dip-project-phase2.1-noise-classifier.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "execution": {
     "iopub.execute_input": "2024-03-14T21:10:10.119363Z",
     "iopub.status.busy": "2024-03-14T21:10:10.118834Z",
     "iopub.status.idle": "2024-03-14T21:10:23.604017Z",
     "shell.execute_reply": "2024-03-14T21:10:23.603028Z",
     "shell.execute_reply.started": "2024-03-14T21:10:10.119334Z"
    },
    "executionInfo": {
     "elapsed": 87096,
     "status": "ok",
     "timestamp": 1710445802006,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "a4vBD_nTgiBP",
    "outputId": "0bf3ec5e-afec-4104-9b22-54ec104503df"
   },
   "outputs": [],
   "source": [
    "!pip install torchmetrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-14T21:10:23.606615Z",
     "iopub.status.busy": "2024-03-14T21:10:23.606320Z",
     "iopub.status.idle": "2024-03-14T21:10:32.273083Z",
     "shell.execute_reply": "2024-03-14T21:10:32.272055Z",
     "shell.execute_reply.started": "2024-03-14T21:10:23.606589Z"
    },
    "id": "rg6Tu4WbBOUp"
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import torchvision\n",
    "import pandas as pd\n",
    "import cv2\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "execution": {
     "iopub.execute_input": "2024-03-14T21:10:32.277705Z",
     "iopub.status.busy": "2024-03-14T21:10:32.277431Z",
     "iopub.status.idle": "2024-03-14T21:10:32.339724Z",
     "shell.execute_reply": "2024-03-14T21:10:32.338530Z",
     "shell.execute_reply.started": "2024-03-14T21:10:32.277680Z"
    },
    "executionInfo": {
     "elapsed": 6,
     "status": "ok",
     "timestamp": 1710443828028,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "379qGSlgH18m",
    "outputId": "22146d7a-582c-442c-93be-09e12f3b8817"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device(type='cuda')"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-14T21:12:39.416961Z",
     "iopub.status.busy": "2024-03-14T21:12:39.416164Z",
     "iopub.status.idle": "2024-03-14T21:12:39.421792Z",
     "shell.execute_reply": "2024-03-14T21:12:39.420688Z",
     "shell.execute_reply.started": "2024-03-14T21:12:39.416927Z"
    },
    "id": "7RAi9gARGTrZ"
   },
   "outputs": [],
   "source": [
    "in_channels = 3\n",
    "num_classes = 3\n",
    "learning_rate = 1e-3\n",
    "batch_size = 64\n",
    "num_epochs = 5\n",
    "load_model = False\n",
    "save_model = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "execution": {
     "iopub.execute_input": "2024-03-14T21:13:23.133295Z",
     "iopub.status.busy": "2024-03-14T21:13:23.132903Z",
     "iopub.status.idle": "2024-03-14T21:13:46.677722Z",
     "shell.execute_reply": "2024-03-14T21:13:46.676632Z",
     "shell.execute_reply.started": "2024-03-14T21:13:23.133264Z"
    },
    "executionInfo": {
     "elapsed": 13630,
     "status": "ok",
     "timestamp": 1710443841654,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "DAE4e2WX-aCd",
    "outputId": "d7ce3761-1a95-413a-9212-7a18fef1673f"
   },
   "outputs": [],
   "source": [
    "!rm -rf *\n",
    "!pip install -U gdown\n",
    "!gdown 1E1poQDBsRp53Ip9qndB2nPLu1po3AbEp\n",
    "!gdown 18UXIh3mGA_M0oxlapPIrUG-VVpF6PN2S\n",
    "!unzip Phase2.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-14T21:14:23.618020Z",
     "iopub.status.busy": "2024-03-14T21:14:23.617587Z",
     "iopub.status.idle": "2024-03-14T21:14:24.716953Z",
     "shell.execute_reply": "2024-03-14T21:14:24.716140Z",
     "shell.execute_reply.started": "2024-03-14T21:14:23.617983Z"
    },
    "id": "l8kVwGJWDnnv"
   },
   "outputs": [],
   "source": [
    "class NoiseDataset(torch.utils.data.Dataset):\n",
    "    def __init__(self, root_dir='DataSet2', spreadsheet_file='Labels.xlsx', transform=None):\n",
    "        self.label_mapping_dict = {'Gaussian': 0, 'Periodic': 1, 'Salt': 2}\n",
    "        self.annotations = pd.read_excel(spreadsheet_file).drop(columns='Denoise Image')[['Noisy Image', 'Noise Type']]\n",
    "        self.annotations['Noise Type'] = self.annotations['Noise Type'].map(self.label_mapping_dict)\n",
