13_Convolution.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch \n",
    "import torch.nn as nn\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from scipy import ndimage, misc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`What is Convolution?  \n",
    "`  \n",
    "Convolution is a linear operation similar to a linear equation, dot product, or matrix multiplication. Convolution has several advantages for analyzing images. As discussed in the video, convolution preserves the relationship between elements, and it requires fewer parameters than other methods.  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can see the relationship between the different methods that you learned:\n",
    "\n",
    "$$linear \\ equation :y=wx+b$$\n",
    "$$linear\\ equation\\ with\\ multiple \\ variables \\ where \\ \\mathbf{x} \\ is \\ a \\ vector \\ \\mathbf{y}=\\mathbf{wx}+b$$\n",
    "$$ \\ matrix\\ multiplication \\ where \\ \\mathbf{X} \\ in \\ a \\ matrix \\ \\mathbf{y}=\\mathbf{wX}+\\mathbf{b} $$\n",
    "$$\\ convolution \\ where \\ \\mathbf{X} \\ and \\ \\mathbf{Y} \\ is \\ a \\ tensor \\  \\mathbf{Y}=\\mathbf{w}*\\mathbf{X}+\\mathbf{b}$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In convolution, the parameter <b>w</b> is called a kernel. You can perform convolution on images where you let the variable image denote the variable X and w denote the parameter."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a two_dimensional convolution object by using the constructor Conv2d, the parameter <code>in_channels</code> and <code>out_channels</code> will be used for this section, and the parameter kernel_size will be three."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "conv = nn.Conv2d(in_channels=1, out_channels=1, kernel_size=3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Because the parameters in <code>nn.Conv2d</code> are randomly initialized and learned through training, give them some values.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('weight',\n",
       "              tensor([[[[ 1.,  0., -1.],\n",
       "                        [ 2.,  0., -2.],\n",
       "                        [ 1.,  0., -1.]]]])),\n",
       "             ('bias', tensor([0.]))])"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv.state_dict()['weight'][0][0] = torch.tensor([[1.0,0.0,-1.0],[2.0,0.0,-2.0],[1.0,0.0,-1.0]])\n",
    "conv.state_dict()['bias'][0] = 0.0\n",
    "conv.state_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.]]]])"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# define dummy tensor to represent an image\n",
    "image=torch.zeros(1,1,5,5)\n",
    "image[0,0,:,2]=1\n",
    "image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[-4.,  0.,  4.],\n",
       "          [-4.,  0.,  4.],\n",
       "          [-4.,  0.,  4.]]]], grad_fn=<ConvolutionBackward0>)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "z = conv(image)\n",
    "z"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`Stride`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('weight',\n",
       "              tensor([[[[1., 1.],\n",
       "                        [1., 1.]]]])),\n",
       "             ('bias', tensor([0.]))])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv1 = nn.Conv2d(in_channels=1, out_channels=1,kernel_size=2,stride=2)\n",
    "\n",
    "conv1.state_dict()['weight'][0][0]=torch.tensor([[1.0,1.0],[1.0,1.0]])\n",
    "conv1.state_dict()['bias'][0]=0.0\n",
    "conv1.state_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[1., 1., 1., 1.],\n",
       "          [1., 1., 1., 1.],\n",
       "          [1., 1., 1., 1.],\n",
       "          [1., 1., 1., 1.]]]])"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "M=4\n",
    "image1=torch.ones(1,1,M,M)\n",
    "image1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[4., 4.],\n",
       "          [4., 4.]]]], grad_fn=<ConvolutionBackward0>)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "z1=conv1(image1)\n",
    "z1\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`Padding`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[1., 2.],\n",
       "          [2., 4.]]]], grad_fn=<ConvolutionBackward0>)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv2 = nn.Conv2d(in_channels=1, out_channels=1,kernel_size=2,stride=3,padding=1)\n",
    "\n",
    "conv2.state_dict()['weight'][0][0]=torch.tensor([[1.0,1.0],[1.0,1.0]])\n",
    "conv2.state_dict()['bias'][0]=0.0\n",
    "conv2.state_dict()\n",
    "z2=conv2(image1)\n",
    "z2\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`Activation Function`  \n",
    "Just like a neural network, we apply an activation function to the activation map.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('weight',\n",
       "              tensor([[[[ 1.,  0., -1.],\n",
       "                        [ 2.,  0., -2.],\n",
       "                        [ 1.,  0., -1.]]]])),\n",
       "             ('bias', tensor([0.]))])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# make kernel (w)\n",
    "conv = nn.Conv2d(in_channels=1, out_channels=1,kernel_size=3)\n",
    "Gx=torch.tensor([[1.0,0,-1.0],[2.0,0,-2.0],[1.0,0,-1.0]])\n",
    "conv.state_dict()['weight'][0][0]=Gx\n",
    "conv.state_dict()['bias'][0]=0.0\n",
    "conv.state_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.]]]])"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# dataset (x)\n",
    "image=torch.zeros(1,1,5,5)\n",
    "image[0,0,:,2]=1\n",
    "image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[-4.,  0.,  4.],\n",
       "          [-4.,  0.,  4.],\n",
       "          [-4.,  0.,  4.]]]], grad_fn=<ConvolutionBackward0>)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# apply kernel or convolution on dataset\n",
    "Z=conv(image)\n",
    "Z"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[0., 0., 4.],\n",
       "          [0., 0., 4.],\n",
       "          [0., 0., 4.]]]], grad_fn=<ReluBackward0>)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Apply the activation function to the activation map. \n",
    "relu = nn.ReLU()\n",
    "relu(Z)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[0., 0., 4.],\n",
       "          [0., 0., 4.],\n",
       "          [0., 0., 4.]]]], grad_fn=<ReluBackward0>)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# or in this way\n",
