Car Racing/notebooks/Data_Preprocessing.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# imports\n",
    "import os\n",
    "from PIL import Image\n",
    "from torchvision import transforms\n",
    "import torch\n",
    "import cv2\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "import shutil\n",
    "import random"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Balancing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Numero minimo di immagini per cartella: 369\n",
      "Classe 0: 1000 immagini disponibili.\n",
      "Copiate 369 immagini in ./train_balanced\\0.\n",
      "Classe 1: 1500 immagini disponibili.\n",
      "Copiate 369 immagini in ./train_balanced\\1.\n",
      "Classe 2: 1500 immagini disponibili.\n",
      "Copiate 369 immagini in ./train_balanced\\2.\n",
      "Classe 3: 2000 immagini disponibili.\n",
      "Copiate 369 immagini in ./train_balanced\\3.\n",
      "Classe 4: 369 immagini disponibili.\n",
      "Copiate 369 immagini in ./train_balanced\\4.\n"
     ]
    }
   ],
   "source": [
    "def balance_datasets(src_dirs, dest_dir):\n",
    "    # Crea la cartella di destinazione se non esiste\n",
    "    if not os.path.exists(dest_dir):\n",
    "        os.makedirs(dest_dir)\n",
    "    \n",
    "    # Trova il numero minimo di immagini tra tutte le cartelle\n",
    "    min_images = min([len(os.listdir(src)) for src in src_dirs if os.path.isdir(src)])\n",
    "    print(f\"Numero minimo di immagini per cartella: {min_images}\")\n",
    "    \n",
    "    # Per ogni cartella di origine, copiamo un numero minimo di immagini\n",
    "    for src_dir in src_dirs:\n",
    "        # Verifica che la cartella sorgente esista ed è una directory\n",
    "        if not os.path.isdir(src_dir):\n",
    "            print(f\"Attenzione: la cartella {src_dir} non è valida.\")\n",
    "            continue\n",
    "\n",
    "        # Crea la cartella di destinazione per ciascuna classe\n",
    "        class_name = os.path.basename(src_dir.rstrip('/'))\n",
    "        dest_class_dir = os.path.join(dest_dir, class_name)\n",
    "        if not os.path.exists(dest_class_dir):\n",
    "            os.makedirs(dest_class_dir)\n",
    "        \n",
    "        # Ottieni la lista di tutte le immagini nella cartella\n",
    "        images = [img for img in os.listdir(src_dir) if os.path.isfile(os.path.join(src_dir, img))]\n",
    "        print(f\"Classe {class_name}: {len(images)} immagini disponibili.\")\n",
    "        \n",
    "        # Verifica se ci sono abbastanza immagini\n",
    "        if len(images) < min_images:\n",
    "            print(f\"Classe {class_name} ha meno immagini di {min_images}. Procedo comunque con {len(images)} immagini.\")\n",
    "            selected_images = images  # Usa tutte le immagini disponibili\n",
    "        else:\n",
    "            selected_images = random.sample(images, min_images)  # Seleziona un numero di immagini pari al minimo\n",
    "        \n",
    "        # Copia le immagini selezionate nella cartella di destinazione\n",
    "        for img in selected_images:\n",
    "            src_img_path = os.path.join(src_dir, img)\n",
    "            dest_img_path = os.path.join(dest_class_dir, img)\n",
    "            shutil.copy(src_img_path, dest_img_path)\n",