    "        self.root_dir = root_dir\n",
    "        self.transform = transform\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.annotations)\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        img_path = os.path.join(self.root_dir, self.annotations.iloc[index, 0])\n",
    "        image = cv2.imread(filename=img_path, flags=cv2.IMREAD_COLOR)\n",
    "        y_label = torch.tensor(int(self.annotations.iloc[index, 1]))\n",
    "        if self.transform:\n",
    "            image = self.transform(image)\n",
    "        return (image, y_label)\n",
    "\n",
    "dataset = NoiseDataset(transform=torchvision.transforms.ToTensor())\n",
    "train_dataset, test_dataset = torch.utils.data.random_split(dataset, [1000, 250])\n",
    "train_loader = torch.utils.data.DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True)\n",
    "test_loader = torch.utils.data.DataLoader(dataset=test_dataset, batch_size=batch_size, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "execution": {
     "iopub.execute_input": "2024-03-14T21:14:24.718838Z",
     "iopub.status.busy": "2024-03-14T21:14:24.718414Z",
     "iopub.status.idle": "2024-03-14T21:14:27.714351Z",
     "shell.execute_reply": "2024-03-14T21:14:27.713353Z",
     "shell.execute_reply.started": "2024-03-14T21:14:24.718810Z"
    },
    "executionInfo": {
     "elapsed": 9841,
     "status": "ok",
     "timestamp": 1710443852201,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "Rsenz-ktHFs0",
    "outputId": "c34b3db0-fb2a-4cc1-9ef2-9b5c57668143",
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "model = torchvision.models.efficientnet_v2_l(pretrained=True)\n",
    "for param in model.parameters():\n",
    "    param.requires_grad = False\n",
    "model.avgpool = torch.nn.Identity()\n",
    "model.classifier[1] = torch.nn.Linear(in_features=327680, out_features=3, bias=True)\n",
    "model.to(device)\n",
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-14T21:14:45.611790Z",
     "iopub.status.busy": "2024-03-14T21:14:45.610875Z",
     "iopub.status.idle": "2024-03-14T21:14:45.624189Z",
     "shell.execute_reply": "2024-03-14T21:14:45.623159Z",
     "shell.execute_reply.started": "2024-03-14T21:14:45.611758Z"
    },
    "id": "uKBi6H6kGYkx"
   },
   "outputs": [],
   "source": [
    "criterion = torch.nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.Adam(params=model.parameters(), lr=learning_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "execution": {
     "iopub.execute_input": "2024-03-14T21:14:48.399913Z",
     "iopub.status.busy": "2024-03-14T21:14:48.399403Z",
     "iopub.status.idle": "2024-03-14T21:19:06.228372Z",
     "shell.execute_reply": "2024-03-14T21:19:06.227372Z",
     "shell.execute_reply.started": "2024-03-14T21:14:48.399873Z"
    },
    "executionInfo": {
     "elapsed": 266641,
     "status": "ok",
     "timestamp": 1710444118839,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "d4qbohRuJDPA",
    "outputId": "3a1cbb72-cbeb-4d66-9f0d-9c8fd918f98c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss at epoch 0 was 3.722674608230591\n",
      "Loss at epoch 1 was 0.0\n",
      "Loss at epoch 2 was 0.0\n",
      "Loss at epoch 3 was 0.0\n",
      "Loss at epoch 4 was 0.3441014587879181\n"
     ]
    }
   ],
   "source": [
    "for epoch in range(num_epochs):\n",
    "    for batch_idx, (data, targets) in enumerate(train_loader):\n",
    "        data = data.to(device=device)\n",
    "        targets = targets.to(device=device)\n",
    "\n",
    "        scores = model(data)\n",
    "        loss = criterion(scores, targets)\n",
    "\n",
    "        optimizer.zero_grad()\n",
    "        loss.backward()\n",
    "\n",
    "        optimizer.step()\n",
    "    print(f'Loss at epoch {epoch} was {loss}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-14T21:19:06.230927Z",
     "iopub.status.busy": "2024-03-14T21:19:06.230558Z",
     "iopub.status.idle": "2024-03-14T21:19:08.579075Z",
