preprocess.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "import glob, os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_label(img_path):\n",
    "    filename, _ = os.path.splitext(os.path.basename(img_path))\n",
    "    \n",
    "    subject_id, etc = filename.split('__')\n",
    "    gender, lr, finger, _ = etc.split('_')\n",
    "    \n",
    "    gender = 0 if gender == 'M' else 1\n",
    "    lr = 0 if lr =='Left' else 1\n",
    "    \n",
    "    if finger == 'thumb':\n",
    "        finger = 0\n",
    "    elif finger == 'index':\n",
    "        finger = 1\n",
    "    elif finger == 'middle':\n",
    "        finger = 2\n",
    "    elif finger == 'ring':\n",
    "        finger = 3\n",
    "    elif finger == 'little':\n",
    "        finger = 4\n",
    "        \n",
    "    return np.array([subject_id, gender, lr, finger], dtype=np.uint16)\n",
    "\n",
    "def extract_label2(img_path):\n",
    "    filename, _ = os.path.splitext(os.path.basename(img_path))\n",
    "    \n",
    "    subject_id, etc = filename.split('__')\n",
    "    gender, lr, finger, _, _ = etc.split('_')\n",
    "    \n",
    "    gender = 0 if gender == 'M' else 1\n",
    "    lr = 0 if lr =='Left' else 1\n",
    "    \n",
    "    if finger == 'thumb':\n",
    "        finger = 0\n",
    "    elif finger == 'index':\n",
    "        finger = 1\n",
    "    elif finger == 'middle':\n",
    "        finger = 2\n",
    "    elif finger == 'ring':\n",
    "        finger = 3\n",
    "    elif finger == 'little':\n",
    "        finger = 4\n",
    "        \n",
    "    return np.array([subject_id, gender, lr, finger], dtype=np.uint16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6000\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f8b4164cc50>"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABlCAYAAAAiRp9EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAHxNJREFUeJztfXtQVEf2/6dneL9ExreC4yPK+qOQMpRSxkSJgqE0KmuishtFV9RQFaNu1kTWxxJZWdRoYlg1UXyyiaAQVDYrShSQgCJEQL4SHiIgykNgEEYYYJj5/P4A70Kixiei66dqam73dPfp+7l9T5/Tfe4dQRIv0TmQPesO/C/hJdmdiJdkdyJekt2JeEl2J+Il2Z2Il2R3Ip4J2UIICiHqhRAbnoX8R4UQYpgQ4rYQQieE8HnY+s9yZI8kufpOQgjxthDi/9pOJlkIMeJeFUUrNgohqts+G4UQ4h5ljYQQEUKIoraLPOF+nRJCGAsh9goh6oQQ5UKIP9/5jWQeSQsAiQ9/ul1EjQghXgHwDYD3AVgDiAZwXAhhcI8qiwHMADASgCOAtwEsuY+IHwG8B6D8AbrjD+AVAAMBuAL4WAjx1gPU+22Q7PQPAAIY2i79AYDv26VlADQAJt6jfjKAxe3SCwGcfwC51wFM+I0ypQDc26UDAIT9okw8AJ+HPe8uMbLbIH5xLAA43KPs/wOQ2S6d2Zb3eB0QojuAvk+jbaCLqBEAPwAYL4SYIIQwAvBXAEYAzO5R3gJAbbt0LQCLe+nth4BFu/bat235mO0C6CJkk8wB4A3gnwDKAPQAkI3W2/5uuA3Aql3aCsBtPv4S5u127bVvW/2Y7QLoImQDAMkIkg4kFQD+BkAJIPUexS+jdXK8g5FteY/bhxq0Xuwn3jbQhcgWQrwqhJALIXoC2AXgeNuIvxsOAvizEKK/EKIfgI8A7L9P28ZCCJO2pJEQwuQ+KucggDVCiO5CCHsAi+7X9kOhK1gjbXk/ovV2VQH4GoD5feoLAJvayqrajsV9yhe1yWz/Ud6jrDGAvQDqAFQA+PNdysTjEayRZ0V2I1onnoBnIf8x+v0KgFsAGgDMf9j6oq2RR0Kbsb8NgBxACMmgR27sfwCPTLYQQg4gD4AbWq2GVABeJLOfXPdeLDzOBDkawBWSV0k2AwgDMP3JdOvFxL3WHh4E/QGUtEtfBzDmfhUsLS2pUCgeQ2TXQ3FxMQCA5G86VI9D9gNBCLEYrQtHUCgUKCoqetoiOxUP47Q+jhq5AcC2XXpAW14HkNxF0pmks6XlE/F6n1s8DtmpAF4RQgxqW8+YA+D4k+nWo6GkpOS3Cz1DPDLZJFvQujR6EsDPAA6TfCJu7aNCLpd3SN++ffuu5SoqKgAAzc3NT71P7fFY7jrJ/5AcRnIIyU7Z4qqt/e+CnF6vBwBcv34dBw8ehLW1tfSbVquFRqPB7du30dTU1KENAwMDXL9+HTExMZ3R5f+iMz0wpVLJx0VLSwtJMjo6Wsprampieno69+7dy02bNv2qjlarpV6v59q1a6nRaDr81tDQwOrqap46dYokWVpa+lD9QZv7z67mrj8JsvV6PUmyurqaZ86c4XfffSddALL1YoSGhlKhUPDLL7/8Vf36+nrOnz//nm3X19c/VH9eWLLXrFlDDw8ParXaDvnm5uZ0dnbmuHHjmJmZyaysLA4ePJjdunUjSdbV1XW4IN26dWNYWBirqqo6tHPnQj4MXiiyf0msWq3mq6++yps3b/L06dPU6XQkycbGRkZFRdHX15ck+dNPPzE/P58kOXz4cI4bN+5Xbfv5+fHy5ctSOjQ0lBs3bmR+fj6rq6sfqH8vDNlNTU308/Pjvn37uGDBAgKgj48Pv/32W6nMd999x4yMDE6YMIGRkZGsrq6mVqu9qzqoq6vjF198wcWLF3cYxceOHWNJSQlJMjU1le+///5dVdDd8MKQfbfb+vDhw3R3d+fAgQOZmprKDRs2MDs7u0OZ5uZmenp6UqlUMjY2lsnJyR1+f+utt1hXVyfdFSSZk5PDjz/+mGFhYQ/Vx+ee7M2bN/PDDz/ksmXLePToUSm/qKiIJHnx4kVGREQwKyuLpqamnDJlCleuXMmkpCSpbE5ODsPCwpiVlSUReOPGDZJkeno6W1pauHTp0g5yDxw4wBEjRnD58uVcvnz5A/X1uSe7sbGRhYWFrKqqYkpKCknSycmJhoaGFEIwKiqK4eHhXLFiBbVaLePi4njy5EkGBQVxwYIFnDx5Mj08PFhTU0Oy1UKJjY2ll5cX+/bty9raWinf3t6eGzZs6CA/MTGRUVFRdHZ2frHJ/stf/kInJydpYtTpdBJpZKtdrNFoWFpaykuXLrFnz56Uy+WsrKyUyqjVakZERHDp0qXs3bs3dTqdZEdXVFSQbFU1d9IODg6UyWTSyL8DvV7PgoKC+6qX55rskpISlpeXMzg4mHZ2djQyMmJbjF6HPcTQ0FDq9foOel2tVrNnz55MTU2lTqeTftPpdCwoKKAQgiYmJgwODua1a9d+JXvr1q1cvnw5vby8GBsbK9XX6/VsaGh48cgm2cEmJsmysjJ+/vnn7NGjB3v06EFzc3MCoEwmo62tLauqqqSRSpLnz5+no6MjbWxsqNPpJK9Ro9EwMTGRQUFB7Nu3r3THtL9zsrKyOHXqVALgjh07JJVzLzyXZOv1eqakpDAmJoaJiYkcOnQoHRwcGB0dzYsXL/Ls2bOMiopibm4udTodGxoauGfPHrq4uNDFxYWmpqYsKyuT2mtubub27dvp4uJChULBJUuW3FXugQMHaGpqSplM1uEiNzU1UaPRcO3atWxububhw4dfHLJJ/srtbmlp4fr167lo0SLOmzeP8+bN4x//+EcCoBCC8+fP54EDB1hXV0eyVY2MGTOGFy9elNoRQnSQsXjxYjo5OVEul7OgoEDKDw4Opq2tLUNCQqhSqUiScrmcsbGxJMkFCxZwxIgRLwbZWq2W69evl1SEoaEhTU1NuXr1aol8nU4n6drAwEAGBgZy8+bNdHZ2poODA728vHjlyhWSrark5s2bJMndu3dTCME1a9bw1KlTVKvV1Gg0vHz5MgFw1apVkrdJtnqWP//8M0nyvffe69DPhQsXdrhIzyXZISEhNDQ0pJGRERUKBX19fWllZcXu3bvzwoULXLJkCX19fenp6Uk/P78OE1h8fDyDg4O5efNmfv755ywpKaGfnx+9vLyo1Wp59uxZ7ty5k6NGjeKoUaPo7e3NxMREkqS/vz8BcOnSpdKdtW3bNnp4eJAkMzIySFIa7cnJyZK3+dySfQdarVZaJFKr1VJ+QkICk5KSGBsby+nTp9PAwIDDhg3j0KFDOXXqVObm5koE5uXlkWy1QoyMjOjq6sqzZ8+yubmZtbW1nD59Om1sbDhp0iRWVFTw0qVLrK6uppeXlyRPr9fTyclJsniio6N5t/4/d2Tr9XoOHDiQffr0kTpvYGBAIQRHjhzJN954g8XFxczNzaVer5d0Z11dHYcOHUo3NzcWFRWxqKiIOp2O/fr1o52dnUR6UVER8/LyCEAy4/R6PXNzcymXy7ls2bIONrZSqeQbb7xBstUUHT58uFTmzkric0t2e9Jra2up0WhYWVnJ6upq6nQ6fvLJJ7S1taWJiQllMhmXLVtGADQ2NmZjYyObm5t55coVtrS0SKqlpKSEjY2NFEKwqalJkqFSqWhkZEQA0mJVWVkZlUolhw4dKq2XBAcHUwjBc+fOsbm5mdHR0QRAtVpNIYS0dPBckl1aWsqrV69KnTc0NCQA7tq1S5roKioqOHPmTGZnZ7Nv377s27cvLSws6ODgQAD09/fn6dOnOzg6Dg4OHDJkCEtLS1lTU0OdTkeVSsXBgwezV69e0oSbmZlJGxsbHjt2THJgnJyc2LNnT2ny7NWrF1etWsUxY8ZIKu65JHvcuHFcv349i4uLmZWVRb1ez8DAQDo7OxMAzczM+M4779DFxYXx8fG8dOkSf/rpJ5aVlVGv17OoqIhqtZrjxo2jsbExfX192djYKJEyZcoUfvjhh3eIIdlqKrq5uXXox+HDh/nuu+8yODiY+fn51Ov19PX1pUajoUql+pXD9cTJRmvIbRaADABpbXk2AGIB5Ld9d39UstVqNYODg7l9+3b27t27g1uuUChYVVXF/fv3MyAggD4+PlywYAE9PT05atQoAqBSqaSvry9PnTrF+vp66nQ6btu2jWvWrKGJiQm3b98u7VkGBgZSo9Fw586d7N27t9SHxMREzp49W3KMQkJCOGfOHMbHx0uWyPz582lpafnIZD/M7rorSSeSzm3pVQBOk3wFwOm29ENj0qRJmDhxIoqLi5Gamop169YhJSUFycnJSEtLwx/+8Ae88soriI+Ph4WFBWxtbWFiYgIfHx8kJSVBr9cjPz8f9vb2iI6OhqOjIyZNmgRbW1t8/PHHKCsrQ3Z2NmbNmoVNmzZBJpPBxMQEixcvRkZGBgDg3LlzGDduHFavXo25c+fi9ddfh6WlJQ4dOoS0tDTY2NjAxMQE+/btw65du9DY2Pgop/pQI7vHL/JyAfRtO+4LIPdRRvZnn33GU6dO8datWywuLpbyV6xYQU9PTwKgXC6nj48P/fz8aGFhQQB0dHTkvHnzGBQURH9/f5JkZWUljx8/TldXV9rb23PevHksKChgRUUFDx48yIiICFpbW0sj/A4yMzPp6+vLpqYmJicnEwBHjx7NnTt3sr6+njNnziQAnj9/nmSrXX9nIsVTUCOFAC4C+Altzx8CuNXud9E+/Yu6iwGkAUhTKBQkW21pnU5HnU7Huro6jh49mgMGDGD37t1pYWHBy5cvMyEhgaWlpZw5cybfeOMN+vr6ct++ffTy8mJeXh4vX77MtLQ0njlzhkIIOjk5MScnhx4eHjx8+DALCwuZkpJCOzs72tracuzYsSRbd2kaGhoYFxfHbdu2cdKkSSTJlJQUGhkZ0c3NjTqdjj///LNEul6vZ1ZWFvv37y8tDdyZhB+G7AeKzxZC9Cd5QwjRC636eSlan3mxblemhmT3+7UzaNAgFhYWAgA8PT2RmJiI6upquLm5YefOnejWrRu6d+8OuVyOsLAwVFVVwczMDLdv34a5uTmam5tx/vx5KBQKbN26FXl5eThx4gTc3d3Rr18/mJmZwdHRETk5OejduzeMjIxw9epVFBcXo7CwEFOmTEF9fT3kcjnWrVuHr7/+Gjdu3ICBgQFMTEywZ88eXL9+HWZmZkhLS0NgYCCUSiUmTJiAI0eO4OrVqwgLC8PmzZsBtEZc3Ylf5ANEsT60RYHWx43/gsdUIzKZjAC4ZcsW/vzzz9TpdLxx4wbd3Ny4YcMG1tTUsKSkhKWlpSwoKGBeXh7VajW1Wi0bGxupUql4+fJlyuVyGhoa0svLi2VlZWxoaODRo0elETd69GhevnyZer2eLi4uTEhIkO4uABw8eDDJ1t14AwMDtrS00NHRkTKZjHq9nmPGjKEQgjY2NtTr9czOzpY8yTttPOjIfhByzQFYtjtOBvAWgM0AVrXlrwKw6WHI1ul0LCsr4/nz52lubs4+ffpwwIABHDlyJF977TUuWLCAAQEBDA8Pp52dHfv378/+/fvT1taWAwYMYEREBK2srAiAp06d4rRp0zhgwACGh4czOTmZ1dXVfPvtt1lQUEClUkkHBwfJbFu2bBmPHz/OpqYmqtVqAmBAQIC0O6TX6/n222/T2dmZ5eXlTExMpFwu59atW6nX6zl9+nRaWVk9tBp5kPjs3gCi2uKQDQB8SzJGCJEK4LAQYiGAYgCzHqAtlJeXo7m5GY6OjrCwsIBCoUBSUhKys7PRrVs3fPTRR6iurkZOTutTeaNHj4ajoyOsra2l293KygpWVlb4/vvvkZqaCj8/P2RlZWHFihXo2bMnCgoKMG/ePHh7e0Or1aKwsBA//vgjDAwMkJiYiM2bN+PTTz/Fe++9h0uXLuH06dNYsmQJZDIZ/Pz8oFKpEBUVheLiYvTp0wckodVqYWBggDfffBNHjx7Fu+++i7lz5z7IKUt4rAeYHhZ3dPbatWvR2NgIe3t79O/fH7Nnz0ZdXR1GjRqFkSNHQqlUok+fPhgyZAiGDRuG4uJi/O53v0N1dTWCg4Oh0WiQn58PMzMzpKen44MPPsAHH3yAwsJC7Nq1CxqNBqtXr0Z4eDh27NgBJycneHp6YvTo0UhPT4cQAtOmTYOlpSUUCgVu3boFIQRef/11NDc346uvvsLvf/97XL16FRs2bEBkZCTi4+ORkZGBCRMmoKKiAr169cI///lPLF26FMBT0tmP87mjRu4sa3700Ud0cXGhkZGRZG3U19dTr9fzyJEjDAgIYGBgIAsLC0m2Oj8nT57kxYsXmZKSwkuXLnHnzp0EwK+//lqycJYuXco5c+Zw7dq1/OqrrwiARkZGzM7OZnNzM62trfnhhx+ypaWF3t7erKqqokql4pkzZyiXy9nS0sK5c+dKDhAArlixQlI5fn5+0jYcurK7rtPpGBMTQ3t7e2mCPHPmDAcMGEAhhLRf6OHhwXXr1nHRokUMCQkh2bqHuHHjRgLg2LFjKZfLOW/ePJaWllKr1VKr1XLPnj3U6/XMycnh+vXrOWLECFZVVfHWrVsEwKlTp7KkpIT79++nEIJ6vZ6JiYlcsWIFZ8yYwYiICHp4eEjls7KyuHfvXs6ePZvFxcWMjo6mTCbj9OnTuz7Zer2eOp2OiYmJPHv2rLSItHfvXsbHx7OpqYnR0dEcNWoUBw8eLC1IjRgxgk5OThwzZgzHjRvHLVu2sLy8nL6+vgRAZ2dnuri4UK/X89ChQywpKWFeXh7j4uIYEBDAadOmUaPRMDQ0lLa2tlSr1SwoKKCRkRHJ1klRJpPRwcGBJ06coBCCJSUlBMCqqiquWLGCAKjRaKRVw4kTJ3Ztsuvq6qhSqbh06VJpmXTy5Mk0NjamsbExDQ0NGRgYyPz8fObm5jIlJYXx8fFUq9VUqVTMyclhSkqK5FlevnyZLS0tLC4upq2tLc3NzfnDDz+wpKSE/fr1o6WlJWtqatjS0kJjY2NmZ2dTp9NRqVRy9+7dLC0tpaGhIW1sbFhVVSVdAABsaGjghQsXaGVlxXPnzrF79+50cnKiSqXi5s2bmZWV1bXJJslvv/2WQgi+//77PHbsmLSB26NHD6anp9PPz4//+Mc/pHgRIQTLy8u5Y8cOGhkZMTk5mXl5eYyOjqYQgpmZmVSpVJJOFUJQrVbz2rVrtLS0ZHl5ORsbG2lhYSGpDjc3N2lU34lJkclk1Gq1vHLlCpctW8b09HTJlrawsGBUVJQ0um1tbbu+GtFqtezduzezsrKkyNQ5c+awsLCQLi4utLa2poeHB8vLy5mcnExfX1+q1Wqq1Wrm5+dzypQpdHJykk4yISGBdXV1NDc3Z2BgIK9cucJPP/2UALh48WKq1WqamprSxMSEpaWl7Nu3L9PT01lbW8u1a9dKoz0yMpI1NTUEwMLCQtbW1ko628PDg3K5nPX19RwyZAhlMhlra2ufvFPzpMm+efMmVSoVTU1NaWBgwKCgILq6utLBwYHz5s3jV199xXXr1tHOzo6mpqZ0cnKiWq2mgYEBAdDb25tpaWnMzs5mYGAgHR0duXPnTtbU1PCTTz6hTCZjaGgoT548yWPHjnHEiBGsra1lWloaAbCgoICjR48mAN66dYsjR47kzZs3uXbtWvr7+0sDICQkhOfOnZN0dklJCbdu3cqmpiaOHz+eY8eOpUql6tpk6/V65ufn08/Pj8eOHeP06dOZl5fHkJAQDh48mPb29kxLS2NCQgLz8vK4fv16bt68mbGxsSwuLmZGRgYnTpxImUzGdevW8datWxw4cCDlcjkXL17MyspKDhkyhMOHD+eZM2fo6urKDRs2sLS0VDIzy8vL6e/vT09PT6pUKubn5/Pjjz/m8OHD+dFHH3Ht2rW0t7dnYmIi7e3tuWDBAubk5NDAwIBFRUXctGkT33//fQYEBHRtsnU6HVNTU5mVlcXJkyfT1NSUycnJtLe356xZsxgcHMwlS5ZwxowZPH/+vBQ/MnnyZAYFBXHlypXcv38/zczM2KtXL2o0Gm7bto1CCPbr148qlYrbt28nAH755Zc8e/YshRBcv349Y2JiaGdnx4SEBFZVVbFXr17MyclhaGgoAfDzzz8nAF64cIGurq6cO3cug4ODOWDAAK5atYpOTk708/OjQqGgn5+ftFzQZcmura1lQUEBX331VQohGBISQiEEPT09pQUkJycnbtu2jRkZGfzmm28YGhrKiIgIRkdHc86cOQRADw8PHjhwgPHx8Zw1axYjIiK4Y8cOZmdns6amhlu2bJHUxvr16/npp59yzpw5DAsLY58+fXjixAkeOXKEMpmMJSUl3LVrFz08PLhy5UoKIejq6spDhw4xNDSU/fr1IwBeunSJhoaGvHHjBo2NjVlVVdW1yW5oaODFixdpYWHB7777jnK5nP7+/uzZsycBMCoqiv7+/lQqlVJM9oQJE1hcXMzCwkKWlJQwOjqa1tbWnDx5Mnfv3s1FixbR2tqaixYt4q5du2hra8u4uDhGRkbS0NCQiYmJHD58OP/2t79J0awAmJiYSAD84osvmJCQQAsLCxoaGjIoKIgHDx7ku+++S6VSyezsbI4bN45HjhxhUlISo6OjWVBQwPz8/K5NdmVlJTMzM5mTk0MHBwdOmzaN7u7unDhxItPT02lhYcGxY8fy0KFDDAgIoEKhoIuLC5OTkxkZGck1