    "        print(f\"Copiate {len(selected_images)} immagini in {dest_class_dir}.\")\n",
    "\n",
    "# Esempio di uso\n",
    "src_dirs = [\"./train/0\", \"./train/1\", \"./train/2\", \"./train/3\", \"./train/4\"]\n",
    "dest_dir = \"./train_balanced\"\n",
    "balance_datasets(src_dirs, dest_dir)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[369, 369, 369, 369, 369]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[len(os.listdir(os.path.join('train_balanced/0'))),\n",
    " len(os.listdir(os.path.join('train_balanced/1'))),\n",
    " len(os.listdir(os.path.join('train_balanced/2'))),\n",
    " len(os.listdir(os.path.join('train_balanced/3'))),\n",
    " len(os.listdir(os.path.join('train_balanced/4')))]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Augmentation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Horizontal Flip"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Implementation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_img_path = os.path.join('./temp.png') # img taken from steer-left class\n",
    "img = cv2.imread(test_img_path)\n",
    "\n",
    "plt.imshow(img)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We want to flip the image in horizontal, but not the panel. And since flipping the image leads to a different class (right to left and viceversa) the green sign need to be adjusted.\n",
    "\n",
    "So, now we crop the green sign and flip on the opposite part.\n",
    "\n",
    "Position of the green sign:\n",
    "48 pixels : the center\n",
    "18 pixels : lenght for steer "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def crop_panel(image):\n",
    "    img = image.copy()\n",
    "    # Ritaglia la parte superiore (macchina) e inferiore (pannello)\n",
    "    \n",
    "    panel_crop = img[84:96, 0:96]  # Prendi i pixel da 84 a 96 lungo l'altezza (H)\n",
    "\n",
    "    return panel_crop"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "panel = crop_panel(img)\n",
    "\n",
    "plt.imshow(panel)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def adjust_left_signal(panel):\n",
    "\n",
    "    updated_panel = panel.copy()\n",
    "\n",
    "    img_height, img_width, _ = panel.shape\n",
    "    center = img_width // 2\n",
    "\n",
    "    sign_width = 18\n",
    "    sign_top_left = [center - sign_width, img_height - 10]\n",
    "    sign_bot_right = [center, img_height - 4]\n",
    "\n",
    "\n",
    "    # Crop the green sign\n",
    "    green_sign = panel[sign_top_left[1]:sign_bot_right[1], sign_top_left[0]:sign_bot_right[0]]\n",
    "\n",
    "    flipped_green_sign = green_sign[:, ::-1]\n",
    "    \n",
    "    # Calcola la nuova posizione per traslare di 18 pixel verso destra\n",
    "    new_sign_top_left = [sign_top_left[0] + sign_width, sign_top_left[1]]\n",
    "    new_sign_bot_right = [sign_bot_right[0] + sign_width, sign_bot_right[1]]\n",
    "\n",
    "    # Rimpiazza la porzione originale con il segnale flippato\n",
    "    updated_panel[sign_top_left[1]:sign_bot_right[1], sign_top_left[0]:sign_bot_right[0]] = 0\n",
    "    updated_panel[new_sign_top_left[1]:new_sign_bot_right[1], new_sign_top_left[0]:new_sign_bot_right[0]] = flipped_green_sign\n",
    "    \n",
    "    return updated_panel"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def adjust_right_signal(panel):\n",
    "\n",
    "    updated_panel = panel.copy()\n",