    "A=torch.relu(Z)\n",
    "A"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`Max Pooling`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[ 1.,  2.,  3., -4.],\n",
       "          [ 0.,  2., -3.,  0.],\n",
       "          [ 0.,  2.,  3.,  1.],\n",
       "          [ 0.,  0.,  0.,  0.]]]])"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "image1=torch.zeros(1,1,4,4)\n",
    "image1[0,0,0,:]=torch.tensor([1.0,2.0,3.0,-4.0])\n",
    "image1[0,0,1,:]=torch.tensor([0.0,2.0,-3.0,0.0])\n",
    "image1[0,0,2,:]=torch.tensor([0.0,2.0,3.0,1.0])\n",
    "\n",
    "image1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[2., 3., 3.],\n",
       "          [2., 3., 3.],\n",
       "          [2., 3., 3.]]]])"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "max1=torch.nn.MaxPool2d(2,stride=1)\n",
    "max1(image1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[2., 3.],\n",
       "          [2., 3.]]]])"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "max1=torch.nn.MaxPool2d(2)\n",
    "max1(image1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`Multiple input and Output Channels`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "conv1 = nn.Conv2d(in_channels=1, out_channels=3,kernel_size=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define kernel\n",
    "Gx=torch.tensor([[1.0,0,-1.0],[2.0,0,-2.0],[1.0,0.0,-1.0]])\n",
    "Gy=torch.tensor([[1.0,2.0,1.0],[0.0,0.0,0.0],[-1.0,-2.0,-1.0]])\n",
    "\n",
    "conv1.state_dict()['weight'][0][0]=Gx\n",
    "conv1.state_dict()['weight'][1][0]=Gy\n",
    "conv1.state_dict()['weight'][2][0]=torch.ones(3,3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([0., 0., 0.])"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv1.state_dict()['bias'][:]=torch.tensor([0.0,0.0,0.0])\n",
    "conv1.state_dict()['bias']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[[ 1.,  0., -1.],\n",
      "         [ 2.,  0., -2.],\n",
      "         [ 1.,  0., -1.]]])\n",
      "tensor([[[ 1.,  2.,  1.],\n",
      "         [ 0.,  0.,  0.],\n",
      "         [-1., -2., -1.]]])\n",
      "tensor([[[1., 1., 1.],\n",
      "         [1., 1., 1.],\n",
      "         [1., 1., 1.]]])\n"
     ]
    }
   ],
   "source": [
    "for x in conv1.state_dict()['weight']:\n",
    "    print(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.],\n",
       "          [0., 0., 1., 0., 0.]]]])"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# dataset\n",
    "image=torch.zeros(1,1,5,5)\n",
    "image[0,0,:,2]=1\n",
    "image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(image[0,0,:,:].numpy(), interpolation='nearest', cmap=plt.cm.gray)\n",
    "plt.colorbar()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([1, 3, 3, 3])"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out=conv1(image)\n",
    "out.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The result is a 1x3x3x3 tensor. This represents one sample with three channels, and each channel contains a 3x3 image.  The same rules that govern the shape of each image were discussed in the last section."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[-4.,  0.,  4.],\n",
      "        [-4.,  0.,  4.],\n",
      "        [-4.,  0.,  4.]], grad_fn=<UnbindBackward0>)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[0., 0., 0.],\n",
      "        [0., 0., 0.],\n",
      "        [0., 0., 0.]], grad_fn=<UnbindBackward0>)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[3., 3., 3.],\n",
      "        [3., 3., 3.],\n",
      "        [3., 3., 3.]], grad_fn=<UnbindBackward0>)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfwAAAGxCAYAAACHlAShAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAA9hAAAPYQGoP6dpAAAws0lEQVR4nO3df1RVdb7/8RcRv1I5jhiIiYKToWI4BiXHtFITgzL7sW7OVIgzWtr1xzLyzhW9d7RWDlZW2tVQV6bj+DMHf3U1bqyVoI4yCkE6pU7dNMhA01FQDPDH/v7Rl3M78fMAB4XP87HWXmv2Pp+99+fjXtOL92fvc7aHZVmWAABAm3bT9e4AAABwPwIfAAADEPgAABiAwAcAwAAEPgAABiDwAQAwAIEPAIABCHwAAAxA4AMAYAACH2ikVatWycPDQzk5Ode7K80iMzNTHh4eyszMrLNdbm6uJk+erDvvvFMdOnRQUFCQHnzwQX3yySct01EAjULgA3DJ+vXrdeDAAf3ud7/Ttm3b9N5778nHx0fDhw/X6tWrr3f3ANTi5uvdAQCty+9//3stWLDAaVt8fLzuuusuvfLKKxo7dux16hmAulDhA7U4evSofvOb3ygoKEg+Pj7q3r27xo4dq4qKCqd2Fy5c0AsvvKDOnTsrICBATzzxhL777junNhs3blRsbKyCg4Pl5+enPn36aObMmSorK3NqN27cOLVv315fffWV4uPj1b59e4WEhOill15yOu+JEyfk4eGhBQsW6K233lJYWJjat28vu92u7OzsamPJycnRo48+qk6dOsnX11cDBgzQBx980Kh/l8DAwGrbPD09FRUVpcLCwkYdE4D7EfhADT777DPdfffdys7O1iuvvKKPPvpIKSkpqqioUGVlpVPbCRMmyMvLS+vWrdPrr7+uzMxMPfvss05tvvzyS8XHx2vFihVKT0/X9OnT9cEHH2jUqFHVzn358mU9+uijGj58uLZt26bf/e53evvtt/Xaa69Va7tkyRJlZGRo4cKFWrt2rcrKyhQfH6+SkhJHm127dunee+/V+fPntXTpUm3btk2/+tWvNGbMGK1atapZ/r2uXLmiPXv2KCIiolmOB8ANLADVDBs2zOrYsaN1+vTpWtusXLnSkmT967/+q9P2119/3ZJkFRUV1bjftWvXrMuXL1tZWVmWJOuzzz5zfJaYmGhJsj744AOnfeLj463w8HDH+vHjxy1J1p133mlduXLFsf3AgQOWJGv9+vWObb1797YGDBhgXb582emYjzzyiBUcHGxdvXrVsizL2rVrlyXJ2rVrV61jrs3s2bMtSdbWrVtd3hdAy6DCB37m0qVLysrK0lNPPaVbb7213vaPPvqo03pkZKQk6ZtvvnFs+/rrr/X000+rS5cu8vT0lJeXl+6//35J0pEjR5z29/DwqFb5R0ZGOh2vysMPPyxPT89az/3VV1/p6NGjeuaZZyT9WIlXLfHx8SoqKtKxY8fqHWNd3nvvPc2bN08vvfSSRo8e3aRjAXAfHtoDfubcuXO6evWqunXr1qD2AQEBTus+Pj6SpB9++EGSdPHiRQ0ZMkS+vr569dVXdccdd+iWW25RYWGhnnjiCUe7Krfccot8fX2rHbO8vNzlc586dUqSNGPGDM2YMaPG/p85c6ZB46zJypUrNXHiRD3//PN64403Gn0cAO5H4AM/06lTJ3l6eurbb79tluN98skn+u6775SZmemo6iXp/PnzzXL8unTu3FmSlJycrCeeeKLGNuHh4Y069sqVKzVhwgQlJiZq6dKl8vDwaHQ/AbgfgQ/8jJ+fn+6//35t2rRJ8+bNc4RmY1UFYVX1XWXZsmVNOm5DhIeHq1evXvrss8/0xz/+sdmOu2rVKk2YMEHPPvus3nvvPcIeaAUIfKAGb731lgYPHqyBAwdq5syZuv3223Xq1Clt375dy5YtU4cOHRp8rEGDBukXv/iFJk2apDlz5sjLy0tr167VZ5995sYR/J9ly5YpLi5OI0eO1Lhx43Tbbbfpn//8p44cOaJPP/1UmzZtcul4mzZt0vjx4/WrX/1KEydO1IEDB5w+HzBgQLU/bgBcfwQ+UIP+/fvrwIEDmjNnjpKTk3XhwgV16dJFw4YNk7e3t0vHCggI0I4dO/TSSy/p2WefVbt27TR69Ght3LhRd911l5tG8H+GDh2qAwcOaN68eZo+fbrOnTungIAA9e3bV0899ZTLx9uxY4euXbumTz/9VPfee2+1z48fP67Q0NBm6DmA5uRhWZZ1vTsBAADci6/lAQBgAAIfAAADEPgAABjArYF/7tw5JSQkyGazyWazKSEhod7vHo8bN04eHh5OS0xMjDu7CQBAi0hNTVVkZKT8/f3l7+8vu92ujz76qNb2e/fu1b333quAgAD5+fmpd+/eevvttxt1brc+pf/000/r22+/VXp6uiTp+eefV0JCgj788MM693vooYe0cuVKx7qrT0UDAHAj6tatm+bPn6/bb79dkvSnP/1Jo0ePVl5eXo0vn2rXrp2mTJmiyMhItWvXTnv37tXEiRPVrl07Pf/88y6d221P6R85ckR9+/ZVdna2Bg4cKEnKzs6W3W7X0aNHa/11r3Hjxun8+fPaunWrO7oFAMANpVOnTnrjjTc0fvz4BrV/4okn1K5dO/35z3926Txuq/D3798vm83mCHtJiomJkc1m0759++r8Oc/MzEwFBgaqY8eOuv/++zVv3rwa38EtSRUVFU7vCb927Zr++c9/KiAggF//AoBWyLIsXbhwQV27dtVNN