     "shell.execute_reply": "2024-03-14T21:19:08.578146Z",
     "shell.execute_reply.started": "2024-03-14T21:19:06.230894Z"
    },
    "executionInfo": {
     "elapsed": 4,
     "status": "ok",
     "timestamp": 1710449591713,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "xT8PLZeKJT0V"
   },
   "outputs": [],
   "source": [
    "from torchmetrics.functional.classification import multiclass_accuracy\n",
    "from torchmetrics.functional.classification import multiclass_precision\n",
    "from torchmetrics.functional.classification import multiclass_recall\n",
    "from torchmetrics.functional.classification import multiclass_f1_score\n",
    "from torchmetrics.functional.classification import multiclass_specificity\n",
    "from torchmetrics.functional.classification import multiclass_auroc\n",
    "from torchmetrics.functional.classification import multiclass_roc\n",
    "\n",
    "def check_performance(loader, model):\n",
    "    model.eval()\n",
    "    full_y = torch.tensor([], device=device, dtype=torch.int8)\n",
    "    full_predictions = torch.tensor([], device=device)\n",
    "    full_scores = torch.tensor([], device=device)\n",
    "\n",
    "    with torch.no_grad():\n",
    "        for x, y in loader:\n",
    "\n",
    "            x = x.to(device=device)\n",
    "            y = y.to(device=device)\n",
    "            scores = model(x)\n",
    "            predictions = scores.max(1)[1]\n",
    "            full_y = torch.cat(tensors=(full_y, y))\n",
    "            full_predictions = torch.cat(tensors=(full_predictions, predictions))\n",
    "            full_scores = torch.cat(tensors=(full_scores, scores))\n",
    "\n",
    "        accuracy = multiclass_accuracy(full_predictions, full_y, num_classes=3, average='macro')\n",
    "        precision = multiclass_precision(full_predictions, full_y, num_classes=3, average='macro')\n",
    "        recall = multiclass_recall(full_predictions, full_y, num_classes=3, average='macro')\n",
    "        f1_score = multiclass_f1_score(full_predictions, full_y, num_classes=3, average='macro')\n",
    "        specificity = multiclass_specificity(full_predictions, full_y, num_classes=3, average='macro')\n",
    "        auroc = multiclass_auroc(full_scores, full_y, num_classes=3, average='macro')\n",
    "        fpr, tpr, thresholds = multiclass_roc(full_scores, full_y, num_classes=3)\n",
    "    model.train()\n",
    "    return accuracy, precision, recall, f1_score, specificity, auroc, full_y, full_predictions, full_scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "execution": {
     "iopub.execute_input": "2024-03-14T21:29:03.203049Z",
     "iopub.status.busy": "2024-03-14T21:29:03.202309Z",
     "iopub.status.idle": "2024-03-14T21:29:52.269927Z",
     "shell.execute_reply": "2024-03-14T21:29:52.268844Z",
     "shell.execute_reply.started": "2024-03-14T21:29:03.203006Z"
    },
    "executionInfo": {
     "elapsed": 361,
     "status": "ok",
     "timestamp": 1710449016812,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "GYoAP4dsJd3O",
    "outputId": "603019cd-9a35-4b38-a381-736cf079bf6a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1 score: 1.0\n",
      "Specificity: 1.0\n",
      "Area under the ROC: 1.0\n"
     ]
    }
   ],
   "source": [
    "accuracy, precision, recall, f1_score, specificity, auroc, full_y, full_predictions, full_scores = check_performance(train_loader, model)\n",
    "print(f\"Accuracy: {accuracy:.4}\\nPrecision: {precision:.4}\\nRecall: {recall:.4}\\nF1 score: {f1_score:.4}\\nSpecificity: {specificity:.4}\\nArea under the ROC: {auroc:.4}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-14T21:29:52.272203Z",
     "iopub.status.busy": "2024-03-14T21:29:52.271871Z",
     "iopub.status.idle": "2024-03-14T21:29:52.618008Z",
     "shell.execute_reply": "2024-03-14T21:29:52.617112Z",
     "shell.execute_reply.started": "2024-03-14T21:29:52.272178Z"
    },
    "id": "wQXqqiQIv12P"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from torchmetrics.classification import MulticlassROC\n",