    "\n",
    "    img_height, img_width, _ = panel.shape\n",
    "    center = img_width // 2\n",
    "\n",
    "    sign_width = 18\n",
    "    sign_top_left = [center, img_height - 10]\n",
    "    sign_bot_right = [center + sign_width, img_height - 4]\n",
    "\n",
    "\n",
    "    # Crop the green sign\n",
    "    green_sign = panel[sign_top_left[1]:sign_bot_right[1], sign_top_left[0]:sign_bot_right[0]]\n",
    "\n",
    "    flipped_green_sign = green_sign[:, ::-1]\n",
    "    \n",
    "    # Calcola la nuova posizione per traslare di 18 pixel verso destra\n",
    "    new_sign_top_left = [sign_top_left[0] - sign_width, sign_top_left[1]]\n",
    "    new_sign_bot_right = [sign_bot_right[0] - sign_width, sign_bot_right[1]]\n",
    "\n",
    "    # Rimpiazza la porzione originale con il segnale flippato\n",
    "    updated_panel[sign_top_left[1]:sign_bot_right[1], sign_top_left[0]:sign_bot_right[0]] = 0\n",
    "    updated_panel[new_sign_top_left[1]:new_sign_bot_right[1], new_sign_top_left[0]:new_sign_bot_right[0]] = flipped_green_sign\n",
    "    \n",
    "    return updated_panel\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAh8AAABsCAYAAADKdNCXAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjAsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvlHJYcgAAAAlwSFlzAAAPYQAAD2EBqD+naQAADhpJREFUeJzt3X9sU/XXwPGzH7RjAq0MaTfZYBrilF8CY2Ng4h9bQhSjqA/RBM3UBIMOZZKogBmSII64mCBIJJCAf4gskog/0ECw4HxIxgYzQ4c6QJYwgRaJWVt+DZ/2PH+Ypw/l3i8UWe/t1vcrOYmc+7m9JxzDTu7upzdDVVUAAAAskml3AQAAIL0wfAAAAEsxfAAAAEsxfAAAAEsxfAAAAEsxfAAAAEsxfAAAAEsxfAAAAEsxfAAAAEsxfAAAAEtlJ+uD169fLw0NDeL3+2XSpEmybt06KSsru+F50WhUTp8+LUOHDpWMjIxklQcAAPqQqko4HJaCggLJzLzBvQ1NgsbGRnU4HLp582Y9cuSIzp8/X91utwYCgRue293drSJCEARBEEQ/jO7u7hv+rM9Q7fsXy5WXl8u0adPkww8/FJF/7mYUFhbKK6+8IkuWLLnuucFgUNxud1+XBAAALNDT0yMul+u6a/r8mY8rV65IW1ubVFVV/f9FMjOlqqpKmpubDet7e3slFArFIhwO93VJAADAIok8MtHnw8e5c+ckEomIx+OJy3s8HvH7/Yb19fX14nK5YlFYWNjXJQEAgBRi+26XpUuXSjAYjEV3d7fdJQEAgCTq890uI0aMkKysLAkEAnH5QCAgXq/XsN7pdIrT6ezrMgAAQIrq8zsfDodDpk6dKj6fL5aLRqPi8/mkoqKiry8HAAD6maR8z8fixYulurpaSktLpaysTNasWSMXLlyQ559/PhmXAwAA/UhSho+nnnpK/vzzT1m+fLn4/X65//77ZdeuXYaHUAEAQPpJyvd83IpQKHTD/cEA+rENFl3nrEluuUXX7rfyTXJv3+JnvmGSC93iZyKVBYNBGTZs2HXX2L7bBQAApBeGDwAAYCmGDwAAYCmGDwAAYCmGDwAAYKl+s9slP9/4FHZubq4h19XVZfq5ZWVlhtyBAwcMubFjxxpyly9fNuT++OMPQ2769Omm125tbTXkIpGI6VpgwLPqf/3jJrl7LLp2v3WvSa7jFj+zwCQXMMlhoGC3CwAASDkMHwAAwFIMHwAAwFIMHwAAwFJJebdLMqxYscKQM3vg9Oq36V5t8uTJhtwdd9xhyM2ePduQczqdhty3335ryN1zj/nTbJcuXTLk2tvbTdcCADDQcecDAABYiuEDAABYiuEDAABYiuEDAABYqt88cHrhwgVD7uzZs4bcxIkTTc/fuHGjIVdZWWnINTQ0GHLLli0z5MaNG5fwtX/44QfTPAAA6Yg7HwAAwFIMHwAAwFIMHwAAwFIMHwAAwFL95oHTQYMGGXJmr+w9deqU6flmD4iGw2FDrry83JCLRIzvAA8Gg4bcpk2bTK9dUlJiyPEQKgAgXXHnAwAAWIrhAwAAWIrhAwAAWIrhAwAAWIrhAwAAWKrf7Hb55ptvDDm3223Ifffdd6bn19TUGHLvv/++ITd37lxDbteuXYZcU1OTIbdw4ULTa69Zs8Y0f60pU6YktO7MmTMJ5W6VcY+OSK5Jzrjv5x+/J3idKTI4oXVn5G+T3P8keBWkjP+y6DrnLbrOgPKHSe5WG9Zzi+djIOLOBwAAsBTDBwAAsBTDBwAAsBTDBwAAsFSGqqrdRVwtFAqJy+WyuwxbmH2Nu5kVK1YYcitXrjRdO8kktz/BenJMcmbTakTyTM/vlVEJXSc3wRl4hRgfql0p/oTOBQBYIxgMmr7+5Grc+QAAAJZi+AAAAJZi+AAAAJZi+AAAAJbqN99wmg52796d0Lrff0/0u0PNv+TxvxM+O1HGbx79z1f/936X3j79PACAPbjzAQAALMXwAQAALMXwAQAALMXwAQAArKU3qampSR955BHNz89XEdEdO3bEHY9Go1pXV6der1dzcnK0srJSjx49mvDnB4NBFRGCIAiCIPphBIPBG/6sv+k7HxcuXJBJkybJ+vXrTY+/9957snbtWtmwYYO0tLTIbbfdJrNmzZLLly/f7KUAAMBAdLN3Pq