7nvznN5eXm11103hmVZ1fLGx8en3h+gunr1qjZt2qSysjLZ7fYGnSsvL0/79u3Tq6++2qiOusW8efOsXr16Vdveq1cv649//GOt+23YsMH67//+b+vw4cPW9u3brf79+1sRERFWeXl5je3nzJljSWJhYWFhaWNLYWFhs2XSz/3www9Wly5dmqWf7du3r7Ztzpw5tZ770KFDVrt27SxPT0/LZrNZO3bsqLe/t912m+Xt7W3ddNNN1iuvvNKoMbtc4c+dO1cvv/xynW0OHjwoSTVW2FYNfwn91JgxYxz/u1+/foqOjlaPHj20Y8eOGl/+kZycrKSkJMd6SUmJunfvrhdffJGf9wSAVqiiokJvv/22Sz9h7arKykoVFxeroKBA/v7+jT5OaWmpunfvrsLCQqfj1JU/4eHhys/P1/nz55WWlqbExERlZWWpb9++te6zZ88eXbx4UdnZ2Y6f+/7Nb37jUl9dDvwpU6bo17/+dZ1tQkNDdejQIcerOX/q+++/V1BQUIPPFxwcrB49eujLL7+s8fPapk18fHyqvWIUANB6tMRt2aqn5VvyON7e3o6H9qKjo3Xw4EEtWrSozhdqhYWFSZLuvPNOnTp1SnPnznV/4Hfu3LlBbw+z2+0qKSnRgQMHdM8990iS/va3v6mkpESDBg1q8PnOnj2rwsJCBQcHu9pVAADqZFmWrCY8u96UfX96jJ8+i9bc7au47aG9Pn366KGHHtJzzz3n+Kvl+eef1yOPPOL0wF7v3r2VkpKixx9/XBcvXtTcuXP15JNPKjg4WCdOnNCsWbPUuXNnPf744+7qKgDAUC0d+LNmzVJcXJxCQkJ04cIFbdiwQZmZmY6vrycnJ+vkyZNavXq1JGnJkiXq3r27evfuLenH7+UvWLBAU6dOdbmvbv0e/tq1azVt2jTFxsZKkh599FEtXrzYqc2xY8dUUlIiSfL09NThw4e1evVqnT9/XsHBwRo6dKg2btzo1ns5AAAztXTgnzp1SgkJCSoqKpLNZlNkZKTS09M1YsQISVJRUZEKCgoc7a9du6bk5GQdP35cN998s375y19q/vz5mjhxost9bXNvyystLZXNZtPMmTO5hw8ArVB5ebnmz5+vkpKSZrm/XpOqrDh79myTH9oLCAhwa1+bi1srfAAAbmQ3wj38lkLgAwCMZVLg87Y8AAAMQIUPADCWSRU+gQ8AMJZJgc+UPgAABqDCBwAYy6QKn8AHABjLpMBnSh8AAANQ4QMAjGVShU/gAwCMReADAGAAkwKfe/gAABiACh8AYCyTKnwCHwBgLJMCnyl9AAAMQIUPADCWSRU+gQ8AMJZJgc+UPgAABqDCBwAYy6QKn8AHABitNYV2UzClDwCAAajwAQDGYkofAAADEPgAABjApMDnHj4AAAagwgcAGMukCp/ABwAYy6TAZ0ofAAADUOEDAIxlUoVP4AMAjGVS4DOlDwCAAajwAQDGMqnCJ/ABAMYyKfCZ0gcAwABU+AAAY5lU4RP4AABjEfgAABjApMDnHj4AAAagwgcAGMukCp/ABwAYy6TAZ0ofAAADUOEDAIxlUoVP4AMAjGVS4DOlDwCAAajwAQDGMqnCJ/ABAEZrTaHdFEzpAwBgACp8AICxmNIHAMAAJgV+i0zpv/vuuwoLC5Ovr6+ioqK0Z8+eOttnZWUpKipKvr6+6tmzp5YuXdoS3QQAGKYq8JuyuCI1NVWRkZHy9/eXv7+/7Ha7Pvroo1rbb968WSNGjNCtt97qaP8///M/jRqr2wN/48aNmj59umbPnq28vDwNGTJEcXFxKigoqLH98ePHFR8fryFDhigvL0+zZs3StGnTlJaW5u6uAgDgVt26ddP8+fOVk5OjnJwcDRs2TKNHj9bnn39eY/vdu3drxIgR2rlzp3JzczV06FCNGjVKeXl5Lp/bw3LzfMTAgQN11113KTU11bGtT58+euyxx5SSklKt/b//+79r+/btOnLkiGPbpEmT9Nlnn2n//v31nq+0tFQ2m00zZ86Ur69v8wwCANBiysvLNX/+fJWUlMjf398t56jKigMHDqh9+/aNPs7Fixd1zz33NKmvnTp10htvvKHx48c3qH1ERITGjBmjP/zhDy6dx60VfmVlpXJzcxUbG+u0PTY2Vvv27atxn/3791drP3LkSOXk5Ojy5cvV2ldUVKi0tNRpAQCgIZprSv/nOVRRUVHvua9evaoNGzaorKxMdru9Qf29du2aLly4oE6dOrk8VrcG/pkzZ3T16lUFBQU5bQ8KClJxcXGN+xQXF9fY/sqVKzpz5ky19ikpKbLZbI4lJCSk+QYAAEADhISEOGVRTTPYVQ4fPqz27dvLx8dHkyZN0pYtW9S3b98GnefNN99UWVmZnnrqKZf72CJP6Xt4eDitW5ZVbVt97WvaLknJyclKSkpyrJeWlhL6AIAGaa6n9AsLC52m9H18fGrdJzw8XPn5+Tp//rzS0tKUmJiorKysekN//fr1mjt3rrZt26bAwECX++rWwO/cubM8PT2rVfOnT5+uVsVX6dKlS43tb775ZgUEBFRr7+PjU+c/LAAAtWmuwK966r4hvL29dfvtt0uSoqOjdfDgQS1atEjLli2rdZ+NGzdq/Pjx2rRpkx588MFG9dWtU/re3t6KiopSRkaG0/aMjAwNGjSoxn3sdnu19h9//LGio6Pl5eXltr4CAHA9WJZV5z3/9evXa9y4cVq3bp0efvjhRp/H7VP6SUlJSkhIUHR0tOx2u5YvX66CggJNmjRJ0o9T8idPntTq1asl/fhE/uLFi5WUlKTnnntO+/fv14oVK7R+/Xp3dxUAYJiW/uGdWbNmKS4uTiEhIbpw4YI2bNigzMxMpaenS6qeievXr9fYsWO1aNEixcTEOGbA/fz8ZLPZXDq32wN/zJgxOnv2rF555RUVFRWpX79+2rlzp3r06CFJKioqcvpOflhYmHbu3KkXX3xRS5YsUdeuXfXOO+/oySefdHdXAQCGaenAP3XqlBISElRUVCSbzabIyEilp6drxIgRkqpn4rJly3TlyhVNnjxZkydPdmxPTEzUqlWrXDq327+H39L4Hj4AtG4t+T38v/71r03+Hv69997r1r42F35LHwBgLJN+S5/ABwAYi8AHAMAAJgV+i7wtDwAAXF9U+AAAY5lU4RP4AABjmRT4TOkDAGAAKnwAgLFMqvAJfACAsUwKfKb0AQAwABU+AMBYJlX4BD4AwGitKbSbgil9AAAMQIUPADAWU/oAABiAwAcAwAAmBT738AEAMAAVPgDAWCZV+AQ+AMBYJgU+U/oAABiACh8AYCyTKnwCHwBgLJMCnyl9AAAMQIUPADCWSRU+gQ8AMJZJgc+UPgAABqDCBwAYy6QKn8AHABiLwAcAwAAmBT738AEAMAAVPgDAWCZV+AQ+AMBYJgU+U/oAABiACh8AYCyTKnwCHwBgLJMCnyl9AAAMQIUPADCWSRU+gQ8AMFprCu2mYEofAAADUOEDAIzFlD4AAAYg8AEAMIBJgc89fAAADECFDwAwlkkVPoEPADCWSYHPlD4AAAagwgcAGMukCp/ABwAYy6TAZ0ofAAADUOEDAIxlUoVP4AMAjGVS4LfIlP67776rsLAw+fr6KioqSnv27Km1bWZmpjw8PKotR48ebYmuAgDgNqmpqYqMjJS/v7/8/f1lt9v10Ucf1dq+qKhITz/9tMLDw3XTTTdp+vTpjT632wN/48aNmj59umbPnq28vDwNGTJEcXFxKigoqHO/Y8eOqaioyLH06tXL3V0FABimqsJvyuKKbt26af78+crJyVFOTo6GDRum0aNH6/PPP6+xfUVFhW699VbNnj1b/fv3b9JY3R74b731lsaPH68JEyaoT58+WrhwoUJCQpSamlrnfoGBgerSpYtj8fT0dHdXAQCGaenAHzVqlOLj43XHHXfojjvu0Lx589S+fXtlZ2fX2D40NFSLFi3S2LFjZbPZmjRWtwZ+ZWWlcnNzFRsb67Q9NjZW+/btq3PfAQMGKDg4WMOHD9euXbtqbVdRUaHS0lKnBQCAhmiuwP95DlVUVNR77qtXr2rDhg0qKyuT3W5391DdG/hnzpzR1atXFRQU5LQ9KChIxcXFNe4THBys5cuXKy0tTZs3b1Z4eLiGDx+u3bt319g+JSVFNpvNsYSEhDT7OAAAqEtISIhTFqWkpNTa9vDhw2rfvr18fHw0adIkbdmyRX379nV7H1vkKX0PDw+ndcuyqm2rEh4ervDwcMe63W5XYWGhFixYoPvuu69a++TkZCUlJTnWS0tLCX0AQIM011P6hYWF8vf3d2z38fGpdZ/w8HDl5+fr/PnzSktLU2JiorKystwe+m4N/M6dO8vT07NaNX/69OlqVX9dYmJitGbNmho/8/HxqfMfFgCA2jRX4Fc9dd8Q3t7euv322yVJ0dHROnjwoBYtWqRly5Y1uh8N4dYpfW9vb0VFRSkjI8Npe0ZGhgYNGtTg4+Tl5Sk4OLi5uwcAwHVnWVaD7vk3ldun9JOSkpSQkKDo6GjZ7XYtX75cBQUFmjRpkqQfp+RPnjyp1atXS5IWLlyo0NBQRUREqLKyUmvWrFFaWprS0tLc3VUAgGFa+od3Zs2apbi4OIWEhOjChQvasGGDMjMzlZ6eLql6JkpSfn6+JOnixYv6/vvvlZ+fL29vb5dvAbg98MeMGaOzZ8/qlVdeUVFRkfr166edO3eqR48ekn78UYGffie/srJSM2bM0MmTJ+Xn56eIiAjt2LFD8fHx7u4qAMAwLR34p06dUkJCgoqKimSz2RQZGan09HSNGDFCUvVMlH781lqV3NxcrVu3Tj169NCJEydcOreH1Zp+F7ABSktLZbPZNHPmTPn6+l7v7gAAXFReXq758+erpKSkwffFXVWVFe+9955uueWWRh/n0qVLmjBhglv72lz4LX0AgLFM+i19Ah8AYCyTAr9FXp4DAACuLyp8AIDRWlOV3hQEPgDAWCZN6RP4AABjmRT43MMHAMAAVPgAAGOZVOET+AAAY5kU+EzpAwBgACp8AICxTKrwCXwAgLFMCnym9AEAMAAVPgDAWCZV+AQ+AMBYBD4AAAYwKfC5hw8AgAGo8AEAxjKpwifwAQDGMinwmdIHAMAAVPgAAGOZVOET+AAAY5kU+EzpAwBgACp8AICxTKrwCXwAgLFMCnym9AEAMAAVPgDAWCZV+AQ+AMBYBD4AAIZoTaHdFNzDBwDAAFT4AABjMaUPAIABTAp8pvQBADAAFT4AwFgmVfgEPgDAWCYFPlP6AAAYgAofAGAskyp8Ah8AYCyTAp8pfQAADECFDwAwlkkVPoEPADAWgQ8AgAFMCnzu4QMAYAAqfACAsUyq8Al8AICxTAp8pvQBADAAFT4AwFgmVfgEPgDAWCYFPlP6AAAYgAofAGAskyp8Ah8AYCyTAt+tU/q7d+/WqFGj1LVrV3l4eGjr1q317pOVlaWoqCj5+vqqZ8+eWrp0qTu7CABAi0lNTVVkZKT8/f3l7+8vu92ujz76qM59misX3Rr4ZWVl6t+/vxYvXtyg9sePH1d8fLyGDBmivLw8zZo1S9OmTVNaWpo7uwkAMFRVhd+UxRXdunXT/PnzlZOTo5ycHA0bNkyjR4/W559/XmP75sxFt07px8XFKS4ursHtly5dqu7du2vhwoWSpD59+ignJ0cLFizQk08+WeM+FRUVqqiocKyXlpY2qc8AAHM015T+z7PHx8dHPj4+1dqPGjXKaX3evHlKTU1Vdna2IiIiqrVvTC7W5oZ6Sn///v2KjY112jZy5Ejl5OTo8uXLNe6TkpIim83mWEJCQlqiqwCANqI5qvuQkBCnLEpJSan3vFevXtWGDRtUVlYmu91eY5vG5GJtbqiH9oqLixUUFOS0LSgoSFeuXNGZM2cUHBxcbZ/k5GQlJSU51ktLSwl9AECLKiwslL+/v2O9puq+yuHDh2W321VeXq727dtry5Yt6tu3b41tG5OLtbmhAl+SPDw8nNar/oL6+fYqtU2bAABQn+aa0q96CK8hwsPDlZ+fr/PnzystLU2JiYnKysqqNfRdzcXa3FCB36VLFxUXFzttO336tG6++WYFBARcp14BANqq6/G1PG9vb91+++2SpOjoaB08eFCLFi3SsmXLqrVtzly8oe7h2+12ZWRkOG37+OOPFR0dLS8vr+vUKwAA3MeyLKeHz3+qOXPRrYF/8eJF5efnKz8/X9KPXy/Iz89XQUGBpB/vv48dO9bRftKkSfrmm2+UlJSkI0eO6P3339eKFSs0Y8YMd3YTAGColv5a3qxZs7Rnzx6dOHFChw8f1uzZs5WZmalnnnlGkntz0a1T+jk5ORo6dKhjverhusTERK1atUpFRUWO8JeksLAw7dy5Uy+++KKWLFmirl276p133nH5qwcAADRES0/pnzp1SgkJCSoqKpLNZlNkZKTS09M1YsQISXJrLro18B944IE6/zFWrVpVbdv999+vTz/91I29AgDg+lixYkWdn7szF2+oh/YAAGhJJv2WPoEPADCWSYF/Qz2lDwAA3IMKHwBgLJMqfAIfAGAsAh8AAAOYFPjcwwcAwABU+AAAY5lU4RP4AABjmRT4TOkDAGAAKnwAgLFMqvAJfACAsUwKfKb0AQAwABU+AMBYJlX4BD4AwFgmBT5T+gAAGIAKHwBgLJMqfAIfAGAsAh8AAEO0ptBuCu7hAwBgACp8AICxmNIHAMAAJgU+U/oAABiACh8AYCyTKnwCHwBgLJMCnyl9AAAMQIUPADCWSRU+gQ8AMJZJgc+UPgAABqDCBwAYy6QKn8AHABiLwAcAwAAmBT738AEAMAAVPgDAWCZV+AQ+AMBYJgU+U/oAABiACh8AYCyTKnwCHwBgLJMCnyl9AAAMQIUPADCWSRU+gQ8AMJZJgc+UPgAABqDCBwAYy6QKn8AHABiLwAcAwBCtKbSbgnv4AAAYgAofAGAspvQBADCASYHPlD4AAAagwgcAGIsKv5ns3r1bo0aNUteuXeXh4aGtW7fW2T4zM1MeHh7VlqNHj7qzmwAAQ1UFflMWV6SkpOjuu+9Whw4dFBgYqMcee0zHjh2rd78lS5aoT58+8vPzU3h4uFavXu3yWN0a+GVlZerfv78WL17s0n7Hjh1TUVGRY+nVq5ebeggAQMvJysrS5MmTlZ2drYyMDF25ckWxsbEqKyurdZ/U1FQlJydr7ty5+vzzz/Xyyy9r8uTJ+vDDD106t1un9OPi4hQXF+fyfoGBgerYsWPzdwgAgJ9o6Sn99PR0p/WVK1cqMDBQubm5uu+++2rc589//rMmTpyoMWPGSJJ69uyp7Oxsvfbaaxo1alSDz31DPrQ3YMAABQcHa/jw4dq1a1edbSsqKlRaWuq0AADQEM01pf/zHKqoqGjQ+UtKSiRJnTp1qrVNRUWFfH19nbb5+fnpwIEDunz5coPHekMFfnBwsJYvX660tDRt3rxZ4eHhGj58uHbv3l3rPikpKbLZbI4lJCSkBXsMAIAUEhLilEUpKSn17mNZlpKSkjR48GD169ev1nYjR47Ue++9p9zcXFmWpZycHL3//vu6fPmyzpw50+A+3lBP6YeHhys8PNyxbrfbVVhYqAULFtQ61ZGcnKykpCTHemlpKaEPAGiQ5prSLywslL+/v2O7j49PvftOmTJFhw4d0t69e+ts95//+Z8qLi5WTEyMLMtSUFCQxo0bp9dff12enp4N7usNVeHXJCYmRl9++WWtn/v4+Mjf399pAQCgIZprSv/nOVRf4E+dOlXbt2/Xrl271K1btzrb+vn56f3339elS5d04sQJFRQUKDQ0VB06dFDnzp0bPNYbqsKvSV5enoKDg693NwAAbVBLP7RnWZamTp2qLVu2KDMzU2FhYQ3e18vLy/HHwYYNG/TII4/oppsaXre7NfAvXryor776yrF+/Phx5efnq1OnTurevbuSk5N18uRJx/cJFy5cqNDQUEVERKiyslJr1qxRWlqa0tLS3NlNAABaxOTJk7Vu3Tpt27ZNHTp0UHFxsSTJZrPJz89Pkqpl4z/+8Q8dOHBAAwcO1Llz5/TWW2/p73//u/70pz+5dG63Bn5OTo6GDh3qWK+6156YmKhVq1apqKhIBQUFjs8rKys1Y8YMnTx5Un5+foqIiNCOHTsUHx/vzm4CAAzV0hV+amqqJOmBBx5w2r5y5UqNGzdOkqpl49WrV/Xmm2/q2LFj8vLy0tChQ7Vv3z6Fhoa6dG4PqzX9LmADlJaWymazaebMmdW+xgAAuPGVl5dr/vz5KikpcdtzWVVZ8fDDD8vLy6vRx7l8+bJ27Njh1r42lxv+oT0AANB0N/xDewAAuItJL88h8AEAxjIp8JnSBwDAAFT4AABjmVThE/gAAGOZFPhM6QMAYAAqfACAsUyq8Al8AICxCHwAAAxgUuBzDx8AAANQ4QMAjNaaqvSmIPABAMZiSh8AALQpVPgAAGOZVOET+AAAY5kU+EzpAwBgACp8AICxTKrwCXwAgLFMCnym9AEAMAAVPgDAWCZV+AQ+AMBYBD4AAAYwKfC5hw8AgAGo8AEAxjKpwifwAQDGMinwmdIHAMAAVPgAAGOZVOET+AAAY5kU+EzpAwBgACp8AICxTKrwCXwAgLFMCnym9AEAMAAVPgDAWCZV+AQ+AMBYBD4AAAYwKfC5hw8AgAGo8AEARmtNVXpTEPgAAGMxpQ8AANoUKnwAgLFMqvAJfACAsUwKfKb0AQAwABU+AMBYJlX4BD4AwFgmBT5T+gAAGIAKHwBgLJMqfAIfAGAsAh8AAAOYFPjcwwcAwABU+AAAY1HhN5OUlBTdfffd6tChgwIDA/XYY4/p2LFj9e6XlZWlqKgo+fr6qmfPnlq6dKk7uwkAMFRV4DdlcUVjc3Ht2rXq37+/brnlFgUHB+u3v/2tzp4969K53Rr4WVlZmjx5srKzs5WRkaErV64oNjZWZWVlte5z/PhxxcfHa8iQIcrLy9OsWbM0bdo0paWlubOrAAC4XWNyce/evRo7dqzGjx+vzz//XJs2bdLBgwc1YcIEl87t1in99PR0p/WVK1cqMDBQubm5uu+++2rcZ+nSperevbsWLlwoSerTp49ycnK0YMECPfnkk+7sLgDAMC09pd+YXMzOzlZoaKimTZsmSQoLC9PEiRP1+uuvu3TuFn1or6SkRJLUqVOnWtvs379fsbGxTttGjhypnJwcXb58uVr7iooKlZaWOi0AADREc03p/zyHKioqGnT+huTioEGD9O2332rnzp2yLEunTp3SX/7yFz388MMujbXFAt+yLCUlJWnw4MHq169fre2Ki4sVFBTktC0oKEhXrlzRmTNnqrVPSUmRzWZzLCEhIc3edwAA6hISEuKURSkpKfXu09BcHDRokNauXasxY8bI29tbXbp0UceOHfVf//VfLvWxxQJ/ypQpOnTokNavX19vWw8PD6f1qr+gfr5dkpKTk1VSUuJYCgsLm6fDAIA2r7kq/MLCQqcsSk5OrvfcDc3FL774QtOmTdMf/vAH5ebmKj09XcePH9ekSZNcGmuLfC1v6tSp2r59u3bv3q1u3brV2bZLly4qLi522nb69GndfPPNCggIqNbex8dHPj4+zdpfAIAZmusevr+/v/z9/Ru8nyu5mJKSonvvvVf/9m//JkmKjIxUu3btNGTIEL366qsKDg5u0DndWuFblqUpU6Zo8+bN+uSTTxQWFlbvPna7XRkZGU7bPv74Y0VHR8vLy8tdXQUAwO0ak4uXLl3STTc5x7Wnp6fjeA3l1sCfPHmy1qxZo3Xr1qlDhw4qLi5WcXGxfvjhB0eb5ORkjR071rE+adIkffPNN0pKStKRI0f0/vvva8WKFZoxY4Y7uwoAMFBLfw+/Mbk4atQobd68Wampqfr666/117/+VdOmTdM999yjrl27Nvjcbp3ST01NlSQ98MADTttXrlypcePGSZKKiopUUFDg+CwsLEw7d+7Uiy++qCVLlqhr16565513+EoeAKDZtfTX8hqTi+PGjdOFCxe0ePFivfTSS+rYsaOGDRum1157zaVze1it6XcBG6C0tFQ2m00zZ86Ur6/v9e4OAMBF5eXlmj9/vkpKSly6L+6Kqqzo2bOnY3q8Ma5evaqvv/7arX1tLrw8BwAAA/DyHACA0drYRHetCHwAgLF4Wx4AAGhTqPABAMYyqcIn8AEAxjIp8JnSBwDAAFT4AABjmVThE/gAAGOZFPhM6QMAYAAqfACAsUyq8Al8AICxCHwAAAxgUuBzDx8AAANQ4QMAjGVShU/gAwCMZVLgM6UPAIABqPABAMYyqcIn8AEAxjIp8JnSBwDAAFT4AABjmVThE/gAAGOZFPhM6QMAYAAqfACAsUyq8Al8AICxCHwAAAxgUuBzDx8AAANQ4QMAjNaaqvSmIPABAMZqati3pj8WmNIHAMAAVPgAAGOZVOET+AAAY5kU+EzpAwBgACp8AICxTKrwCXwAgLFMCnym9AEAMAAVPgDAWCZV+AQ+AMBYBD4AAAYwKfC5hw8AgAGo8AEAxjKpwifwAQDGMinwmdIHAMAAVPgAAGOZVOET+AAAY5kU+EzpAwBgACp8AICxTKrwCXwAgLFMCnym9AEAMAAVPgDAWFT4zSQlJUV33323OnTooMDAQD322GM6duxYnftkZmbKw8Oj2nL06FF3dhUAYCDLspq8uKIxuThu3LgaczEiIsKlc7s18LOysjR58mRlZ2crIyNDV65cUWxsrMrKyurd99ixYyoqKnIsvXr1cmdXAQAGaunAb0wuLlq0yCkPCwsL1alTJ/3Lv/yLS+d265R+enq60/rKlSsVGBio3Nxc3XfffXXuGxgYqI4dO7qxdwAAtKzG5KLNZpPNZnOsb926VefOndNvf/tbl87dovfwS0pKJEmdOnWqt+2AAQNUXl6uvn376j/+4z80dOjQGttVVFSooqKi2jl+ug0A0HpU/fe7pe6PN8d5SktLndZ9fHzk4+NT736u5GKVFStW6MEHH1SPHj1c66TVQq5du2aNGjXKGjx4cJ3tjh49ai1fvtzKzc219u3bZ73wwguWh4eHlZWVVWP7OXPmWJJYWFhYWNrY8r//+7/uiCPLsizrhx9+sLp06dIs/Wzfvn21bXPmzKm3Dw3NxZ/67rvvLE9PT2vjxo0uj9nDslrmT6jJkydrx44d2rt3r7p16+bSvqNGjZKHh4e2b99e7bOfV/jnz59Xjx49VFBQ4DQF0taVlpYqJCREhYWF8vf3v97daTEmjtvEMUtmjtvEMUs/Vr3du3fXuXPn3Hprt7y8XJWVlU0+jmVZ8vDwcNrWkAq/MbmYkpKiN998U9999528vb1d6meLTOlPnTpV27dv1+7du10Oe0mKiYnRmjVravystn9Um81m1P9Bqvj7+zNuQ5g4ZsnMcZs4Zkm66Sb3/lSMr6+vfH193XqO2jQmFy3L0vvvv6+EhASXw15yc+BblqWpU6dqy5YtyszMVFhYWKOOk5eXp+Dg4GbuHQAALaspuZiVlaWvvvpK48ePb9S53Rr4kydP1rp167Rt2zZ16NBBxcXFkn6svv38/CRJycnJOnnypFavXi1JWrhwoUJDQxUREaHKykqtWbNGaWlpSktLc2dXAQBwu8bkYpUVK1Zo4MCB6tevX6PO7dbAT01NlSQ98MADTttXrlypcePGSZKKiopUUFDg+KyyslIzZszQyZMn5efnp4iICO3YsUPx8fENOqePj4/mzJnToKcj2xLGbc64TRyzZOa4TRyz1LbH3ZhclH58riEtLU2LFi1q9Llb7KE9AABw/fDyHAAADEDgAwBgAAIfAAADEPgAABiAwAcAwABtIvDPnTunhIQExxuFEhISdP78+Tr3qen9wjExMS3T4UZ69913FRYWJl9fX0VFRWnPnj11ts/KylJUVJR8fX3Vs2dPLV26tIV62nxcGXNmZmaN74w+evRoC/a46Xbv3q1Ro0apa9eu8vDw0NatW+vdp7Vfa1fH3BaudWPeiy61/mvdmHG3het9I2gTgf/0008rPz9f6enpSk9PV35+vhISEurd76GHHnJ6x/DOnTtboLeNs3HjRk2fPl2zZ89WXl6ehgwZori4uGrf1axy/PhxxcfHa8iQIcrLy9OsWbM0bdq0VvUDRq6OucqxY8ecrmuvXr1aqMfNo6ysTP3799fixYsb1L4tXGtXx1ylNV/rxrwXvS1c68aMu0prvt43BJdft3OD+eKLLyxJVnZ2tmPb/v37LUnW0aNHa90vMTHRGj16dAv0sHncc8891qRJk5y29e7d25o5c2aN7X//+99bvXv3dto2ceJEKyYmxm19bG6ujnnXrl2WJOvcuXMt0LuWIcnasmVLnW3awrX+qYaMuS1e69OnT1uSan0zqGW1vWttWQ0bd1u83tdDq6/w9+/fL5vNpoEDBzq2xcTEyGazad++fXXum5mZqcDAQN1xxx167rnndPr0aXd3t1EqKyuVm5ur2NhYp+2xsbG1jnH//v3V2o8cOVI5OTm6fPmy2/raXBoz5ioDBgxQcHCwhg8frl27drmzmzeE1n6tm6ItXeuGvBe9LV5rV94H35au9/XQ6gO/uLhYgYGB1bYHBgY6fqO4JnFxcVq7dq0++eQTvfnmmzp48KCGDRvm9KrdG8WZM2d09epVBQUFOW0PCgqqdYzFxcU1tr9y5YrOnDnjtr42l8aMOTg4WMuXL1daWpo2b96s8PBwDR8+XLt3726JLl83rf1aN0Zbu9aWZSkpKUmDBw+u83fS29q1bui429r1vl5a5PW4jTF37ly9/PLLdbY5ePCgJFV7D7FU8/uJf2rMmDGO/92vXz9FR0erR48e2rFjh5544olG9tq9fj6e+sZYU/uatt/IXBlzeHi4wsPDHet2u12FhYVasGCB7rvvPrf283prC9faFW3tWk+ZMkWHDh3S3r17623blq51Q8fd1q739XLDBv6UKVP061//us42oaGhOnTokE6dOlXts++//77aX8J1CQ4OVo8ePfTll1+63Fd369y5szw9PatVtqdPn651jF26dKmx/c0336yAgAC39bW5NGbMNYmJidGaNWuau3s3lNZ+rZtLa73WrrwXvS1d68a8D/6nWuv1vp5u2MDv3LmzOnfuXG87u92ukpISHThwQPfcc48k6W9/+5tKSko0aNCgBp/v7NmzKiwsVHBwcKP77C7e3t6KiopSRkaGHn/8ccf2jIwMjR49usZ97Ha7PvzwQ6dtH3/8saKjo+Xl5eXW/jaHxoy5Jnl5eTfkNW1Orf1aN5fWdq2tRrwXvS1c68aMuyat7XrfEK7X04LN6aGHHrIiIyOt/fv3W/v377fuvPNO65FHHnFqEx4ebm3evNmyLMu6cOGC9dJLL1n79u2zjh8/bu3atcuy2+3WbbfdZpWWll6PIdRrw4YNlpeXl7VixQrriy++sKZPn261a9fOOnHihGVZljVz5kwrISHB0f7rr7+2brnlFuvFF1+0vvjiC2vFihWWl5eX9Ze//OV6DcFlro757bfftrZs2WL94x//sP7+979bM2fOtCRZaWlp12sIjXLhwgUrLy/PysvLsyRZb731lpWXl2d98803lmW1zWvt6pjbwrV+4YUXLJvNZmVmZlpFRUWO5dKlS442bfFaN2bcbeF63wjaROCfPXvWeuaZZ6wOHTpYHTp0sJ555plqX9+QZK1cudKyLMu6dOmSFRsba916662Wl5eX1b17dysxMdEqKCho+c67YMmSJVaPHj0sb29v66677nL6GktiYqJ1//33O7XPzMy0BgwYYHl7e1uhoaFWampqC/e46VwZ82uvvWb98pe/tHx9fa1f/OIX1uDBg60dO3Zch143TdVXkH6+JCYmWpbVNq+1q2NuC9e6pvH+9L9TltU2r3Vjxt0WrveNwMOy/v8THwAAoM1q9V/LAwAA9SPwAQAwAIEPAIABCHwAAAxA4AMAYAACHwAAAxD4AAAYgMAHAMAABD4AAAYg8AEAMACBDwCAAf4fBH1UDkWMogQAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for channel,image in enumerate(out[0]):\n",
    "    plt.imshow(image.detach().numpy(), interpolation='nearest', cmap=plt.cm.gray)\n",
    "    print(image)\n",
    "    plt.title(\"channel {}\".format(channel))\n",
    "    plt.colorbar()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Different kernels can be used to detect various features in an image. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[[[0., 0., 0., 0., 0.],\n",
      "          [0., 0., 0., 0., 0.],\n",
      "          [1., 1., 1., 1., 1.],\n",
      "          [0., 0., 0., 0., 0.],\n",
      "          [0., 0., 0., 0., 0.]]]])\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "image1=torch.zeros(1,1,5,5)\n",
    "image1[0,0,2,:]=1\n",
    "print(image1)\n",
    "plt.imshow(image1[0,0,:,:].detach().numpy(), interpolation='nearest', cmap=plt.cm.gray)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[0., 0., 0.],\n",
      "        [0., 0., 0.],\n",
      "        [0., 0., 0.]], grad_fn=<UnbindBackward0>)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhkAAAGxCAYAAADPvaSVAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAA9hAAAPYQGoP6dpAABAtUlEQVR4nO3de1xVdb7/8fcOFdRkpyIXC5AaB/FSKZZgmZYTXjKzy6RZZDNmOaam5PFI1kO0KdLjMacxr1nmqGkzZHUeOSTNCNZD8BZoF3XshELGzjAFNQV11u+PfuzTls2dLyLr9Xw81uPh/u7vd63vcs+0336+a6/lsCzLEgAAQD274lJPAAAANE2EDAAAYAQhAwAAGEHIAAAARhAyAACAEYQMAABgBCEDAAAYQcgAAABGEDIAAIARhAygllatWiWHw6Fdu3Zd6qnUi/T0dDkcDqWnp1fZ99y5c5o9e7Y6deokX19fdenSRX/+85/NTxLAZaXZpZ4AgMvPhAkT9Je//EUvvPCCbrrpJn300Ud6+umndfLkST377LOXenoAGglCBoAa+fLLL7Vy5Uq9+OKL+o//+A9J0oABA3Ts2DH98Y9/1Pjx49WuXbtLPEsAjQHLJUAF9u/fr4ceekhBQUHy9fVVWFiYHn30UZWUlHj0O3nypP7whz8oICBA7du313333afvvvvOo8+GDRsUFxenkJAQtWzZUlFRUZoxY4ZOnz7t0e+xxx7TlVdeqa+//lpDhw7VlVdeqdDQUD3zzDMexz106JAcDofmz5+vBQsWKCIiQldeeaViY2OVlZVV7lx27dql4cOHq127dvLz81PPnj31zjvv1Orv5b333pNlWfrd737n0f673/1OZ86cUWpqaq32C6DpIWQAXuzZs0c33XSTsrKyNGfOHP39739XcnKySkpKVFpa6tH38ccfV/PmzbVu3TrNmzdP6enpeuSRRzz6HDx4UEOHDtXKlSuVmpqqKVOm6J133tHdd99d7tjnzp3T8OHDNXDgQL3//vv6/e9/r1deeUVz584t1/e1115TWlqaFi5cqLVr1+r06dMaOnSoioqK3H22bNmiW265RSdOnNDSpUv1/vvv68Ybb9TIkSO1atWqGv/dfPHFF+rQoYOCg4M92q+//nr3+wAgSbIAlHPHHXdYV111lXX06NEK+7z55puWJGvChAke7fPmzbMkWQUFBV7H/fvf/7bOnTtnZWRkWJKsPXv2uN8bM2aMJcl65513PMYMHTrUioyMdL/Ozc21JFk9evSwzp8/727fsWOHJcl6++233W1dunSxevbsaZ07d85jn8OGDbNCQkKsCxcuWJZlWVu2bLEkWVu2bKnwnC3Lsu68806PufxSixYtrCeeeKLS8QDsg0oGcJGffvpJGRkZevDBB9WhQ4cq+w8fPtzjddm/6A8fPuxu++abbzR69GgFBwfLx8dHzZs3V//+/SVJ+/bt8xjvcDjKVTiuv/56j/2Vueuuu+Tj41Phsb/++mvt379fDz/8sCTp/Pnz7m3o0KEqKCjQgQMHqjzHizkcjlq9B8BeuPATuMjx48d14cIFXXPNNdXq3759e4/Xvr6+kqQzZ85Ikk6dOqV+/frJz89Pf/zjH/XrX/9arVq1Un5+vu677z53vzKtWrWSn59fuX2ePXu2xsf+/vvvJUnTpk3TtGnTvM6/sLCwWuf5y2Pm5OSUaz99+rRKS0u56BOAGyEDuEi7du3k4+Ojb7/9tl72989//lPfffed0tPT3dULSTpx4kS97L8yAQEBkqTExETdd999XvtERkbWaJ89evTQ+vXr5XK5PK7L+PzzzyVJ3bt3r+VsATQ1LJcAF2nZsqX69++vv/71rzX+V743ZcsHZVWGMsuWLavzvqsSGRmpzp07a8+ePerdu7fXrU2bNjXa5z333COHw6G33nrLo33VqlVq2bKlBg8eXJ+nAOAyRiUD8GLBggW69dZb1adPH82YMUO/+tWv9P333+uDDz7QsmXLavTF3LdvX7Vt21bjx4/XrFmz1Lx5c61du1Z79uwxeAb/Z9myZRoyZIgGDRqkxx57TFdffbV+/PFH7du3T5999pn++te/1mh/3bp109ixYzVr1iz5+Pjopptu0ubNm7V8+XL98Y9/ZLkEgBshA/Dihhtu0I4dOzRr1iwlJibq5MmTCg4O1h133KEWLVrUaF/t27fXhx9+qGeeeUaPPPKIWrdurXvuuUcbNmxQr169DJ3B/7n99tu1Y8cOvfjii5oyZYqOHz+u9u3bq2vXrnrwwQdrtc/Fixfr6quv1p///Ge5XC516tRJf/rTnzRp0qR6nj2Ay5nDsizrUk8CAAA0PVyTAQAAjCBkAAAAIwgZAADACKMh4/jx44qPj5fT6ZTT6VR8fHyV9wZ47LHH5HA4PLaYmBiT0wQAoEEtXrxYERER8vPzU3R0tD755JMK+xYUFGj06NGKjIzUFVdcoSlTpnjtl5KSoq5du8rX11ddu3bVxo0b63Tc+mA0ZIwePVo5OTlKTU1VamqqcnJyFB8fX+W4wYMHq6CgwL1t2rTJ5DQBAGgwGzZs0JQpUzRz5kxlZ2erX79+GjJkiPLy8rz2LykpUYcOHTRz5kzdcMMNXvtkZmZq5MiRio+P1549exQfH68HH3xQ27dvr/Vx64OxX5fs27dPXbt2VVZWlvr06SNJysrKUmxsrPbv31/hXQYfe+wxnThxQu+9956JaQEAcEn16dNHvXr10pIlS9xtUVFRGjFihJKTkysdO2DAAN14441auHChR/vIkSNVXFysv//97+62wYMHq23btnr77bfrfNzaMnafjMzMTDmdTnfAkKSYmBg5nU5t27at0lsZp6enKzAwUFdddZX69++vF198UYGBgV77lpSUqKSkxP363//+t3788Ue1b9+eBzUBwGXIsiydPHlSHTt21BVXmCu4nz17VqWlpXXej2VZ5b5vfH19y93lV5JKS0u1e/duzZgxw6M9Li5O27Ztq/UcMjMzNXXqVI+2QYMGucOIqeNWxVjIcLlcXoNBYGCgXC5XheOGDBmi3/72twoPD1dubq6ef/553XHHHdq9e7fXDyw5OVmzZ8+u17kDAC69/Pz8aj+osKbOnj2riIiISr+PquvKK6/UqVOnPNpmzZqlpKSkcn0LCwt14cIFBQUFebQHBQXVaS4ul6vSfZo6blVqHDKSkpKq/FLfuXOnJO+PfPaW+H5p5MiR7j93795dvXv3Vnh4uD788EOvD3hKTExUQkKC+3VRUZHCwsI0depUr6EEANC4lZSU6JVXXqnxc3VqorS0VC6XS3l5efL396/1foqLixUWFqb8/HyP/VT1/XPx92BV343VUZ19mjhuZWocMiZOnKhRo0ZV2qdTp07au3ev+zHTv/TDDz+US1KVCQkJUXh4uA4ePOj1/YpKUr6+vuUelw0AuHw0xJK3v79/nUJGTfcTEBAgHx+fctWDo0eP1ui78WLBwcGV7tPUcatS48WugIAAdenSpdLNz89PsbGxKioq0o4dO9xjt2/frqKiIvXt27faxzt27Jjy8/MVEhJS06kCAFApy7LqvNVEixYtFB0drbS0NI/2tLS0Gn03Xiw2NrbcPjdv3uzep6njVsXYNRlRUVEaPHiwxo0b536k9RNPPKFhw4Z5XPTZpUsXJScn695779WpU6eUlJSk+++/XyEhITp06JCeffZZBQQE6N577zU1VQCATdUmKFw8vqYSEhIUHx+v3r17KzY2VsuXL1deXp7Gjx8v6efLAI4cOaLVq1e7x+Tk5EiSTp06pR9++EE5OTlq0aKFunbtKkl6+umnddttt2nu3Lm655579P777+vjjz/Wp59+Wu3jmmD0Kaxr167V5MmTFRcXJ0kaPny4Fi1a5NHnwIEDKioqkiT5+Pjo888/1+rVq3XixAmFhITo9ttv14YNG4yuzQEA7OlShIyRI0fq2LFjmjNnjgoKCtS9e3dt2rRJ4eHhkn6++dbF967o2