a9YQAFeuXMnMzEwCkNQIACmkzNDQkBs3bmRYWBjlcjlPnTpFHx8fent7MyUlhV5eXjQ0NOSlS5d48uRJpqWl0djYmJ6enhRC8IcffiAA5uTkcPny5czKyqKFhQUrKyv5xRdfMDU1lSS7NtkNDQ3MyMiQ7OgTJ04QAJOTk7l//37KZDIeOXKEBw8epBCCdnZ2lMlkHD9+PF977TUpH21Lq3dCFUJDQ6lUKpmQkMCgoCCJhMLCQm7dupU5OTlS7MfBgwclpygzM5Px8fGcMmUKJ0+eTEtLSwKgtbU1HRwcePz4cR49epSenp48d+4cCwoKpLiWLm/6FRcX8/z58/zPf/7Do0ePUi6X87PPPqOPjw8dHR3p6+tLhUJBNzc3Xr16lSkpKdy7dy8nTJjAPn360MTEhN9//70UgBkeHs6FCxdKXp5SqeS0adM4ZcoUHjlyhCdOnGB+fj5v3rzJ8PBw9urVi+PHj+eBAwekC52YmMi+ffty7NixHDNmDFevXk13d3d+/fXXlMlkPHnyJGNiYujj48Pg4GBWVFRQrVZzx44dXZtsjUbD06dP88yZM1y1ahXHjx9Pd3d3rl69mj179uSIESOoVCppZmbGyMhIBgUFcfbs2TQzM6MQgm5ubgTAM2fO8J133iEAuri40M/Pj2ZmZty9ezcBMCUlhRcvXmRQUBCzsrK4fft2zpo1izExMezWrRvNzc2ZmZlJZ2dnLl68mDExMdIdNnjwYKamprJ79+5MSEigEIIbNmygk5MTc3Nzped9Kisrn/we5JPCoEGDmJ2djcbGRhQVFWHLli0YNGgQ/v73v8POzg4ajQaenp6wsbHBa6+9hmPHjsHExATGxsZoaGjAtWvXEBcXhz/96U/497//DXd3dygUChgbG2Pjxo0IDw+HlZUVNm7cCAMDA8ycORORkZFYvnw5Jk2ahG+++QZhYWFYuHAhLly4gIiICOzbtw+urq4YNGgQpk2bhsLCQjg4OCA4OBhbtmyBj48PVqxYgd27d+PHH3+EXC7H0KFDUVdXBxMTE5iZtb7BlA+wnt3pL1EUQqCsrAzm5ub461//iv79+6Nfv364du0aZs6cid27d8PT0xPffPMN/vWvfyEkJARffvkl9u3bh5aWFrz55pu4fv06SkpK0NDQAFdXV1y/fh0ajQYhISEIDw+Ht7c3oqOj0djYiKSkJNy8eRP9+/fHiRMn4OvrCwCwsLBAfn4+cnNz4e7ujpycHEydOhXFxcUYNmwY1Go1+vXrBxsbG8yYMQPvvPMObt++jWPHjkEmk8HU1BRGRkYPd/LPYoKsrq5mYWEhKysrGRcXx6ysLMbFxXHq1KkMCQnhqFGj6O/vz40bN/Knn37i3r17CYBDhw5lZGQkjY2NuXXrVgJgTEwM58+fzxkzZtDd3Z1eXl5UKBQ8ePAglUolZ82aRblcznXr1tHT05NJSUnct28f9+zZI6muixcv0sHBgWPHjuWuXbtobW3NAwcOEABjY2OZlJTEuLg4ent7MycnhxkZGdRoNCwvL++6aqTNingh0SXVyP8ynvrbz36B22iNN+ks9ABQ1QlyBj5Ioc4mO5f/Dcx86hBCpHWmvN/CSzXSiXhJdieis8ne9YLLuy861fT7X8dLNdKJeEl2J6LTyBZCvCWEyBVCXBFCPNLzNw8go0gIkSWEyBBCpLXl2QghYoUQ+W3f9w3Yf6rojDURtL6mvwDAYLT+2U8mgBFPQU4Rfv3szyZ0jCPf2JnrQe0/nTWyn+Vb5KcDONB2fACtfwD3TNBZZN/tLfL9n4IcAjglhPip7SXpANCbZFnbcTlag/ufCTrbXX/aGMd2D1oJITr8qRApxfU9E3TWyH6gt8g/LkjeaPu+CSAKreqrQgjRFwDavm8+abkPis4i+6m/RV4IYS6EsLxzDMAdwP+1yfFuK+YN4NiTlPsw6BQ1QrJFCHHnLfJyAHv55N8i/0QftHoaeOmudyJeepCdiJdkdyJekt2JeEl2J+Il2Z2Il2R3Il6S3Yn4/zmbZPKQIkFbAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8b4164ca20>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img_list = sorted(glob.glob('Real/*.BMP'))\n",
    "print(len(img_list))\n",
    "\n",
    "imgs = np.empty((len(img_list), 96, 96), dtype=np.uint8)\n",
    "labels = np.empty((len(img_list), 4), dtype=np.uint16)\n",
    "\n",
    "for i, img_path in enumerate(img_list):\n",
    "    img = cv2.imread(img_path, cv2.IMREAD_GRAYSCALE)\n",
    "    img = cv2.resize(img, (96, 96))\n",
    "    imgs[i] = img\n",
    "    \n",
    "    # subject_id, gender, lr, finger\n",
    "    labels[i] = extract_label(img_path)\n",
    "    \n",
    "np.save('dataset/x_real.npy', imgs)\n",
    "np.save('dataset/y_real.npy', labels)\n",
    "\n",
    "plt.figure(figsize=(1, 1))\n",
    "plt.title(labels[-1])\n",
    "plt.imshow(imgs[-1], cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "17931\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f8b41624b70>"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8b41652be0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img_list = sorted(glob.glob('Altered/Altered-Easy/*.BMP'))\n",
    "print(len(img_list))\n",
    "\n",
    "imgs = np.empty((len(img_list), 96, 96), dtype=np.uint8)\n",
    "labels = np.empty((len(img_list), 4), dtype=np.uint16)\n",
    "\n",
    "for i, img_path in enumerate(img_list):\n",
    "    img = cv2.imread(img_path, cv2.IMREAD_GRAYSCALE)\n",
    "    img = cv2.resize(img, (96, 96))\n",