4mEn/nIxqNqtfr1YaGhliup6dHnU6nbtu2LaHP5M4HQRAEQfTfSMqdj+vp6uoSv98vVVVVsZzL5ZLy8nJpbm42Pae3t1dCoVBcAACAgatPhw+//583jHo8nri8x+OJHbtWfX29uFyuWBQWFvZlSQAAIMXYvttl6dKlEgwGY9Hd3W13SQAAIIn6dPjwer0iIhIIBOLygUAgduxaTqdThg0bFhcAAGDg6tPho7i4WLxer/h8vlguFApJS0uLVFRU9OWlAABAP3XTL5Y7f/68HD9+PPbnrq4uaW9vl+HDh0tRUZHU1tbKO++8I2PHjpXi4mKpq6uTgoICmTNnTl/WDQAA+qub3V67b98+06011dXVse22dXV16vF41Ol0amVlpXZ2dib8+Wy1JQiCIIj+G4lstc1QVZUUEgwGxe12210GAAD4F3p6esTlcl13je27Xa4VDoftLgEAAPxLifwcT7k7H9FoVE6fPi1Dhw6VcDgshYWF0t3dzS6YFBMKhehNCqM/qYvepC56c2tUVcLhsBQUFEhm5vXvbdz0A6fJlpmZKaNGjRIRkYyMDBERtuCmMHqT2uhP6qI3qYve/Hs3+nXL/0m5X7sAAICBjeEDAABYKqWHD6fTKW+//bY4nU67S8E16E1qoz+pi96kLnpjnZR74BQAAAxsKX3nAwAADDwMHwAAwFIMHwAAwFIMHwAAwFIMHwAAwFIpO3ysX79exowZIzk5OVJeXi6tra12l5SW6uvrZdq0aTJ06FAZOXKkzJkzRzo7O+PWXL58WWpqaiQvL0+GDBkiTz75pAQCAZsqTl+rV6+WjIwMqa2tjeXojX1OnTolzzzzjOTl5cngwYNlwoQJcujQodhxVZXly5dLfn6+DB48WKqqquTYsWM2VpweIpGI1NXVSXFxsQwePFjuvvtuWblypVy98ZPeWCDhd91bqLGxUR0Oh27evFmPHDmi8+fPV7fbrYFAwO7S0s6sWbN0y5Yt2tHRoe3t7frwww9rUVGRnj9/PrZmwYIFWlhYqD6fTw8dOqTTp0/XGTNm2Fh1+mltbdUxY8boxIkTddGiRbE8vbHHX3/9paNHj9bnnntOW1pa9MSJE7p79249fvx4bM3q1avV5XLpF198oYcPH9ZHH31Ui4uL9dKlSzZWPvCtWrVK8/LydOfOndrV1aXbt2/XIUOG6AcffBBbQ2+SLyWHj7KyMq2pqYn9ORKJaEFBgdbX19tYFVRVz549qyKiTU1Nqqra09OjgwYN0u3bt8fW/Prrryoi2tzcbFeZaSUcDuvYsWN1z549+uCDD8aGD3pjnzfffFMfeOCB/3g8Go2q1+vVhoaGWK6np0edTqdu27bNihLT1uzZs/WFF16Iyz3xxBM6b948VaU3Vkm5X7tcuXJF2trapKqqKpbLzMyUqqoqaW5utrEyiIgEg0ERERk+fLiIiLS1tcnff/8d16+SkhIpKiqiXxapqamR2bNnx/VAhN7Y6auvvpLS0lKZO3eujBw5UiZPniybNm2KHe/q6hK/3x/XG5fLJeXl5fQmyWbMmCE+n0+OHj0qIiKHDx+W/fv3y0MPPSQi9MYqKfdW23PnzkkkEhGPxxOX93g88ttvv9lUFUREotGo1NbWysyZM2X8+PEiIuL3+8XhcIjb7Y5b6/F4xO/321BlemlsbJQff/xRDh48aDhGb+xz4sQJ+eijj2Tx4sWybNkyOXjwoLz66qvicDikuro69vdv9u8cvUmuJUuWSCgUkpKSEsnKypJIJCKrVq2SefPmiYjQG4uk3PCB1FVTUyMdHR2yf/9+u0uBiHR3d8uiRYtkz549kpOTY3c5uEo0GpXS0lJ59913RURk8uTJ0tHRIRs2bJDq6mqbq0tvn332mWzdulU+/fRTGTdunLS3t0ttba0UFBTQGwul3K9dRowYIVlZWYYn8gOBgHi9XpuqwsKFC2Xnzp2yb98+GTVqVCzv9XrlypUr0tPTE7eefiVfW1ubnD17VqZMmSLZ2dmSnZ0tTU1NsnbtWsnOzhaPx0NvbJKfny/33XdfXO7ee++VkydPiojE/v75d856r7/+uixZskSefvppmTBhgjz77LPy2muvSX19vYjQG6uk3PDhcDhk6tSp4vP5YrloNCo+n08qKipsrCw9qaosXLhQduzYIXv37pXi4uK441OnTpVBgwbF9auzs1NOnjxJv5KssrJSfv75Z2lvb49FaWmpzJs3L/bf9MYeM2fONGxJP3r0qIwePVpERIqLi8Xr9cb1JhQKSUtLC71JsosXL0pmZvyPvqysLIlGoyJCbyxj9xOvZhobG9XpdOrHH3+sv/zyi7744ovqdrvV7/fbXVraeemll9Tlcun333+vZ86cicXFixdjaxYsWKBFRUW6d+9ePXTokFZUVGhFRYWNVaevq3e7qNIbu7S2tmp2drauWrVKjx07plu3btXc3Fz95JNPYmtWr16tbrdbv/zyS/3pp5/0scceYzunBaqrq/XOO++MbbX9/PPPdcSIEfrGG2/E1tCb5EvJ4UNVdd26dVpUVKQOh0PLysr0wIEDdpeUlkTENLZs2RJbc+nSJX355Zf19ttv19zcXH388cf1zJkz9hWdxq4dPuiNfb7++msdP368Op1OLSkp0Y0bN8Ydj0ajWldXpx6PR51Op1ZWVmpnZ6dN1aaPUCikixYt0qKiIs3JydG77rpL33rrLe3t7Y2toTfJl6F61de6AQAAJFnKPfMBAAAGNoYPAABgKYYPAABgKYYPAABgKYYPAABgKYYPAABgKYYPAABgKYYPAABgKYYPAABgKYYPAABgKYYPAABgqf8FRRggF3nysekAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(adjust_left_signal(panel=panel))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(crop_panel(img))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let us now flip the entire image and adjust the panel"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def flip_right(image):\n",
    "    img = image.copy()\n",
    "\n",
    "    panel = crop_panel(img)\n",
    "    adjusted_panel = adjust_left_signal(panel)\n",
    "\n",
    "    # Flippa l'intera immagine\n",
