[4,    10] loss: 0.726\n",
      "[4,    20] loss: 0.915\n",
      "[4,    30] loss: 0.844\n",
      "[4,    40] loss: 0.783\n",
      "[4,    50] loss: 0.955\n",
      "[4,    60] loss: 0.786\n",
      "[4,    70] loss: 0.821\n",
      "[4,    80] loss: 0.670\n",
      "[4,    90] loss: 0.957\n",
      "[4,   100] loss: 0.742\n",
      "[4,   110] loss: 0.713\n",
      "[4,   120] loss: 0.925\n",
      "[4,   130] loss: 0.975\n",
      "[4,   140] loss: 0.887\n",
      "[4,   150] loss: 0.779\n",
      "[4,   160] loss: 0.639\n",
      "[4,   170] loss: 0.833\n",
      "[4,   180] loss: 0.782\n",
      "[4,   190] loss: 0.775\n",
      "[4,   200] loss: 0.698\n"
     ]
    }
   ],
   "source": [
    "from tqdm import tqdm\n",
    "import time\n",
    "\n",
    "tic = time.time()\n",
    "train_acc_list = []\n",
    "val_acc_list = []\n",
    "loss_list = []\n",
    "print_probe_num = 10\n",
    "\n",
    "for epoch in range(epoch):  # loop over the dataset multiple times\n",
    "\n",
    "    running_loss = 0.0\n",
    "    \n",
    "    for i, data in enumerate(Train_DataLoader, 0):\n",
    "        # Select input and output pair\n",
    "        inputs, labels = data[0].to(device), data[1].to(device)\n",
    "        \n",
    "        # Clear gradient\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        # Forward Propagation\n",
    "        outputs = lenet(inputs.float())\n",
    "        \n",
    "        # Compute Loss\n",
    "        loss = criterion(outputs, labels)\n",
    "        \n",
    "        # Backward Propagation\n",
    "        loss.backward()\n",
    "        \n",
    "        # Update Weight\n",
    "        optimizer.step()\n",
    "\n",
    "        # Just want to calculate the running loss 移動平均 loss!!\n",
    "        running_loss += loss.item()\n",
    "        if i % print_probe_num == (print_probe_num - 1):\n",
    "            print('[%d, %5d] loss: %.3f' % (epoch + 1, i + 1, running_loss / print_probe_num))\n",
    "            loss_list.append(running_loss / print_probe_num)\n",
    "            running_loss = 0.0\n",
    "            \n",
    "    correct = 0\n",
    "    total = 0\n",
    "    \n",
    "    # Train\n",
    "    with torch.no_grad(): # since we're not training, we don't need to calculate the gradients for our outputs\n",
    "        for datum in tqdm(Train_DataLoader):\n",
    "\n",
    "            imgs, labs = datum[0].to(device), datum[1].to(device)\n",
    "            # calculate outputs by running images through the network \n",
    "            outputs = lenet(imgs.float())\n",
    "            # the class with the highest energy is what we choose as prediction\n",
    "            _, preds = torch.max(outputs.data, 1)\n",
    "            # [0.1 0.1 0.1 .... ]\n",
    "            total += labs.size(0)\n",
    "            correct += (preds == labs).sum().item()\n",
    "        train_acc_list.append(float(correct)/float(total))\n",
    "        print('Accuracy of the network on the train images: %d %%' % (100 * correct / total))\n",
    "\n",
    "        \n",
    "toc = time.time()\n",
    "print(f\"Spend {round(toc - tic, 2)} (sec)\")\n",
    "print('Finished Training')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6ee2e9c3",
   "metadata": {},
   "source": [
    "# Step 10: Plot Accuracy & Loss Curves"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cc7680bb",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "## Accuracy\n",
    "\n",
    "plt.figure(figsize = (20, 10))\n",
    "plt.title(\"LeNet5: Accuracy Curve\", fontsize = 24)\n",
    "plt.xlabel(\"Epochs\"    , fontsize = 20)\n",
    "plt.ylabel(\"Accuracy %\", fontsize = 20)\n",
    "plt.plot(train_acc_list, label = \"train acc.\")\n",
    "plt.legend(loc = 2, fontsize = 20)\n",
    "plt.show()\n",
    "\n",
    "## Loss\n",
    "\n",
    "plt.figure(figsize=(20, 10))\n",
    "plt.title(\"LeNet5: Loss Curve\", fontsize = 24)\n",
    "plt.plot(loss_list)\n",
    "plt.xlabel(\"Probes\", fontsize = 20)\n",
    "plt.ylabel(\"Loss\", fontsize = 20)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ff26ad63",
   "metadata": {},
   "source": [
    "# Step 11: Testing Phase"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "697f1195",
   "metadata": {},
   "outputs": [],
   "source": [
    "transform_test = transforms.Compose([\n",
    "    transforms.ToTensor(),\n",
    "])\n",
    "\n",
    "Test_Dataset = AOI_Dataset(target_csv = test_csv, root_path = root_test, height = height, width = width, transform = transform_test)\n",
    "Test_DataLoader = torch.utils.data.DataLoader(dataset = Test_Dataset, batch_size = 1, shuffle = False)\n",
    "Name_of_csv_file = \"AI.FREE.SUCCESS.csv\"\n",
    "\n",
    "df_test = pd.read_csv(test_csv)\n",
    "df_test_np = df_test.to_numpy()\n",
    "\n",
    "count = -1\n",
    "with torch.no_grad(): # since we're not training, we don't need to calculate the gradients for our outputs\n",
    "    for datum in tqdm(Test_DataLoader):\n",
    "        count = count + 1\n",
    "        imgs = datum[0].to(device)\n",
    "        # calculate outputs by running images through the network \n",
    "        outputs = lenet(imgs.float())\n",
    "        # the class with the highest energy is what we choose as prediction\n",
    "        _, preds = torch.max(outputs.data, 1)\n",
    "        df_test_np[count][1] = float(preds)\n",
    "        \n",
    "df = pd.DataFrame(df_test_np, columns = ['ID','Label'])\n",
    "df.to_csv(Name_of_csv_file, index=False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}