top-channels-used-per-layer.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "layers = {\n",
    "    'mixed3a': 256,\n",
    "    'mixed3b': 480,\n",
    "    'mixed4a': 508,\n",
    "    'mixed4b': 512,\n",
    "    'mixed4c': 512,\n",
    "    'mixed4d': 528,\n",
    "    'mixed4e': 832,\n",
    "    'mixed5a': 832,\n",
    "    'mixed5b': 1024\n",
    "}\n",
    "\n",
    "with open('data/imagenet.json') as f:\n",
    "    imagenet = json.load(f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compute top channel usage statistics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed3a\n",
      "mixed3b\n",
      "mixed4a\n",
      "mixed4b\n",
      "mixed4c\n",
      "mixed4d\n",
      "mixed4e\n",
      "mixed5a\n",
      "mixed5b\n"
     ]
    }
   ],
   "source": [
    "top_channels = {}\n",
    "\n",
    "for layer in list(layers.keys()):\n",
    "    print(layer)\n",
    "    top_channels[layer] = {\n",
    "        'channels': layers[layer],\n",
    "        'channels_used': 0,\n",
    "        'channels_not_used': 0,\n",
    "        'percent_channels_used': 0\n",
    "    }\n",
    "    all_channels = []\n",
    "    \n",
    "    for c in imagenet:\n",
    "        for ch in c['topChannels'][layer]:\n",
    "            all_channels.append(ch['channel'])\n",
    "            \n",
    "    (channels, counts) = np.unique(all_channels, return_counts=True)\n",
    "    sorted_counts, sorted_channels = (list(t) for t in zip(*sorted(zip(counts, channels), reverse=True)))\n",
    "    \n",
    "    top_channels[layer]['channels_used'] = int(len(np.unique(sorted_channels)))\n",
    "    top_channels[layer]['channels_not_used'] = layers[layer] - top_channels[layer]['channels_used']\n",
    "    top_channels[layer]['percent_channels_used'] = top_channels[layer]['channels_used'] / layers[layer]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed3a\n",
      "{'channels': 256, 'channels_used': 140, 'channels_not_used': 116, 'percent_channels_used': 0.546875}\n",
      "mixed3b\n",
      "{'channels': 480, 'channels_used': 243, 'channels_not_used': 237, 'percent_channels_used': 0.50625}\n",
      "mixed4a\n",
      "{'channels': 508, 'channels_used': 393, 'channels_not_used': 115, 'percent_channels_used': 0.7736220472440944}\n",
      "mixed4b\n",
      "{'channels': 512, 'channels_used': 507, 'channels_not_used': 5, 'percent_channels_used': 0.990234375}\n",
      "mixed4c\n",
      "{'channels': 512, 'channels_used': 510, 'channels_not_used': 2, 'percent_channels_used': 0.99609375}\n",
      "mixed4d\n",
      "{'channels': 528, 'channels_used': 526, 'channels_not_used': 2, 'percent_channels_used': 0.9962121212121212}\n",
      "mixed4e\n",
      "{'channels': 832, 'channels_used': 713, 'channels_not_used': 119, 'percent_channels_used': 0.8569711538461539}\n",
      "mixed5a\n",
      "{'channels': 832, 'channels_used': 832, 'channels_not_used': 0, 'percent_channels_used': 1.0}\n",
      "mixed5b\n",
      "{'channels': 1024, 'channels_used': 1024, 'channels_not_used': 0, 'percent_channels_used': 1.0}\n"
     ]
    }
   ],
   "source": [
    "for layer in top_channels:\n",
    "    print(layer)\n",