    "    flipped_img = img[:, ::-1]\n",
    "\n",
    "    # Inserisci il pannello corretto nella nuova immagine\n",
    "    flipped_img[84:96, 0:96] = adjusted_panel\n",
    "\n",
    "    return flipped_img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def flip_left(image):\n",
    "    img = image.copy()\n",
    "\n",
    "    panel = crop_panel(img)\n",
    "    adjusted_panel = adjust_right_signal(panel)\n",
    "\n",
    "    # Flippa l'intera immagine\n",
    "    flipped_img = img[:, ::-1]\n",
    "\n",
    "    # Inserisci il pannello corretto nella nuova immagine\n",
    "    flipped_img[84:96, 0:96] = adjusted_panel\n",
    "\n",
    "    return flipped_img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# original\n",
    "plt.imshow(img)\n",
    "plt.show()\n",
    "\n",
    "#flipped\n",
    "plt.imshow(flip_right(img))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Dataset creation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from PIL import Image\n",
    "\n",
    "def create_flipped_steer_dataset(input_dir, output_dir, side):\n",
    "    \"\"\"Crea un nuovo dataset con immagini flippate e pannelli regolati.\"\"\"\n",
    "    if not os.path.exists(output_dir):\n",
    "        os.makedirs(output_dir)\n",
    "\n",
    "    for filename in os.listdir(input_dir):\n",
    "        if not filename.endswith((\".jpg\", \".png\")):\n",
    "            continue\n",
    "\n",
    "        # Carica l'immagine\n",
    "        img_path = os.path.join(input_dir, filename)\n",
    "        img = np.array(Image.open(img_path))\n",
    "\n",
    "        flipped_img = None\n",
    "        if side == 'LEFT':\n",
    "            flipped_img = flip_left(img)\n",
    "        elif side == 'RIGHT':\n",
    "            flipped_img = flip_right(img)\n",
    "        else:\n",
    "            print(\"Check the side!\")\n",
    "            return\n",
    "\n",
    "        # Salva il risultato\n",
    "        output_path = os.path.join(output_dir, filename)\n",
    "        Image.fromarray(img).save(output_path)\n",
    "        output_path = os.path.join(output_dir, \"aug_\" + filename)\n",
    "        Image.fromarray(flipped_img.astype('uint8')).save(output_path)\n",
    "        \n",
    "        #print(f\"Processed and saved: {filename}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "create_flipped_steer_dataset('./train_balanced/2', './train_balanced_augmented/1', 'RIGHT')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "create_flipped_steer_dataset('./train_balanced/1', './train_balanced_augmented/2', 'LEFT')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Flipped not steered"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "def flip_no_steer(image):\n",
    "    img = image.copy()\n",
    "\n",
    "    panel = crop_panel(img)\n",
    "\n",
    "    # Flippa l'intera immagine\n",
    "    flipped_img = img[:, ::-1]\n",
    "\n",
    "    # Inserisci il pannello corretto nella nuova immagine\n",
    "    flipped_img[84:96, 0:96] = panel\n",
    "\n",
    "    return flipped_img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_flipped_dataset(input_dir, output_dir):\n",
    "    \"\"\"Crea un nuovo dataset con immagini flippate e pannelli regolati.\"\"\"\n",
    "    if not os.path.exists(output_dir):\n",