    "    imgs[i] = img\n",
    "    \n",
    "    # subject_id, gender, lr, finger\n",
    "    labels[i] = extract_label2(img_path)\n",
    "    \n",
    "np.save('dataset/x_easy.npy', imgs)\n",
    "np.save('dataset/y_easy.npy', labels)\n",
    "\n",
    "plt.figure(figsize=(1, 1))\n",
    "plt.title(labels[-1])\n",
    "plt.imshow(imgs[-1], cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "17067\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f8b4136c198>"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8b41368b00>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img_list = sorted(glob.glob('Altered/Altered-Medium/*.BMP'))\n",
    "print(len(img_list))\n",
    "\n",
    "imgs = np.empty((len(img_list), 96, 96), dtype=np.uint8)\n",
    "labels = np.empty((len(img_list), 4), dtype=np.uint16)\n",
    "\n",
    "for i, img_path in enumerate(img_list):\n",
    "    img = cv2.imread(img_path, cv2.IMREAD_GRAYSCALE)\n",
    "    img = cv2.resize(img, (96, 96))\n",
    "    imgs[i] = img\n",
    "    \n",
    "    # subject_id, gender, lr, finger\n",
    "    labels[i] = extract_label2(img_path)\n",
    "    \n",
    "np.save('dataset/x_medium.npy', imgs)\n",
    "np.save('dataset/y_medium.npy', labels)\n",
    "\n",
    "plt.figure(figsize=(1, 1))\n",
    "plt.title(labels[-1])\n",
    "plt.imshow(imgs[-1], cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "14272\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f8b401d86d8>"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABlCAYAAAAiRp9EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJztfXtcVVX6/rPOAQ73lFsqggiOkBFSMmhqIaOCZBeYzMvMpDGhE59+TpqaUtmQhkGO5iXL0DAhC1REpBJl4hKKIMjFkwQiN48i99vhfjjn+f0Bnq82XSSNqPH5fM7nnLX2uy772Wu/+13vetc+giTuYnAg+bU78L+Eu2QPIu6SPYi4S/Yg4i7Zg4i7ZA8i7pI9iPhVyBZCUAjRLoQI+TXa/7kQQowXQrQJIdRCiICBlv81R/ZEkq9dTwghnhBCfNN/MhlCiAk/VFD0IUwI0dD/CRNCiB+Q1RNCHBZCVPRf5Bk/1ikhhEwIESGEaBVCVAshXr5+jORFksYA0gd+ukNEjQgh/gDgAIAXAAwDkADgmBBC5weKLAPgC2AiABcATwD4x480cQrA3wBU30J3ggH8AcAYAJ4AXhFCzLmFcj8NkoP+AUAA425I/z8AX9yQlgDoBDDzB8pnAFh2Q/p5AJm30O4VADN+QqYKgNcN6Y0Aor8jkwogYKDnPSRGdj/Ed34LAM4/IHs/gIIb0gX9ebfXASGGAxj5S9QNDBE1AuA/ADyEEDOEEHoAXgWgB8DwB+SNAbTckG4BYPxDensAML6hvhvrNrnNegEMEbJJFgFYAuA9ANcAWAAoRN9t/31oA2B6Q9oUQBtv34XZdkN9N9atvM16AQwRsgGA5GGSziTNAfwLgB2A7B8Qv4C+h+N1TOzPu90+NKHvYt/xuoEhRLYQYpIQQiqEsAQQDuBY/4j/PkQCeFkIYS2EGAVgFYCPf6RumRBCvz+pJ4TQ/xGVEwngdSHEcCGEE4ClP1b3gDAUrJH+vFPou10bAXwIwOhHygsA7/TLNvb/Fj8iX9Hf5o0fux+QlQGIANAKoAbAy98jk4qfYY38WmR3oe/Bs/HXaP82+v0HAM0AOgA8N9Dyor+Sn4V+Y387ACmAvSRDf3Zl/wP42WQLIaQALgKYjT6rIRvAIpKFd657vy/czgPSHcAlkmUkewBEA3jqznTr94kf8j3cCqwBKG5IXwEw+ccKmJiY0Nzc/DaaHHqorKwEAJD8yQnV7ZB9SxBCLEOf4wjm5uaoqKj4pZscVAxk0no7auQqAJsb0qP7824CyXCSbiTdTEzuyKz3N4vbITsbwB+EEGP7/RkLARy7M936eVAoFD8t9CviZ5NNshd9rtETAL4FcJDkHZnW/lxIpdKb0m1tbd8rV1NTAwDo6en5xft0I25ruk7yS5LjSTqQHJQlrpaW/3PIaTQaAMCVK1cQGRmJYcOGaY+pVCp0dnaira0N3d3dN9Who6ODK1euIDExcTC6/H8YzBmYnZ0dbxe9vb0kyYSEBG1ed3c38/LyGBERwXfeeee/yqhUKmo0Gq5fv56dnZ03Hevo6GBDQwNPnjxJkqyqqhpQf9A//edQm67fCbI1Gg1JsqGhgcnJyTxy5Ij2ApB9FyMqKorm5ubcsWPHf5Vvb2/nc88994N1t7e3D6g/v1uyX3/9dfr4+FClUt2Ub2RkRDc3N06fPp0FBQWUy+W0t7fnPffcQ5JsbW296YLcc889jI6OZn19/U31XL+QA8HviuzvEqtUKjlp0iTW1tbyq6++olqtJkl2dXUxLi6OgYGBJMlz586xpKSEJOno6Mjp06f/V91BQUG8cOGCNh0VFcWwsDCWlJSwoaHhlvr3uyG7u7ubQUFB3LdvH/39/QmAAQEB/PTTT7UyR44cYX5+PmfMmMHY2Fg2NDRQpVJ9rzpobW3ltm3buGzZsptGcXx8PBUKBUkyOzubL7zwwveqoO/D74bs77utDx48SC8vL44ZM4bZ2dkMCQlhYWHhTTI9PT308/OjnZ0dk5KSmJGRcdPxOXPmsLW1VXtXkGRRURFfeeUVRkdHD6iPv3myN2/ezH/+85