bOn+8+7d+/WunXrFB4erkOHDkmS+vbtq/Xr1+u5557T888/r+uuu04bNmzw+IVnVcc1ock9hbW4uFhOp1MzZszgmgwAuAydPXtWL7/8soqKiurleglvyr4rjh07VucLP9u3b290rpczo5UMAAAas0tRybATQgYAwLYIGWbxFFYAAGAElQwAgG1RyTCLkAEAsC1ChlkslwAAACOoZAAAbItKhlmEDACAbREyzGK5BAAAGEElAwBgW1QyzCJkAABsi5BhFiEDAGBbhAyzuCYDAAAYQSUDAGBbVDLMImQAAGyLkGEWyyUAAMAIKhkAANuikmEWIQMAYFuEDLNYLgEAAEZQyQAA2BaVDLMIGQAAWyMomMNyCQAAMIJKBgDAtlguMYuQAQCwLUKGWYQMAIBtETLM4poMAABgBJUMAIBtUckwi5ABALAtQoZZLJcAAAAjqGQAAGyLSoZZhAwAgG0RMsxiuQQAABhBJQMAYFtUMswiZAAAbIuQYRbLJQAAwAhCBgDAtsoqGXXZamPx4sWKiIiQn5+foqOj9cknn1TaPyMjQ9HR0fLz89O1116rpUuXerw/YMAAORyOcttdd93l7pOUlFTu/eDg4FrNv7oIGQAA27oUIWPDhg2aMmWKZs6cqezsbPXr109DhgxRXl6e1/65ubkaOnSo+vXrp+zsbD377LOaPHmyUlJS3H3effddFRQUuLcvvvhCPj4++u1vf+uxr27dunn0+/zzz2s8/5rgmgwAgG1dimsyFixYoLFjx+rxxx+XJC1cuFAfffSRlixZouTk5HL9ly5dqrCwMC1cuFCSFBUVpV27dmn+/Pm6//77JUnt2rXzGLN+/Xq1atWqXMho1qyZ8erFL1HJAACgjoqLiz22kpISr/1KS0u1e/duxcXFebTHxcVp27ZtXsdkZmaW6z9o0CDt2rVL586d8zpm5cqVGjVqlFq3bu3RfvDgQXXs2FEREREaNWqUvvnmm+qeYq0QMgAAtlVfyyWhoaFyOp3uzVtFQpIKCwt14cIFBQUFebQHBQXJ5XJ5HeNyubz2P3/+vAoLC8v137Fjh7744gt3paRMnz59tHr1an300UdasWKFXC6X+vbtq2PHjlX776umWC4BANhWfS2X5Ofny9/f393u6+tb6TiHw1FuPxe3VdXfW7v0cxWje/fuuvnmmz3ahwwZ4v5zjx49FBsbq+uuu05vvfWWEhISKp1vbREyAACoI39/f4+QUZGAgAD5+PiUq1ocPXq0XLWiTHBwsNf+zZo1U/v27T3af/rpJ61fv15z5sypci6tW7dWjx49dPDgwSr71hbLJQAA22roX5e0aNFC0dHRSktL82hPS0tT3759vY6JjY0t13/z5s3q3bu3mjdv7tH+zjvvqKSkRI888kiVcykpKdG+ffsUEhJSo3OoCUIGAMC2LsVPWBMSEvT666/rjTfe0L59+zR16lTl5eVp/PjxkqTExEQ9+uij7v7jx4/X4cOHlZCQoH379umNN97QypUrNW3atHL7XrlypUaMGFGuwiFJ06ZNU0ZGhnJzc7V9+3Y98MADKi4u1pgxY2p8DtXFcgkAAA1o5MiROnbsmObMmaOCggJ1795dmzZtUnh4uCSpoKDA454ZERER2rRpk6ZOnarXXntNHTt21Kuvvur++WqZf/3rX/r000+1efNmr8f99ttv9dBDD6mwsFAdOnRQTEyMsrKy3Mc1wWE1sRuvFxcXy+l0asaMGfLz87vU0wEA1NDZs2f18ssvq6ioqFrXOdRG2XfFZ599pjZt2tR6PydPnlSvXr2MzvVyRiUDAGBrTezf2o0K12QAAAAjqGQAAGyLR72bRcgAANgWIcOsBlkuqe9H2gIAUB8u1aPe7cJ4yDDxSFsAAND4GQ8Zv3ykbVRUlBYuXKjQ0FAtWbLEa/9fPtI2KipKjz/+uH7/+99r/vz5pqcKALAZKhlmGQ0ZDfFI25KSknKP2AUAoDoIGWYZDRkN8Ujb5ORkj8frhoaG1t8JAACAWmuQCz9NPtI2MTFRRUVF7i0/P78eZgwAsAMqGWYZ/Qmr6UfaSpKvr698fX3rb9IAANvgJ6xmGa1kmH6kLQAAaLyML5eYfKQtAAB1wXKJWcbv+GnqkbYAANQVyyVmNchtxSdMmKAJEyZ4fW/VqlXl2vr376/PPvvM8KwAAIBJPLsEAGBbVDLMImQAAGyLkGEWIQMAYFuEDLMa5GZcAADAfqhkAABsi0qGWYQMAIBtETLMYrkEAAAYQSUDAGBbVDLMImQAAGyLkGEWyyUAAMAIKhkAANuikmEWIQMAYGsEBXNYLgEAAEZQyQAA2BbLJWYRMgAAtkXIMIuQAQCwLUKGWVyTAQAAjKCSAQCwLSoZZhEyAAC2Rcgwi+USAABgBCEDAGBbZZWMumy1sXjxYkVERMjPz0/R0dH65JNPKu2fkZGh6Oho+fn56dprr9XSpUs93l+1apUcDke57ezZs3U6bl0RMgAAtnUpQsaGDRs0ZcoUzZw5U9nZ2erXr5+GDBmivLw8r/1zc3M1dOhQ9evXT9nZ2Xr22Wc1efJkpaSkePTz9/dXQUGBx+bn51fr49YHQgYAAA1owYIFGjt2rB5//HFFRUVp4cKFCg0N1ZIlS7z2X7p0qcLCwrRw4UJFRUXp8ccf1+9//3vNnz/fo5/D4VBwcLDHVpfj1gdCBgDAtuqrklFcXOyxlZSUeD1eaWmpdu/erbi4OI/2uLg4bdu2zeuYzMzMcv0HDRqkXbt26dy5c+62U6dOKTw8XNdcc42GDRum7OzsOh23PhAyAAC2VV8hIzQ0VE6n070lJyd7PV5hYaEuXLigoKAgj/agoCC5XC6vY1wul9f+58+fV2FhoSSpS5cuWrVqlT744AO9/fbb8vPz0y233KKDBw/W+rj1gZ+wAgBQR/n5+fL393e/9vX1rbS/w+HweG1ZVrm2qvr/sj0mJkYxMTHu92+55Rb16tVLf/7zn/Xqq6/W+rh1RcgAANhWfd0nw9/f3yNkVCQgIEA+Pj7lqgdHjx4tV2UoExwc7LV/s2bN1L59e69jrrjiCt10003uSkZtjlsfWC4BANhWQ/+6pEWLFoqOjlZaWppHe1pamvr27et1TGxsbLn+mzdvVu/evdW8efMKzysnJ0chISG1Pm59oJIBALCtS3HHz4SEBMXHx6t3796KjY3V8uXLlZeXp/Hjx0uSEhMTdeTIEa1evVqSNH78eC1atEgJCQkaN26cMjMztXLlSr399tvufc6ePVsxMTHq3LmziouL9eqrryonJ0evvfZatY9rAiEDAIAGNHLkSB07dkxz5sxRQUGBunfvrk2bNik8PFySVFBQ4HHvioiICG3atElTp07Va6+9po4dO+rVV1/V/fff7+5z4sQJPfHEE3K5XHI6nerZs6e2bt2qm2++udrHNcFhNbEbrxcXF8vpdGrGjBkeNyEBAFwezp49q5dffllFRUXVus6hNsq+K9577z21bt261vs5ffq0RowYYXSulzMqGQAA2+IBaWZx4ScAADCCSgYAwLaoZJhFyAAA2BYhwyyWSwAAgBFUMgAAtkUlwyxCBgDA1ggK5rBcAgAAjKCSAQCwLZZLzCJkAABsi5BhFiEDAGBbhAyzuCYDAAAYQSUDAGBbVDLMImQAAGyLkGEWyyUAAMAIKhkAANuikmEWIQMAYFuEDLNYLgEAAEZQyQAA2BaVDLMIGQAA2yJkmNUgyyWLFy9WRESE/Pz8FB0drU8++aTCvunp6XI4HOW2/fv3N8RUAQBAPTFeydiwYYOmTJmixYsX65ZbbtGyZcs0ZMgQffXVVwoLC6tw3IEDB+Tv7+9+3aFDB9NTBQDYDJUMs4xXMhYsWKCxY8fq8ccfV1RUlBYuXKjQ0FAtWbKk0nGBgYEKDg52bz4+PqanCgCwmbKQUZcNFTMaMkpLS7V7927FxcV5tMfFxWnbtm2Vju3Zs6dCQkI0cOBAbdmypcJ+JSUlKi4u9tgAAKgOQoZZRkNGYWGhLly4oKCgII/2oKAguVwur2NCQkK0fPlypaSk6N1331VkZKQGDhyorVu3eu2fnJwsp9Pp3kJDQ+v9PAAAQM01yK9LHA6Hx2vLssq1lYmMjFRkZKT7dWxsrPLz8zV//nzddttt5fonJiYqISHB/bq4uJigAQCoFq7JMMtoyAgICJCPj0+5qsXRo0fLVTcqExMTozVr1nh9z9fXV76+vnWaJwDAnggZZhldLmnRooWio6OVlpbm0Z6Wlqa+fftWez/Z2dkKCQmp7+kBAACDjC+XJCQkKD4+Xr1791ZsbKyWL1+uvLw8jR8/XtLPyx1HjhzR6tWrJUkLFy5Up06d1K1bN5WWlmrNmjVKSUlRSkqK6akCAGyGSoZZxkPGyJEjdezYMc2ZM0cFBQXq3r27Nm3apPDwcElSQUGB8vLy3P1LS0s1bdo0HTlyRC1btlS3bt304YcfaujQoaanCgCwGUKGWQ1y4eeECRM0YcIEr++tWrXK4/X06dM1ffr0BpgVAAAwiWeXAABsi0qGWYQMAIBtETLMapAHpAEAAPuhkgEAsDWqEeZQyQAA2NalenbJ4sWLFRERIT8/P0VHR+uTTz6ptH9GRoaio6Pl5+ena6+9VkuXLvV4f8WKFerXr5/atm2rtm3b6je/+Y127Njh0ScpKUkOh8NjCw4OrtX8q4uQAQCwrUsRMjZs2KApU6Zo5syZys7OVr9+/TRkyBCP2zn8Um5uroYOHap+/fopOztbzz77rCZPnuxx/6j09HQ99NBD2rJlizIzMxUWFqa4uDgdOXLEY1/dunVTQUGBe/v8889rPP+aYLkEAIAGtGDBAo0dO1aPP/64pJ9vQvnRRx9pyZIlSk5OLtd/6dKlCgsL08KFCyVJUVFR2rVrl+bPn6/7779fkrR27VqPMStWrNDf/vY3/eMf/9Cjjz7qbm/WrJnx6sUvUckAANhWfVUyiouLPbaSkhKvxystLdXu3bsVFxfn0R4XF6dt27Z5HZOZmVmu/6BBg7Rr1y6dO3fO65iffvpJ586dU7t27TzaDx48qI4dOyoiIkKjRo3SN998U62/p9oiZAAAbKu+QkZoaKicTqd781aRkKTCwkJduHCh3ENCg4KCyj1MtIzL5fLa//z58yosLPQ6ZsaMGbr66qv1m9/8xt3Wp08frV69Wh999JFWrFghl8ulvn376tixY9X++6oplksAAKij/Px8+fv7u19X9XRwh8Ph8dqyrHJtVfX31i5J8+bN09tvv6309HT5+fm524cMGeL+c48ePRQbG6vrrrtOb731lhISEiqdb20RMgAAtlVfN+Py9/f3CBkVCQgIkI+PT7mqxdGjR8tVK8oEBwd77d+sWTO1b9/eo33+/Pl66aWX9PHHH+v666+vdC6tW7dWjx49dPDgwSrnXVsslwAAbKuhf13SokULRUdHKy0tzaM9LS1Nffv29TomNja2XP/Nmzerd+/eat68ubvtv/7rv/TCCy8oNTVVvXv3rnIuJSUl2rdvn0JCQmp0DjVByAAAoAElJCTo9ddf1xtvvKF9+/Zp6tSpysvL0/jx4yVJiYmJHr8IGT9+vA4fPqyEhATt27dPb7zxhlauXKlp06a5+8ybN0/PPfec3njjDXXq1Ekul0sul0unTp1y95k2bZoyMjKUm5ur7du364EHHlBxcbHGjBlj7FxZLgEA2NaleHbJyJEjdezYMc2ZM0cFBQXq3r27Nm3apPDwcElSQUGBxz0zIiIitGnTJk2dOlWvvfaaOnbsqFdffdX981Xp55t7lZaW6oEHHvA41qxZs5SUlCRJ+vbbb/XQQw+psLBQHTp0UExMjLKystzHNYGQAQCwrUv1gLQJEyZowoQJXt9btWpVubb+/fvrs88+q3B/hw4dqvKY69evr+706g0hAwBgWzyF1SyuyQAAAEZQyQAA2BaVDLMIGQAA2yJkmMVyCQAAMIJKBgDAtqhkmEXIAADYFiHDLJZLAACAEVQyAAC2RSXDLEIGAMC2CBlmsVwCAACMoJIBALAtKhlmETIAALZFyDCLkAEAsDWCgjlckwEAAIygkgEAsC2WS8wiZAAAbIuQYRbLJQAAwAgqGQAA26KSYRYhAwBgW4QMs1guAQAARlDJAADYFpUMswgZAADbImSYxXIJAAAwgkoGAMC2qGSYRcgAANgWIcMsQgYAwLYIGWZxTQYAADCCSgYAwLaoZJhFyAAA2BYhwyyWSwAAgBFUMgAAtkUlwyxCBgDAtggZZrFcAgAAjKCSAQCwLSoZZhEyAAC2Rcgwy+hyydatW3X33XerY8eOcjgceu+996ock5GRoejoaPn5+enaa6/V0qVLTU4RAIAGt3jxYkVERMjPz0/R0dH65JNPKu1fne/GlJQUde3aVb6+vuratas2btxY5+PWldGQcfr0ad1www1atGhRtfrn5uZq6NCh6tevn7Kzs/Xss89q8uTJSklJMTlNAIBNlVUy6rLV1IYNGzRlyhTNnDlT2dnZ6tevn4YMGaK8vDyv/avz3ZiZmamRI0cqPj5ee/bsUXx8vB588EFt37691setDw6rgWo9DodDGzdu1IgRIyrs85//+Z/64IMPtG/fPnfb+PHjtWfPHmVmZnodU1JSopKSEvfr4uJihYaGasaMGfLz86u3+QMAGsbZs2f18ssvq6ioSP7+/kaOUVxcLKfTWefvirK55ufne8zV19dXvr6+Xsf06dNHvXr10pIlS9xtUVFRGjFihJKTk8v1r85348iRI1VcXKy///3v7j6DBw9W27Zt9fbbb9fquPWhUf26JDMzU3FxcR5tgwYN0q5du3Tu3DmvY5KTk+V0Ot1baGhoQ0wVANBE1EcVIzQ01OO7qKIv7dLSUu3evbvcd11cXJy2bdvmdUx1vhsr6lO2z9octz40qgs/XS6XgoKCPNqCgoJ0/vx5FRYWKiQkpNyYxMREJSQkuF+XVTIAAGgo3ioZ3hQWFurChQtev+tcLpfXMdX5bqyoT9k+a3Pc+tCoQob087LKL5UlxYvby1RWkgIAoDL19esSf3//Gi3tePuuq+h7rqL+F7dXZ581PW5dNaqQERwcXC5RHT16VM2aNVP79u0v0awAAE1VQ/+ENSAgQD4+Pl6/6y6uMpSpzndjRX3K9lmb49aHRnVNRmxsrNLS0jzaNm/erN69e6t58+aXaFYAANSPFi1aKDo6utx3XVpamvr27et1THW+GyvqU7bP2hy3PhitZJw6dUpff/21+3Vubq5ycnLUrl07hYWFKTExUUeOHNHq1asl/Xy17KJFi5SQkKBx48YpMzNTK1eudF8ZCwBAfboUN+NKSEhQfHy8evfurdjYWC1fvlx5eXkaP368JNXqu/Hpp5/Wbbfdprlz5+qee+7R+++/r48//liffvpptY9rgtGQsWvXLt1+++3u12UXaI4ZM0arVq1SQUGBx+9zIyIitGnTJk2dOlWvvfaaOnbsqFdffVX333+/yWkCAGzqUoSMkSNH6tixY5ozZ44KCgrUvXt3bdq0SeHh4ZJUq+/Gvn37av369Xruuef0/PPP67rrrtOGDRvUp0+fah/XhAa7T0ZDqa/fPgMALo2GvE/GtGnT6vTjgZKSEs2fP9/oXC9njerCTwAAGhLPLjGLkAEAsC1ChlmN6tclAACg6aCSAQCwLSoZZhEyAAC2Rcgwi5ABALAtQoZZXJMBAACMoJIBALAtKhlmETIAALZFyDCL5RIAAGAElQwAgG1RyTCLkAEAsC1ChlkslwAAACOoZAAAbItKhlmEDACAbREyzGK5BAAAGEElAwBgW1QyzCJkAABsi5BhFiEDAGBrBAVzuCYDAAAYQSUDAGBbLJeYRcgAANgWIcMslksAAIARVDIAALZFJcMsQgYAwLYIGWaxXAIAAIygkgEAsC0qGWYRMgAAtkXIMIvlEgAAYASVDACAbVHJMIuQAQCwLUKGWYQMAIBtETLM4poMAABgBJUMAIBtUckwi5ABALAtQoZZLJcAANBIHT9+XPHx8XI6nXI6nYqPj9eJEycqHWNZlpKSktSxY0e1bNlSAwYM0Jdfful+/8cff9SkSZMUGRmpVq1aKSwsTJMnT1ZRUZHHfjp16iSHw+GxzZgxo0bzJ2QAAGyrrJJRl82k0aNHKycnR6mpqUpNTVVOTo7i4+MrHTNv3jwtWLBAixYt0s6dOxUcHKw777xTJ0+elCR99913+u677zR//nx9/vnnWrVqlVJTUzV27Nhy+5ozZ44KCgrc23PPPVej+bNcAgCwrca8XLJv3z6lpqYqKytLffr0kSStWLFCsbGxOnDggCIjI73OZ+HChZo5c6buu+8+SdJbb72loKAgrVu3Tk8++aS6d++ulJQU95jrrrtOL774oh555BGdP39ezZr9XzRo06aNgoODa30OVDIAAKij4uJij62kpKTO+8zMzJTT6XQHDEmKiYmR0+nUtm3bvI7Jzc2Vy+VSXFycu83X11f9+/evcIwkFRUVyd/f3yNgSNLcuXPVvn173XjjjXrxxRdVWlpao3OgkgEAsK36qmSEhoZ6tM+aNUtJSUl1mZpcLpcCAwPLtQcGBsrlclU4RpKCgoI82oOCgnT48GGvY44dO6YXXnhBTz75pEf7008/rV69eqlt27basWOHEhMTlZubq9dff73a50DIAADYVn2FjPz8fPn7+7vbfX19KxyTlJSk2bNnV7rfnTt3SpIcDofXY3pr/6WL369oTHFxse666y517dpVs2bN8nhv6tSp7j9ff/31atu2rR544AF3daM6CBkAANSRv7+/R8iozMSJEzVq1KhK+3Tq1El79+7V999/X+69H374oVylokzZ9RMul0shISHu9qNHj5Ybc/LkSQ0ePFhXXnmlNm7cqObNm1c6p5iYGEnS119/TcgAAKAql+LCz4CAAAUEBFTZLzY2VkVFRdqxY4duvvlmSdL27dtVVFSkvn37eh0TERGh4OBgpaWlqWfPnpKk0tJSZWRkaO7cue5+xcXFGjRokHx9ffXBBx/Iz8+vyvlkZ2dLkkd4qQohAwBgW4351yVRUVEaPHiwxo0bp2XLlkmSnnjiCQ0bNszjlyVdunRRcnKy7r33XjkcDk2ZMkUvvfSSOnfurM6dO+ull15Sq1atNHr0aEk/VzDi4uL0008/ac2aNe6LVSWpQ4cO8vHxUWZmprKysnT77bfL6XRq586dmjp1qoYPH66wsLBqnwMhAwBga435rp1r167V5MmT3b8WGT58uBYtWuTR58CBAx430po+fbrOnDmjCRMm6Pjx4+rTp482b96sNm3aSJJ2796t7du3S5J+9atfeewrNzdXnTp1kq+vrzZs2KDZs2erpKRE4eHhGjdunKZPn16j+RMyAABopNq1a6c1a9ZU2ufikORwOJSUlFThr1sGDBhQZbDq1auXsrKyajRXbwgZAADbaszLJU0BIQMAYFuEDLO44ycAADCCSgYAwLaoZJhltJKxdetW3X333erYsaMcDofee++9Svunp6eXe6ysw+HQ/v37TU4TAGBTjf0prJc7o5WM06dP64YbbtDvfvc73X///dUed+DAAY87p3Xo0MHE9AAAgEFGQ8aQIUM0ZMiQGo8LDAzUVVddVf8TAgDgF1guMatRXvjZs2dPhYSEaODAgdqyZUulfUtKSso9YhcAgOpgucSsRhUyQkJCtHz5cqWkpOjdd99VZGSkBg4cqK1bt1Y4Jjk5WU6n071d/LhdAABwaTSqX5dERkZ63I89NjZW+fn5mj9/vm677TavYxITE5WQkOB+XVxcTNAAAFQLyyVmNaqQ4U1MTEylt1T19fWVr69vA84IANBUEDLMavQhIzs7u0aPlQUAoLoIGWYZDRmnTp3S119/7X6dm5urnJwctWvXTmFhYUpMTNSRI0e0evVqSdLChQvVqVMndevWTaWlpVqzZo1SUlKUkpJicpoAAMAAoyFj165duv32292vy66dGDNmjFatWqWCggLl5eW53y8tLdW0adN05MgRtWzZUt26ddOHH36ooUOHmpwmAMCmqGSYZTRkVPU42VWrVnm8nj59eo2fVQ8AQG0RMsxqVD9hBQAATUejv/ATAABTqGSYRcgAANgWIcMslksAAIARVDIAALZFJcMsQgYAwLYIGWaxXAIAAIygkgEAsC0qGWYRMgAAtkXIMIuQAQCwLUKGWVyTAQAAjKCSAQCwNaoR5hAyAAC2xXKJWSyXAAAAI6hkAABsi0qGWYQMAIBtETLMYrkEAAAYQSUDAGBbVDLMImQAAGyLkGEWyyUAAMAIQgYAwLbKKhl12Uw6fvy44uPj5XQ65XQ6FR8frxMnTlR5TklJSerYsaNatmypAQMG6Msvv/ToM2DAADkcDo9t1KhRdT72xQgZAADbauwhY/To0crJyVFqaqpSU1OVk5Oj+Pj4SsfMmzdPCxYs0KJFi7Rz504FBwfrzjvv1MmTJz36jRs3TgUFBe5t2bJldT72xbgmAwBgW435mox9+/YpNTVVWVlZ6tOnjyRpxYoVio2N1YEDBxQZGel1PgsXLtTMmTN13333SZLeeustBQUFad26dXryySfdfVu1aqXg4OB6O7Y3VDIAAKij4uJij62kpKTO+8zMzJTT6XR/yUtSTEyMnE6ntm3b5nVMbm6uXC6X4uLi3G2+vr7q379/uTFr165VQECAunXrpmnTpnlUOmpzbG+oZAAAbKu+KhmhoaEe7bNmzVJSUlJdpiaXy6XAwMBy7YGBgXK5XBWOkaSgoCCP9qCgIB0+fNj9+uGHH1ZERISCg4P1xRdfKDExUXv27FFaWlqtj+0NIQMAYFv1FTLy8/Pl7+/vbvf19a1wTFJSkmbPnl3pfnfu3ClJcjgcXo/prf2XLn7/4jHjxo1z/7l79+7q3Lmzevfurc8++0y9evWq07F/iZABAEAd+fv7e4SMykycOLHcLzku1qlTJ+3du1fff/99ufd++OGHcpWKMmXXWLhcLoWEhLjbjx49WuEYSerVq5eaN2+ugwcPqlevXgoODq7xsb0hZAAAbOtSXPgZEBCggICAKvvFxsaqqKhIO3bs0M033yxJ2r59u4qKitS3b1+vY8qWQNLS0tSzZ09JUmlpqTIyMjR37twKj/Xll1/q3Llz7mBSm2N7w4WfAADbasw/YY2KitLgwYM1btw4ZWVlKSsrS+PGjdOwYcM8ft3RpUsXbdy4UdLPSxxTpkzRSy+9pI0bN+qLL77QY489platWmn06NGSpP/93//VnDlztGvXLh06dEibNm3Sb3/7W/Xs2VO33HJLjY5dFSoZAAA0UmvXrtXkyZPdvxYZPny4Fi1a5NHnwIEDKioqcr+ePn26zpw5owkTJuj48ePq06ePNm/erDZt2kiSWrRooX/84x/605/+pFOnTik0NFR33XWXZs2aJR8fnxoduyoOq4ndeL24uFhOp1MzZsyQn5/fpZ4OAKCGzp49q5dffllFRUXVvs6hpsq+K/r06aNmzWr/7+3z58+7lxFMzfVyRiUDAGBbjflmXE0B12QAAAAjqGQAAGyLSoZZhAwAgG0RMswiZAAAbIuQYRbXZAAAACOoZAAAbI1qhDmEDACAbbFcYhbLJQAAwAgqGQAA26KSYRYhAwBgW4QMs1guAQAARlDJAADYFpUMswgZAADbImSYxXIJAAAwgkoGAMC2qGSYRcgAANgWIcMsQgYAwLYIGWZxTQYAADCCSgYAwLaoZJhltJKRnJysm266SW3atFFgYKBGjBihAwcOVDkuIyND0dHR8vPz07XXXqulS5eanCYAwKbKQkZdNlTMaMjIyMjQU089paysLKWlpen8+fOKi4vT6dOnKxyTm5uroUOHql+/fsrOztazzz6ryZMnKyUlxeRUAQBAPTO6XJKamurx+s0331RgYKB2796t2267zeuYpUuXKiwsTAsXLpQkRUVFadeuXZo/f77uv/9+k9MFANgMyyVmNeiFn0VFRZKkdu3aVdgnMzNTcXFxHm2DBg3Srl27dO7cuXL9S0pKVFxc7LEBAFAdLJeY1WAhw7IsJSQk6NZbb1X37t0r7OdyuRQUFOTRFhQUpPPnz6uwsLBc/+TkZDmdTvcWGhpa73MHAAA112AhY+LEidq7d6/efvvtKvs6HA6P12VJ8eJ2SUpMTFRRUZF7y8/Pr58JAwCaPCoZZjXIT1gnTZqkDz74QFu3btU111xTad/g4GC5XC6PtqNHj6pZs2Zq3759uf6+vr7y9fWt1/kCAOyBazLMMlrJsCxLEydO1Lvvvqt//vOfioiIqHJMbGys0tLSPNo2b96s3r17q3nz5qamCgAA6pnRkPHUU09pzZo1Wrdundq0aSOXyyWXy6UzZ864+yQmJurRRx91vx4/frwOHz6shIQE7du3T2+88YZWrlypadOmmZwqAMCGWC4xy2jIWLJkiYqKijRgwACFhIS4tw0bNrj7FBQUKC8vz/06IiJCmzZtUnp6um688Ua98MILevXVV/n5KgCg3hEyzDJ6TUZ1/vJXrVpVrq1///767LPPDMwIAID/wzUZZvGANAAAYAQPSAMA2BrVCHMIGQAA22K5xCyWSwAAgBGEDACAbTX2X5ccP35c8fHx7kdnxMfH68SJE1WeU1JSkjp27KiWLVtqwIAB+vLLL93vHzp0SA6Hw+v217/+1d2vU6dO5d6fMWNGjeZPyAAA2FZjDxmjR49