    "        os.makedirs(output_dir)\n",
    "\n",
    "    for filename in os.listdir(input_dir):\n",
    "        if not filename.endswith((\".jpg\", \".png\")):\n",
    "            continue\n",
    "\n",
    "        # Carica l'immagine\n",
    "        img_path = os.path.join(input_dir, filename)\n",
    "        img = np.array(Image.open(img_path))\n",
    "\n",
    "        flipped_img = flip_no_steer(img)\n",
    "\n",
    "        # Salva il risultato\n",
    "        output_path = os.path.join(output_dir, filename)\n",
    "        Image.fromarray(img).save(output_path)\n",
    "        output_path = os.path.join(output_dir, \"aug_\" + filename)\n",
    "        Image.fromarray(flipped_img.astype('uint8')).save(output_path)\n",
    "        \n",
    "        #print(f\"Processed and saved: {filename}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "create_flipped_dataset('./train_balanced/0', './train_balanced_augmented/0')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "create_flipped_dataset('./train_balanced/3', './train_balanced_augmented/3')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "create_flipped_dataset('./train_balanced/4', './train_balanced_augmented/4')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[738, 738, 738, 738, 738]"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[len(os.listdir(os.path.join('train_balanced_augmented/0'))),\n",
    " len(os.listdir(os.path.join('train_balanced_augmented/1'))),\n",
    " len(os.listdir(os.path.join('train_balanced_augmented/2'))),\n",
    " len(os.listdir(os.path.join('train_balanced_augmented/3'))),\n",
    " len(os.listdir(os.path.join('train_balanced_augmented/4')))]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Mark edges"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess_images(input_dir, output_dir):\n",
    "    os.makedirs(output_dir, exist_ok=True)  # Creare la directory di output\n",
    "    \n",
    "    i = 0\n",
    "    for filename in os.listdir(input_dir):\n",
    "        \n",
    "        if filename.endswith(('.jpg', '.png', '.jpeg')):\n",
    "            filepath = os.path.join(input_dir, filename)\n",
    "            \n",
    "            # Leggere l'immagine\n",
    "            img = cv2.imread(filepath, cv2.IMREAD_GRAYSCALE)\n",
    "            \n",
    "            # Applica il filtro Sobel per evidenziare le linee\n",
    "            sobelx = cv2.Sobel(img, cv2.CV_64F, 1, 0, ksize=3)  # Bordi verticali\n",
    "            sobely = cv2.Sobel(img, cv2.CV_64F, 0, 1, ksize=3)  # Bordi orizzontali\n",
    "            sobel_combined = cv2.magnitude(sobelx, sobely)      # Combina i bordi\n",
    "            \n",
    "            # Normalizza il risultato per salvarlo\n",
    "            sobel_combined = np.uint8(cv2.normalize(sobel_combined, None, 0, 255, cv2.NORM_MINMAX))\n",
    "            \n",
    "            # Salva l'immagine preprocessata\n",
    "            output_path = os.path.join(output_dir, filename)\n",
    "            cv2.imwrite(output_path, sobel_combined)\n",
    "            i += 1\n",
    "\n",
    "            # Facoltativo: mostra una preview\n",
    "            #plt.imshow(sobel_combined, cmap='gray')\n",
    "            #plt.title(f\"Preprocessed: {filename}\")\n",
    "            #plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess_images_top_only(input_dir, output_dir):\n",