    "metric = MulticlassROC(num_classes=3)\n",
    "metric.update(full_scores, full_y)\n",
    "fig_, ax_ = metric.plot(score=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 367
    },
    "execution": {
     "iopub.execute_input": "2024-03-14T21:30:02.947644Z",
     "iopub.status.busy": "2024-03-14T21:30:02.947240Z",
     "iopub.status.idle": "2024-03-14T21:30:03.249245Z",
     "shell.execute_reply": "2024-03-14T21:30:03.247890Z",
     "shell.execute_reply.started": "2024-03-14T21:30:02.947611Z"
    },
    "executionInfo": {
     "elapsed": 866,
     "status": "error",
     "timestamp": 1710450354012,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "IOIiG-NQw_7A",
    "outputId": "44cab0ad-b687-46a9-8727-4c085d08dd0b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from torchmetrics.classification import MulticlassConfusionMatrix\n",
    "metric = MulticlassConfusionMatrix(num_classes=3).to(device)\n",
    "metric.update(full_predictions, full_y)\n",
    "fig_, ax_ = metric.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "execution": {
     "iopub.execute_input": "2024-03-14T21:30:11.432705Z",
     "iopub.status.busy": "2024-03-14T21:30:11.431997Z",
     "iopub.status.idle": "2024-03-14T21:30:23.424885Z",
     "shell.execute_reply": "2024-03-14T21:30:23.423809Z",
     "shell.execute_reply.started": "2024-03-14T21:30:11.432672Z"
    },
    "executionInfo": {
     "elapsed": 12803,
     "status": "ok",
     "timestamp": 1710449637557,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "ztsf2d-1Jf3s",
    "outputId": "502b7023-5a40-48ca-b054-399603a4ee5b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.9959\n",
      "Precision: 0.996\n",
      "Recall: 0.9959\n",
      "F1 score: 0.996\n",
      "Specificity: 0.998\n",
      "Area under the ROC: 1.0\n"
     ]
    }
   ],
   "source": [
    "accuracy, precision, recall, f1_score, specificity, auroc, full_y, full_predictions, full_scores = check_performance(test_loader, model)\n",
    "print(f\"Accuracy: {accuracy:.4}\\nPrecision: {precision:.4}\\nRecall: {recall:.4}\\nF1 score: {f1_score:.4}\\nSpecificity: {specificity:.4}\\nArea under the ROC: {auroc:.4}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 452
    },
    "execution": {
     "iopub.execute_input": "2024-03-14T21:30:23.426897Z",
     "iopub.status.busy": "2024-03-14T21:30:23.426614Z",
     "iopub.status.idle": "2024-03-14T21:30:23.718616Z",
     "shell.execute_reply": "2024-03-14T21:30:23.717664Z",
     "shell.execute_reply.started": "2024-03-14T21:30:23.426872Z"
    },
    "executionInfo": {
     "elapsed": 553,
     "status": "ok",
     "timestamp": 1710449658109,
     "user": {
      "displayName": "Ali Nafisi",
      "userId": "08017426569548477520"
     },
     "user_tz": -210
    },
    "id": "m4XKH_NCu8aI",
    "outputId": "68070427-1e6d-4908-ae17-f40c7b4e12d1"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from torchmetrics.classification import MulticlassROC\n",
    "metric = MulticlassROC(num_classes=3)\n",
    "metric.update(full_scores, full_y)\n",
    "fig_, ax_ = metric.plot(score=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-14T21:30:23.720008Z",
     "iopub.status.busy": "2024-03-14T21:30:23.719733Z",
     "iopub.status.idle": "2024-03-14T21:30:24.032555Z",
     "shell.execute_reply": "2024-03-14T21:30:24.031156Z",
     "shell.execute_reply.started": "2024-03-14T21:30:23.719984Z"
    },
    "id": "SuhFSrm0igMF"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from torchmetrics.classification import MulticlassConfusionMatrix\n",
    "metric = MulticlassConfusionMatrix(num_classes=3).to(device)\n",
    "metric.update(full_predictions, full_y)\n",
    "fig_, ax_ = metric.plot()"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "authorship_tag": "ABX9TyMCk8O4McIuKxcqbd6gt/8d",
   "gpuType": "T4",
   "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": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}