986aWXePToUW1+RUUFSTI3N5eHDx+mXC6ngYEB586dyzVr1vD06dNa2aKiIkZHR1Mul2sJvHr1KkkyLy+Pvb29XL58+U3t7t+/nxMmTOCKFSu4YsWKW+rrb57srq4ulpeXs76+nllZWSRJV1dX6urqUgjBuLg4xsTEcOXKlVSpVExJSeGJEycYGhpKf39/ent708fHh01NTST7LJSkpCQuWrSII0eOZEtLizbfycmJISEhN7Wfnp7OuLg4urm5/b7JXr16NV1dXbUPRrVarSWN7LOLOzs7WVVVxfPnz9PS0pJSqZR1dXVaGaVSyZMnT7Kjo4Pt7e3s7u5md3c3lUolr127RrJP1ZB9d1FCQgJPnz5Nf3//m/qi0WhYWlr6o+rlN022QqFgdXU1d+7cSVtbW+rp6bE/Rk97YjKZjF9//TV7enqYkpJCjUZDtVpNpVJJS0tLxsXFcdu2bayqqmJDQwM7OjqoVqsZEhJCNzc3Ojk5UaFQMDc3l11dXbx69SpjYmJYVlZGtVrNd955h56enrSwsKChoSE9PDzo6enJF1988fdFNsmbbGKSvHbtGt99911aWFjQwsKC586dY01NDdeuXUsTExN+8cUXVCqVLC8vZ1JSElevXs2enh5qNBqWlZVxw4YNHDNmDGfOnMmIiAjW19ezsbGRu3fvppeXF21sbGhubs7a2lpWVFTwb3/7GwMDA5mamsq6ujqmp6f/oCk4ELJvaw1yoBg7dizLy8u/9xhJZGdno6mpCUZGRvD394e+vj7efvttWFtbo62tDQ0NDcjLy8OxY8fwxBNPYNq0adDR0YFSqcTnn3+Oxx57DBMnTkRBQQFOnToFuVyOefPmYeHChWhtbcW5c+dw7NgxPP744/jwww/x8MMPY/ny5VCpVNDX14dcLoetrS2Ki4vR2toKa2trPPzww9BoNFCpVJDJZP/V7+v+bN7C4sGQIRsA1Gq11nOnVqsBAJs2bYJCodA6kz744AN8+eWXiIiIwLhx42BkZIRVq1Zh69atcHR0xP79+7F27Vrcf//9uHbtGj7//HPtaoqfnx+srKxw8eJFFBQUoKioCA4ODsjMzIRMJsODDz4Id3d3/PGPf4SRkRH09fUhkUhQXl6O6upqHDt2DOHh4Tf1+TdJdm9vL95++22EhYWhvb0durq60NHRwcsvv4y33noLarUan3/+OVasWAE/Pz8kJSVh7dq1KCwsRE9PDxwdHSGTydDY2IiLFy/CzMwMtra28PLywgcffIC8vDxYWVlBrVajsLAQ4eHhGD9+PC5cuIDm5mY0NDTA0NAQY8aMwUcffYQlS5bgypUrGDt2LJyd++I7DQwMEBAQgFdffRX29vZ9BP4Wyf7oo48QGBgIIQRMTEwwf/58HDhwAFKpFCdOnEBkZCRyc3Px3HPPYcSIEaioqIC+vj4iIyOxbds2XLlyBdHR0Whra4O/vz+mTZsGExMTvPXWW9i1axemTp2K4OBgfPPNNwD6iFMqlfjkk0+wfv16tLS0YMSIEdiwYQPmzp0LNzc3pKenw8rKCv7+/mhpacHw4cNx5swZ2NjYYPTo0X0E/hbJvo7e3l7ExsZi1qxZkMlkMDbuCyy9cuUKTE1NsXDhQlRVVSEmJgbm5uZ49dVXIZfLsXLlSowbNw6bNm1Cd3c3HnnkESxcuBA9PT0YMWIEysrKsGzZMvzxj39Ec3MzbGxs8NhjjyE1NRUeHh6wtraGqakpTExMcOrUKVRUVGDkyJFwdHSEjY0NTp48ieXLl+O7/R8I2UPCGtFoNBwzZgxHjBihfbrr6OhQCMGJEyfy0UcfpY2NDV1dXZmcnMwFCxbQwsKCTk5OLC8vZ05ODouKijhv3jwmJiYyPz+fHh4eHD16NH18fFhSUsKqqioeOHCAs2bNokKhoFqtZnNzMzs7O9nb28vGxkY+++yzTE1NpUKh4KVLl+jp6cmQkBA6OjrypZde4tWrV7WexJ9jjQwJsm8kvaWlhZ2dnayrq2NDQwPVajVzc3N59epV+vj40MTEhJ9++imrq6vp7u5OOzs7pqSksLy8nBEREXRycuLy5cuZl5dHf39/uri4MDY2lsXFxWxpaWFTUxN7e3u5a9cu6urqcufOnfTx8WFLSwtrampYX19PtVrN3t5e1tXVMT8/nz09PUxISCAAKpVKCiG0roPfJNlVVVUsKyvTdl5XV5cAGB4ezkmTJtHIyIiWlpbct28fCwsLaWVlxTFjxjAnJ4dqtZoZGRn88MMPuXTpUq5cuZL29vacNGkS5XI56+rqeOnSJdbW1vLAgQPcvXs33d3duXjxYpaWlpLsm3WSZH19PQ8ePMiYmBi6ubnR0tJS66q1srLiunXrOHnyZK38QMgeMjr7kUcegZeXF5YsWYLW1lbcf//9CA0NxSeffILt27dj1KhROH/+POLj43H//fejoKAAQgi8/fbbkMlkWLx4sdZCeOSRR2Bra4tTp07hL3/5C2praxEVFYUHH3wQixYtQkdHB5qbmyGXy5GXl4fLly/j8uXL2L59OzIyMmBtbQ1XV1cIITB69Gi8+OKL2Lp1Kzo7O2FqanrTwvJAdPYtBekIISrQF86rBtBL0k0IYQYgBn1B6xUA5rMvmHzAaGtrw4IFCyCRSODu7q5d/QaARYsWITIyEiRhbm4OIQSKi4vh6emJK1euIC4uDtOmTYOhoSGOHDmCBQsWYNq0acjLy0NTUxNMTU1hZ2cHpVKJ8+fP491338U333wDGxsbTJw4EQsXLtQeb21tRV1dHfT19bFmzRro6ekhICAAISEh0NfXR2BgIGJjY9Ha2vpzTnNAEVGeJOtvSK8D8BXJUCHEuv702oF2YNasWVAqlXj00UdRX1+PN954A25ublCr1ejo6MCuXbugp6eH/Px8PPTQQ3jggQfw8ccfw87ODmPHjkV4eDjKysqwbds26OrqwsTEBJmZmfD19cXTTz+NrVu3giRkMhnmzJkDXV1dWFhYoLe3F3V1dfDz84ODgwMMDQ0xf/58+Pr6oqenB56enrCzs8OWLVswY8YMyGQydHV1wdvbG11dXdDX1//pk/sObkmN9I9stxvJFkIUo2+b2zUhxEgAqSQdf6ye71MjW7ZsgYuLC9zd3dHS0gJbW1sAwMsvvwy1Wg17e3vs3r0bq1evRm1tLfbu3Yu//vWvqKioQHFxMebPnw8vLy90d3fD3t5eO/JdXFxw5MgR6OvrY8qUKfjmm2+go6OD9957D6WlpfDz88Ojjz6K0tJSuLi4wMTEBGPGjEFWVhaOHTsGY2Nj2NraYvHixVi8eDFiY2ORmZmJyZMnIy0tDY888ggkEsmdt7OFEOUAmtD3MPiQZLgQopnksP7jAkDT9fR3yt4Y6zepvr4evb29kEj6Qlba29sxa9YsVFVVob29HSqVCllZWbh48SIuX74MS0tLhIeHw8DAAL6+vvjqq6/wpz/9CUZGRvDw8ICpqSk6OjpQV1eHrKwsxMXF4bPPPoOhoSHCwsJw9OhRmJqawsbGBk888QQ8PT3R1NSEtrY21NbWYvr06UhKSsJHH32El156CefPn8f06dNhYmKC++67D+7u7sjMzMSFCxcwZ84cfPvttzAxMel74Alx5+1sANb931bo2xf4KIDm78g0DcQa8fX1pbm5OQFw9uzZvHTpEuvq6rQev9raWi5atIgbN25kbGwsp06dysDAQHZ0dLCmpoaVlZWMioqii4sLAwMDGRQUxPj4eJaUlHD8+PE8ePAgq6ur2dzczOLiYk6bNo3e3t60tbXl3LlzmZOTQ5VKRZVKxZKSEmZnZ7OmpoZz587l9u3b+eyzz1KhUFAqlXLmzJlsbGxkTk4OV69erT0HpVL5y5p+6NtuvBpAMYCR/XkjARQPhGyJREIA3LJlC7/99luq1WpevXqVs2fPZkhICC0tLenn58fw8HCOGjWKubm57O7u5tWrV6lQKJiRkcF9+/bx4sWLPH/+PE1NTZmRkcGgoCB2d3eztraW69at47fffss5c+Zw3LhxDAoKYlNTE5VKJVtbW0mSmZmZdHBwoIWFBXt7e/nuu+/SxcWFEomEGo2GkydPphCCZmZm1Gg0LCws5PXzUKlUd5ZsAEYATG74nQFgDoDNANb1568D8M5AyFar1bx27RozMzNpZGTEESNGcPTo0Zw4cSJff/11urm50czMjLt27WJQUBB37NjBmpoalpWVUaVSsbOzU7va0trayqSkJKalpdHb25tPP/00161bx4MHD2oXB+RyORsbG+ng4KBdTFCpVKyrq2NQUBB9fHx45MgRkn2TqyeeeIJubm6srq5meno6pVIpt27dSo1Gw6eeeoqmpqYk77CdLYSwBxDXn9QB8CnJECGEOYCDAGwBVKLP9Gv8sbrGjh3LM2fOoKenBy4uLjA2Noa5uTkiIyNRWFiIe+65B6tWrYJSqURERAT09PQQHByM2NhYqFQqNDY2Yvjw4di5cycMDQ0REBCAyspKXLt2Debm5jAyMsKDDz6IpqYmJCcnIzs7GxEREXj44YexdOlSPPDAA6isrERycjLOnTuH3bt3Q1dXF4WFhWhvb8fDDz8MMzMzNDY2YtiwYaisrISDg4OWLB0dHeTm5mLixIl45plnoK+vj08++eS6Oh6ajqj169ejq6sLTk5OsLa2xoIFC9Da2oqHHnoIM2fOREhICORyOSwsLPDFF1/g5MmTeOWVV0ASZmZmqK2txbhx41BaWor77rsP6enp2LFjB6qrq/Hiiy9CoVBg3rx50NPTg5GREc6cOQM/Pz8kJydj+PDh+Oyzz2Bvbw+1Wo3U1FRUVFRAT08Ppqam6Onpwe7du/HnP/8ZZWVlCAkJQWxsLFJTU5Gfn48ZM2agpqYGVlZWeO+997B8+XIAQ5jsN998E8HBwVi1ahVOnz6N3NxcPP/883B3d8dDDz0EIQTi4+NhZWUFCwsLjB8/HuPGjQNJREdHo729Hb6+vgD6woTnzJkDX19faDQajBw5EnPmzNG6QC9duoTDhw/jmWeegYWFBfbv34/q6mqMHj0ajz32GM6ePQupVApDQ0N0d3fjL3/5C7q7u+Hv74/77rsPK1euhIGBAVauXIkNGzbAxMQEQUFBePPNN6Grqzu0vX5qtZqJiYl0cnLSPiCTk5M5evRoCiF48uRJ1tfXMzc3lwUFBUxLS+PZs2epVqvZ3d3NsLAwAuDUqVMplUp56NAhVlVVaS2LS5cuUaFQUKVSccOGDZwwYQLr6+vZ3NxMAJw3bx5bW1vZ3t5OPz8/ajQaqlQqnjlzhr6+vjx8+DB9fHy08nK5nBEREVywYAErKyuZkJBAiUTCp556asA6e9DJvr4Snp6ezq+//prOzs4EwIiICKamprK7u5sJCQncuXMnN2/erHVITZgwga6urpw8eTKnT5/OLVu2sLq6moGBgQRANzc3TpkyhRqNhsXFxdpV+pSUFG7cuJFPPvkkOzs7GRUVRRsbGyqVSpaWlnLp0qXUaDTUaDSUSCR0dnbm8ePHKYSgQqEgANbX13PlypUEwM7OTurp6REAZ86cObTJbm1tZWNjI5cvX64NS/D29qZMJqNMJqOuri43bdrEkpISFhcXMysri6mpqVQqlWxsbGRRURGzsrIIgFKplBcuXGBvby8rKytpY2NDIyMjLdmjRo2iiYmJ1q0qk8lYWFhItVpNOzs77tmzh1VVVdTV1aWZmRnr6+tZWlqqJbOjo4Nnz56lqakpz5w5w+HDh9PV1ZWNjY3cvHkz5XL50CabJD/99FMKIfjCCy8wPj6eACiEoIWFBfPy8hgUFMS3335bGy8ihGB1dTXff/996unpMSMjgxcvXmRCQgKFECwoKGBjY6N2kiGEoFKp5OXLl2liYsLq6mp2dXXR2NiYQghqNBrOnj2benp6JKmNSZFIJFpV9NJLLzEvL09rSxsbGzMuLk47um1sbIa+GlGpVLz33nspl8u1kakLFy5keXk5p0yZwmHDhtHHx4fV1dXMyMhgYGAglUollUolS0pKOHfuXLq6umpPMi0tja2trTQyMuKmTZt46dIlvvnmmwTAZcuWUalU0sDAgPr6+qyqquLIkSOZl5fHlpYWrl+/XjvaY2Nj2dTURAAsLy9nS0uLVmf7+PhQKpWyvb2dDg4OlEgkbGlpufOTmjtNdm1tLRsbG2lgYEAdHR2GhobS09OTzs7OXLx4MXfv3s033niDtra2NDAwoKurK5VKJXV0dAiAS5YsYU5ODgsLC7lp0ya6uLjwgw8+YFNTE9euXUuJRMKoqCieOHGC8fHxnDBhAltaWpiTk0MALC0tpbu7OwGwubmZEydOZG1tLdevX8/g4GDtANi7dy/PnDmj1dkKhYJbt25ld3c3PTw8OHXqVDY2Ng5tsjUaDUtKSrS+jKeeeooXL17k3r17aW9vTycnJ+bk5DAtLY0XL17khg0buHnzZiYlJbGyspL5+fmcOXMmJRIJ33jjDTY3N3PMmDGUSqVctmwZ6+rq6ODgQEdHRyYnJ2vXEauqqjhlyhTq6emxurqawcHB9PPzY2NjI0tKSvjKK6/Q0dGRq1at4vr16+nk5MT09HQ6OTnR39+fRUVF1NHRYUVFBd955x2+8MIL3Lhx49AmW61WMzs7m3K5nN7e3jQwMGBGRgadnJw4f/587ty5k//4xz/o6+vLzMxM6urq0sDAgN7e3gwNDeWaNWv48ccf09DQkFZWVuzs7OT27dsphOCoUaPY2NjIXbt2EQB37NjBr7/+mkIIbtiwgYmJibS1tWVaWhrr6+tpZWXFoqIiRkVFEQDfffddAuDZs2fp6enJZ599ljt37uTo0aO5bt06urq6MigoiObm5gwKCqKpqenQJrulpYWlpaWcNGkShRDcu3cvhRD08/Pj0aNHCYCurq7cvn078/PzeeDAAUZFRfHw4cNMSEjgwoULCYA+Pj7cv38/U1NTOX/+fB4+fJjvv/8+CwsL2dTUxC1btmjVxoYNG/jmm29y4cKFjI6O5ogRI3j8+HEeOnSIEomECoWC4eHh9PHx4Zo1ayiEoKenJz/77DNGRUVx1KhRBMDz589TV1eXV69epUwmY319/dAmu6Ojg7m5uTQ2NuaRI0colUoZHBxMS0tLAmBcXByDg4NpZ2enjcmeMWMGKysrWV5eToVCwYSEBA4bNoze3t7cs2cPly5dymHDhnHp0qUMDw+njY0NU1JSGBsbS11dXaanp9PR0ZH/+te/WFpaqrU+0tPTCYDbtm1jWloajY2Nqaury9DQUEZGRvKZZ56hnZ0dCwsLOX36dB46dIinT59mQkICS0tLWVJSMrTJrqurY0FBAYuKiujs7Mwnn3ySXl5enDlzJvPy8mhsbMypU6fys88+48aNG2lubs4pU6YwIyODsbGxfP311wmAa9asYUFBAQFo1QgAXrhwQbs6HxYWxujoaEqlUp48eZIBAQFcsmQJs7KyuGjRIurq6vL8+fM8ceIEc3JyKJPJ6OfnRyEE//Of/xAAi4qKuGLFCsrlchobG7Ouro7btm1jdnY2SQ5tsjs6Opifn6+1o48fP04AzMjI4Mcff0yJRMJDhw4xMjKSQgja2tpSIpHQw8OD06ZN0+YD4MmTJ+ns7EwHBwdGRUXRzs6OaWlpDA0N1ZJQXl7OrVu3sqioSBv7ERkZqZ0UFRQUMDU1lXPnzqW3tzdNTEwIgMOGDaOzszOPHTvGo0eP0s/Pj2fOnGFpaak2rmXIm36VlZXMzMzkl19+yaNHj1IqlfLf//43AwICtKsu5ubmnD17NsvKypiVlcWIiAjOmDGDI0aMoL6+Pr/44gsaGRkRAGNiYvj8889rZ3l2dnZ88sknOXfuXB46dIjHjx9nSUkJa2trGRMTQysrK3p4eHD//v3aC52ens6RI0dy6tSpnDx5Ml977TV6eXnxww8/pEQi4YkTJ5iYmMiAgADu3LmTNTU1VCqVfP/994c22Z2dnfzqq6+YnJzMdevW0cPDg15eXnzttddoaWnJCRMm0M7OjoaGhoyNjWVoaCgXLFhAQ0NDCiE4e/ZsAmBycjLnzZtHAJwyZQqDgoJoaGjIPXv2EACzsrKYm5vL0NBQyuVy7tq1i/Pnz2diYiLvueceGhkZsaCggG5ubly2bBkTExO1d5i9vT2zs7M5fPhwpqWlUQjBkJAQurq6sri4WLvfp66ubmgH6RQWFqKrqwsVFRXYsmULxo4di7feegu2trbo7OyEn58fzMzMMG3aNMTHx0NfXx8ymQwdHR24fPkyUlJS8Pe//x2ff/45vLy8YG5uDplMhrCwMMTExMDU1BRhYWHQ0dHB008/jdjYWKxYsQKzZs3CgQMHEB0djeeffx5nz57F4cOHsW/fPnh6emLs2LF48sknUV5eDmdnZ+zcuRNbtmxBQEAAVq5ciT179uDUqVOQSqUYN24cWltboa+vD0PDvjeY8hZcrIP+EkUhBK5duwYjIyO8+uqrsLa2xqhRo3D58mU8/fTT2LNnD/z8/HDgwAF88skn2Lt3L3bs2IF9+/aht7cXf/rTn3DlyhUoFAp0dHRog3U6Ozuxd+9exMTEYMmSJUhISEBXVxdOnz6N2tpaWFtb4/jx4wgMDAQAGBsbo6SkBMXFxfDy8kJRUREef/xxVFZWYvz48VAqlRg1ahTMzMzg6+uLefPmoa2tDfHx8ZBIJDAwMICent7ATv7XeEA2NDSwvLycdXV1TElJoVwuZ0pKCh9//HHu3buXDz30EIODgxkWFsZz584xIiKCADhu3DjGxsZSJpNx69atBMDExEQ+99xz9PX1pZeXFxctWkRzc3NGRkbSzs6O8+fPp1Qq5RtvvEE/Pz+ePn2a+/bt40cffaRVXbm5uXR2dubUqVMZHh7OYcOGcf/+/QTApKQknj59mikpKVyyZAmLioqYn5/Pzs5OVldXD1010m9F/C4xJNXI/zJ+8beffQdt6Is3GSxYAKj/Sanbx5hbERpssotJug1WY0KInMFs76dwV40MIu6SPYgYbLLDf1rkN93ej2JQTb//ddxVI4MyX+eXAAABPklEQVSIu2QPIgaNbCHEHCFEsRDiUv8enF+ijQohhFwIkS+EyOnPMxNCJAkhSvq/h/8Sbd8SBsMngr7X9JcCsEffn/0UAJjwC7RTAcDiO3nv4OY48rDB9Afd+Bmskf1rvkX+KQD7+3/vR98fwP0qGCyyv+8t8ta/QDsEcFIIca5/4xQA3EvyWv/vagD3/gLt3hIGe7r+S2M6yatCCCsASUKIm/5UiNTG9f0qGKyRfUtvkb9dkLza/12Lvq0p7gBq+vdpov+79k63e6sYLLJ/8bfICyGMhBAm138D8ALwTX87S/rFlgCIv5PtDgSDokZI9gohrr9FXgoggnf+LfL3Aojr33ZxfaNVohAiG8BBIcTz6N9odYfbvWXcna4PIu7OIAcRd8keRNwlexBxl+xBxF2yBxF3yR5E3CV7EPH/AcD96RjuCkQYAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8b4136eb00>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img_list = sorted(glob.glob('Altered/Altered-Hard/*.BMP'))\n",
    "print(len(img_list))\n",
    "\n",
    "imgs = np.empty((len(img_list), 96, 96), dtype=np.uint8)\n",
    "labels = np.empty((len(img_list), 4), dtype=np.uint16)\n",
    "\n",
    "for i, img_path in enumerate(img_list):\n",
    "    img = cv2.imread(img_path, cv2.IMREAD_GRAYSCALE)\n",
    "    img = cv2.resize(img, (96, 96))\n",
    "    imgs[i] = img\n",
    "    \n",
    "    # subject_id, gender, lr, finger\n",
    "    labels[i] = extract_label2(img_path)\n",
    "    \n",
    "np.save('dataset/x_hard.npy', imgs)\n",
    "np.save('dataset/y_hard.npy', labels)\n",
    "\n",
    "plt.figure(figsize=(1, 1))\n",
    "plt.title(labels[-1])\n",
    "plt.imshow(imgs[-1], cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tf",
   "language": "python",
   "name": "tf"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}