WTk6OUlNTlZqaqpycHMXHx1c6Zt68eVqwYIEWLVqknTt3Kjg4WHfeeadOnjwpSQoNDVVBQYHHNnv2bLVu3VpDhgzx2NecOXM8+j333HM1mj/XZAAA0Ajt27dPqampysrKUp8+fSRJK1asUGxsrA4cOKDIyMhyYyzL0sKFCzVz5kzdd999kqS33npLQUFBWrdunZ588kn5+PgoODjYY9zGjRs1cuRIXXnllR7tbdq0Kde3JqhkAABsq74qGcXFxR5bSUlJneeWmZkpp9PpDhiSFBMTI6fTqW3btnkdk5ubK5fLpbi4OHebr6+v+vfvX+GY3bt3KycnR2PHji333ty5c9W+fXvdeOONevHFF1VaWlqjc6CSAQCwrfr6dUloaKhH+6xZs5SUlFSXqcnlcikwMLBce2BgYLkHif5yjCQFBQV5tAcFBenw4cNex6xcuVJRUVHq27evR/vTTz+tXr16qW3bttqxY4cSExOVm5ur119/vdrnQMgAAKCO8vPz5e/v735d2dPBk5KSNHv27Er3t3PnTkmSw+Eo955lWV7bf+ni9ysac+bMGa1bt07PP/98ufemTp3q/vP111+vtm3b6oEHHnBXN6qDkAEAsK36qmT4+/t7hIzKTJw4UaNGjaq0T6dOnbR37159//335d774YcfylUqypRdP+FyuRQSEuJuP3r0qNcxf/vb3/TTTz95PKi0IjExMZKkr7/+mpABAEBVLsXNuAICAhQQEFBlv9jYWBUVFWnHjh26+eabJUnbt29XUVFRuaWNMhEREQoODlZaWpp69uwpSSotLVVGRobmzp1brv/KlSs1fPhwdejQocr5ZGdnS5JHeKkKIQMAYFuN+Y6fUVFRGjx4sMaNG6dly5ZJkp544gkNGzbM45clXbp0UXJysu699145HA5NmTJFL730kjp37qzOnTvrpZdeUqtWrTR69GiP/X/99dfaunWrNm3aVO7YmZmZysrK0u233y6n06mdO3dq6tSpGj58uMLCwqp9DoQMAAAaqbVr12ry5MnuX4sMHz5cixYt8uhz4MABFRUVuV9Pnz5dZ86c0YQJE3T8+HH16dNHmzdvVps2bTzGvfHGG7r66qs9folSxtfXVxs2bNDs2bNVUlKi8PBwjRs3TtOnT6/R/B1WE7vxenFxsZxOp2bMmCE/P79LPR0AQA2dPXtWL7/8soqKiqp9nUNNlX1XhISE6Ioran83h3//+98qKCgwOtfLGZUMAIBtNeblkqaAm3EBAAAjqGQAAGyLSoZZhAwAgG0RMsxiuQQAABhBJQMAYFtUMswiZAAAbIuQYRbLJQAAwAgqGQAA26KSYRYhAwBgW4QMswgZAADbImSYxTUZAADACCoZAABboxphDiEDAGBbdQ0YBJTKsVwCAACMoJIBALAtKhlmETIAALZFyDCL5RIAAGAElQwAgG1RyTCLkAEAsC1ChlkslwAAACOoZAAAbItKhlmEDACAbREyzCJkAABsi5BhFtdkAAAAI6hkAABsi0qGWYQMAIBtETLMYrkEAAAYQSUDAGBbVDLMImQAAGyLkGEWyyUAAMAIKhkAANuikmEWIQMAYFuEDLNYLgEAAEZQyQAA2BaVDLOMVjKSk5N10003qU2bNgoMDNSIESN04MCBSsekp6fL4XCU2/bv329yqgAAG7Isq86bScePH1d8fLycTqecTqfi4+N14sSJSse8++67GjRokAICAuRwOJSTk1OuT0lJiSZNmqSAgAC1bt1aw4cP17ffflvnY1/MaMjIyMjQU089paysLKWlpen8+fOKi4vT6dOnqxx74MABFRQUuLfOnTubnCoAwIYae8gYPXq0cnJylJqaqtTUVOXk5Cg+Pr7SMadPn9Ytt9yil19+ucI+U6ZM0caNG7V+/Xp9+umnOnXqlIYNG6YLFy7U6dgXM7pckpqa6vH6zTffVGBgoHbv3q3bbrut0rGBgYG66qqrDM4OAIDGa9++fUpNTVVWVpb69OkjSVqxYoViY2N14MABRUZGeh1XFgQOHTrk9f2ioiKtXLlSf/nLX/Sb3/xGkrRmzRqFhobq448/1qBBg2p97Is16DUZRUVFkqR27dpV2bdnz546e/asunbtqueee0633367134lJSUqKSkpd4xftgEALh9l//1uqOsd6uM4xcXFHq99fX3l6+tbp31mZmbK6XS6v+QlKSYmRk6nU9u2bav2F/3Fdu/erXPnzikuLs7d1rFjR3Xv3l3btm3ToEGD6u3YDRYyLMtSQkKCbr31VnXv3r3CfiEhIVq+fLmio6NVUlKiv/zlLxo4cKDS09O9Vj+Sk5M1e/bscu2vvPJKvc4fANCwjh07JqfTaWTfLVq0UHBwsFwuV533deWVVyo0NNSjbdasWUpKSqrTfl0ulwIDA8u1BwYG1mneLpdLLVq0UNu2bT3ag4KC3Putr2M3WMiYOHGi9u7dq08//bTSfpGRkR4JKTY2Vvn5+Zo/f77XkJGYmKiEhAT36xMnTig8PFx5eXnG/sfZGBUXFys0NFT5+fny9/e/1NNpMHY8bzues2TP87bjOUs/V6TDwsKqVfWuLT8/P+Xm5qq0tLTO+7IsSw6Hw6OtsipGUlKS138c/9LOnTslqdx+Kzpefbh4v/Vx7AYJGZMmTdIHH3ygrVu36pprrqnx+JiYGK1Zs8brexWVpJxOp63+T1nG39+f87YJO56zZM/ztuM5S9IVV5i9lZOfn5/8/PyMHsObiRMnatSoUZX26dSpk/bu3avvv/++3Hs//PCDgoKCan384OBglZaW6vjx4x7VjKNHj6pv377uPvVxbKMhw7IsTZo0SRs3blR6eroiIiJqtZ/s7GyFhITU8+wAAGh4AQEBCggIqLJfbGysioqKtGPHDt18882SpO3bt6uoqMgdBmojOjpazZs3V1pamh588EFJUkFBgb744gvNmzevXo9tNGQ89dRTWrdund5//321adPGvY7jdDrVsmVLST8vdxw5ckSrV6+WJC1cuFCdOnVSt27dVFpaqjVr1iglJUUpKSkmpwoAQKMSFRWlwYMHa9y4cVq2bJkk6YknntCwYcM8Livo0qWLkpOTde+990qSfvzxR+Xl5em7776TJPf9qYKDgxUcHCyn06mxY8fqmWeeUfv27dWuXTtNmzZNPXr0cP/apLrHrpJlkCSv25tvvunuM2bMGKt///7u13PnzrWuu+46y8/Pz2rbtq116623Wh9++GG1j3n27Flr1qxZ1tmzZ+vxTBo/zts+523Hc7Yse563Hc/Zsux73t4cO3bMevjhh602bdpYbdq0sR5++GHr+PHjHn0u/l598803vX73zpo1y93nzJkz1sSJE6127dpZLVu2tIYNG2bl5eXV+NhVcfz/CQIAANQrHpAGAACMIGQAAAAjCBkAAMAIQgYAADCCkAEAAIxoEiGjNs+8f+yxx+RwODy2mJiYhplwLS1evFgRERHy8/NTdHS0Pvnkk0r7Z2RkKDo6Wn5+frr22mu1dOnSBppp/anJOaenp5f7TB0Oh/bv39+AM667rVu36u6771bHjh3lcDj03nvvVTnmcv+sa3rOTeGzTk5O1k033aQ2bdooMDBQI0aMcN/PoDKX+2ddm/NuCp+3XTWJkFHbZ94PHjxYBQUF7m3Tpk0NMNva2bBhg6ZMmaKZM2cqOztb/fr105AhQ5SXl+e1f25uroYOHap+/fopOztbzz77rCZPnnxZ3dSspudc5sCBAx6fa+fOnRtoxvXj9OnTuuGGG7Ro0aJq9W8Kn3VNz7nM5fxZZ2Rk6KmnnlJWVpbS0tJ0/vx5xcXF6fTp0xWOaQqfdW3Ou8zl/HnbVo3uqtEIffXVV5YkKysry92WmZlpSbL2799f4bgxY8ZY99xzTwPMsH7cfPPN1vjx4z3aunTpYs2YMcNr/+nTp1tdunTxaHvyySetmJgYY3OsbzU95y1btliSanyzmMZMkrVx48ZK+zSFz/qXqnPOTfGzPnr0qCXJysjIqLBPU/usLat6590UP2+7uOwrGVU9874y6enpCgwM1K9//WuNGzdOR48eNT3dWiktLdXu3bsVFxfn0R4XF1fhOWZmZpbrP2jQIO3atUvnzp0zNtf6UptzLtOzZ0+FhIRo4MCB2rJli8lpNgqX+2ddF03psy4qKpKkSp882hQ/6+qcd5mm9HnbxWUfMmr7zPshQ4Zo7dq1+uc//6n//u//1s6dO3XHHXeopKTE5HRrpbCwUBcuXCj35LugoKAKz9Hlcnntf/78eRUWFhqba32pzTmHhIRo+fLlSklJ0bvvvqvIyEgNHDhQW7dubYgpXzKX+2ddG03ts7YsSwkJCbr11lvVvXv3Cvs1tc+6uufd1D5vO2mQR73XRlJSkmbPnl1pn507d0qq3TPvR44c6f5z9+7d1bt3b4WHh+vDDz/UfffdV8tZm3Xx+VR1jt76e2tvzGpyzpGRkR4P7omNjVV+fr7mz5+v2267zeg8L7Wm8FnXRFP7rCdOnKi9e/fq008/rbJvU/qsq3veTe3ztpNGGzImTpyoUaNGVdqnU6dO2rt3b7088z4kJETh4eE6ePBgjedqWkBAgHx8fMr9C/7o0aMVnmNwcLDX/s2aNVP79u2NzbW+1OacvYmJidGaNWvqe3qNyuX+WdeXy/WznjRpkj744ANt3bpV11xzTaV9m9JnXZPz9uZy/bztptGGjICAAAUEBFTZr76eeX/s2DHl5+crJCSk1nM2pUWLFoqOjlZaWpr7Ub6SlJaWpnvuucfrmNjYWP3P//yPR9vmzZvVu3dvNW/e3Oh860Ntztmb7OzsRvmZ1qfL/bOuL5fbZ21ZliZNmqSNGzcqPT1dERERVY5pCp91bc7bm8vt87atS3XFaX0aPHiwdf3111uZmZlWZmam1aNHD2vYsGEefSIjI613333XsizLOnnypPXMM89Y27Zts3Jzc60tW7ZYsbGx1tVXX20VFxdfilOo0vr1663mzZtbK1eutL766itrypQpVuvWra1Dhw5ZlmVZM2bMsOLj4939v/nmG6tVq1bW1KlTra+++spauXKl1bx5c+tvf/vbpTqFGqvpOb/yyivWxo0brX/961/WF198Yc2YMcOSZKWkpFyqU6iVkydPWtnZ2VZ2drYlyVqwYIGVnZ1tHT582LKspvlZ1/Scm8Jn/Yc//MFyOp1Wenq6VVBQ4N5++uknd5+m+FnX5rybwudtV00iZFTnmfeSrDfffNOyLMv66aefrLi4OKtDhw5W8+bNrbCwMGvMmDFWXl5ew0++Bl577TUrPDzcatGihdWrVy+Pn3yNGTPG6t+/v0f/9PR0q2fPnlaLFi2sTp06WUuWLGngGdddTc557ty51nXXXWf5+flZbdu2tW699Vbrww8/vASzrpuyn+tdvI0ZM8ayrKb5Wdf0nJvCZ+3tfH/53ynLapqfdW3Ouyl83nblsKz/f9UQAABAPbrsf8IKAAAaJ0IGAAAwgpABAACMIGQAAAAjCBkAAMAIQgYAADCCkAEAAIwgZAAAACMIGQAAwAhCBgAAMIKQAQAAjPh/g3avAg8gh2YAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[-4., -4., -4.],\n",
      "        [ 0.,  0.,  0.],\n",
      "        [ 4.,  4.,  4.]], grad_fn=<UnbindBackward0>)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[3., 3., 3.],\n",
      "        [3., 3., 3.],\n",
      "        [3., 3., 3.]], grad_fn=<UnbindBackward0>)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "out1=conv1(image1)\n",
    "for channel,image in enumerate(out1[0]):\n",
    "    plt.imshow(image.detach().numpy(), interpolation='nearest', cmap=plt.cm.gray)\n",
    "    print(image)\n",
    "    plt.title(\"channel {}\".format(channel))\n",
    "    plt.colorbar()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`Deep CNN`"
   ]
  }
 ],
 "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}