    "    os.makedirs(output_dir, exist_ok=True)  # Creare la directory di output\n",
    "    \n",
    "    for filename in os.listdir(input_dir):\n",
    "        if filename.endswith(('.jpg', '.png', '.jpeg')):\n",
    "            filepath = os.path.join(input_dir, filename)\n",
    "            \n",
    "            # Leggi l'immagine\n",
    "            img = cv2.imread(filepath, cv2.IMREAD_GRAYSCALE)\n",
    "            height, width = img.shape\n",
    "            \n",
    "            # Ritaglia il pannello inferiore e copia\n",
    "            \n",
    "            panel = crop_panel(img)\n",
    "            top_img = img[:height - panel.shape[0], :]  # Parte superiore\n",
    "            \n",
    "            # Applica il filtro Sobel solo alla parte superiore\n",
    "            sobelx = cv2.Sobel(top_img, cv2.CV_64F, 1, 0, ksize=3)  # Bordi verticali\n",
    "            sobely = cv2.Sobel(top_img, cv2.CV_64F, 0, 1, ksize=3)  # Bordi orizzontali\n",
    "            sobel_combined = cv2.magnitude(sobelx, sobely)          # Combina i bordi\n",
    "            \n",
    "            # Normalizza la parte superiore\n",
    "            sobel_combined = np.uint8(cv2.normalize(sobel_combined, None, 0, 255, cv2.NORM_MINMAX))\n",
    "            \n",
    "            # Ricombina il pannello originale\n",
    "            combined_image = np.vstack((sobel_combined, panel))\n",
    "            \n",
    "            # Salva l'immagine finale\n",
    "            output_path = os.path.join(output_dir, filename)\n",
    "            cv2.imwrite(output_path, combined_image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Cartelle del dataset\n",
    "input_dir = 'train_balanced/4'\n",
    "output_dir = 'train_balanced_edges/4'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Preprocessing\n",
    "preprocess_images_top_only(input_dir, output_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "369\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "369"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(len(os.listdir(os.path.join(input_dir))))\n",
    "len(os.listdir(os.path.join(output_dir)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Remove augmented files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'os' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m\n",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)\n",
      "Cell \u001b[1;32mIn[1], line 1\u001b[0m\n",
      "\u001b[1;32m----> 1\u001b[0m \u001b[38;5;28mlen\u001b[39m(\u001b[43mos\u001b[49m\u001b[38;5;241m.\u001b[39mlistdir(os\u001b[38;5;241m.\u001b[39mpath\u001b[38;5;241m.\u001b[39mjoin(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mtrain/edges/0\u001b[39m\u001b[38;5;124m'\u001b[39m)))\n",
      "\n",
      "\u001b[1;31mNameError\u001b[0m: name 'os' is not defined"
     ]
    }
   ],
   "source": [
    "len(os.listdir(os.path.join('train/edges/0')))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for filename in os.listdir('train/edges/0'):\n",
    "    if filename.startswith('aug'):\n",
    "        os.remove(os.path.join('train/edges/0', filename))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".env",
   "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": "3.12.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}