    "    print(top_channels[layer])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Inspect top channels of the top channels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed3a\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA68AAAD8CAYAAACYapItAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x113d4f828>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed3b\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA68AAAD8CAYAAACYapItAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAKOneAUAAGD0FK8AAACMnuIVAACA0fv/XgmeLscbJdcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x13b452860>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed4a\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6gAAAD9CAYAAACx6lrxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAH+hJREFUeJzt3X/QZXddH/D3hywECCIJrNuQH26cWaSJFdA14PijSAYT\njXVjbdOlysSSTtQGCmrVjbY21sk0/qjamYqdjGLXKmS2WCepWDFEKXUU4gYSJAkxC0lM0iQbYSiK\nbWjCt3/cs+TmyXPuPfc+zyVnn+f1mnnmOffcc7738z3f7/me+7nn3HOrtRYAAAB4uj3j6Q4AAAAA\nEgkqAAAAIyFBBQAAYBQkqAAAAIyCBBUAAIBRkKACAAAwCnMT1Kr68qq6Zerv01X1lqo6papuqKq7\nuv8nT61zRVUdqao7q+r81VYBAACAraAW+R3UqjohyQNJXpnk8iSfbK1dXVUHkpzcWvvRqjo7yTuS\nnJvkxUnek+QlrbXHNz16AAAAtoxFL/E9L8nHWmv3JtmX5GA3/2CSi7rpfUmuba092lq7O8mRTJJV\nAAAA6LVogro/k7OjSbKrtfZgN/1Qkl3d9GlJ7pta5/5uHgAAAPTaMXTBqnpWkm9PcsXa51prraqG\nXys8Ke+yJJclyUknnfTVL33pSxdZHQAAgOPEzTff/JettZ3zlhucoCb5liQfbK093D1+uKpOba09\nWFWnJjnazX8gyRlT653ezXuS1to1Sa5Jkr1797bDhw8vEAoAAADHi6q6d8hyi1zi+7o8cXlvklyf\n5JJu+pIk103N319VJ1bVWUn2JLlpgdcBAABgGxp0BrWqTkry2iTfOzX76iSHqurSJPcmuThJWmu3\nVdWhJLcneSzJ5e7gCwAAwDyDEtTW2meSvHDNvE9kclff9Za/KslVG44OAACAbWPRu/gCAADASkhQ\nAQAAGAUJKgAAAKMgQQUAAGAUJKgAAACMggQVAACAURj0MzMkuw+8a0Pr33P1hZsUCQAAwNbkDCoA\nAACjIEEFAABgFCSoAAAAjIIEFQAAgFGQoAIAADAKElQAAABGQYIKAADAKEhQAQAAGAUJKgAAAKMg\nQQUAAGAUJKgAAACMggQVAACAUZCgAgAAMAoSVAAAAEZBggoAAMAoDEpQq+oFVfXOqvpoVd1RVV9b\nVadU1Q1VdVf3/+Sp5a+oqiNVdWdVnb+68AEAANgqhp5B/fdJfq+19tIkL0tyR5IDSW5sre1JcmP3\nOFV1dpL9Sc5JckGSt1bVCZsdOAAAAFvL3AS1qr44yTcm+dUkaa19trX2qST7khzsFjuY5KJuel+S\na1trj7bW7k5yJMm5mx04AAAAW8uQM6hnJXkkya9V1Yeq6leq6qQku1prD3bLPJRkVzd9WpL7pta/\nv5sHAAAAvYYkqDuSfFWSX26tvSLJZ9JdzntMa60laYu8cFVdVlWHq+rwI488ssiqAAAAbEFDEtT7\nk9zfWvtA9/idmSSsD1fVqUnS/T/aPf9AkjOm1j+9m/ckrbVrWmt7W2t7d+7cuWz8AAAAbBFzE9TW\n2kNJ7quqL+9mnZfk9iTXJ7mkm3dJkuu66euT7K+qE6vqrCR7kty0qVEDAACw5ewYuNybkvxmVT0r\nyceT/JNMkttDVXVpknuTXJwkrbXbqupQJknsY0kub609vumRAwAAsKUMSlBba7ck2bvOU+f1LH9V\nkqs2EBcAAADbzNDfQQUAAICVkqACAAAwChJUAAAARkGCCgAAwChIUAEAABgFCSoAAACjIEEFAABg\nFCSoAAAAjIIEFQAAgFGQoAIAADAKElQAAABGQYIKAADAKEhQAQAAGAUJKgAAAKMgQQUAAGAUJKgA\nAACMggQVAACAUZCgAgAAMAoSVAAAAEZBggoAAMAoSFABAAAYBQkqAAAAozAoQa2qe6rqz6rqlqo6\n3M07papuqKq7uv8nTy1/RVUdqao7q+r8VQUPAADA1rHIGdRvaq29vLW2t3t8IMmNrbU9SW7sHqeq\nzk6yP8k5SS5I8taqOmETYwYAAGAL2sglvvuSHOymDya5aGr+ta21R1trdyc5kuTcDbwOAAAA28DQ\nBLUleU9V3VxVl3XzdrXWHuymH0qyq5s+Lcl9U+ve380DAACAXjsGLvf1rbUHqupLktxQVR+dfrK1\n1qqqLfLCXaJ7WZKceeaZi6wKAADAFjToDGpr7YHu/9Ekv53JJbsPV9WpSdL9P9ot/kCSM6ZWP72b\nt7bMa1pre1tre3fu3Ll8DQAAANgS5iaoVXVSVX3Rsekk35zkI0muT3JJt9glSa7rpq9Psr+qTqyq\ns5LsSXLTZgcOAADA1jLkEt9dSX67qo4t//bW2u9V1Z8mOVRVlya5N8nFSdJau62qDiW5PcljSS5v\nrT2+kugBAADYMuYmqK21jyd52TrzP5HkvJ51rkpy1YajAwAAYNvYyM/MAAAAwKaRoAIAADAKElQA\nAABGQYIKAADAKEhQAQAAGAUJKgAAAKMgQQUAAGAUJKgAAACMggQVAACAUZCgAgAAMAoSVAAAAEZB\nggoAAMAo7Hi6A9iudh9414bLuOfqCzchEgAAgHFwBhUAAIBRkKACAAAwChJUAAAARkGCCgAAwChI\nUAEAABgFCSoAAACjIEEFAABgFCSoAAAAjIIEFQAAgFEYnKBW1QlV9aGq+p3u8SlVdUNV3dX9P3lq\n2Suq6khV3VlV568icAAAALaWRc6gvjnJHVOPDyS5sbW2J8mN3eNU1dlJ9ic5J8kFSd5aVSdsTrgA\nAABsVYMS1Ko6PcmFSX5lava+JAe76YNJLpqaf21r7dHW2t1JjiQ5d3PCBQAAYKsaegb1F5P8SJLP\nTc3b1Vp7sJt+KMmubvq0JPdNLXd/Nw8AAAB6zU1Qq+rbkhxtrd3ct0xrrSVpi7xwVV1WVYer6vAj\njzyyyKoAAABsQUPOoH5dkm+vqnuSXJvkNVX1G0kerqpTk6T7f7Rb/oEkZ0ytf3o370laa9e01va2\n1vbu3LlzA1UAAABgK5iboLbWrmitnd5a253JzY/+oLX23UmuT3JJt9glSa7rpq9Psr+qTqyqs5Ls\nSXLTpkcOAADAlrJjA+teneRQVV2a5N4kFydJa+22qjqU5PYkjyW5vLX2+IYjBQAAYEtbKEFtrb03\nyXu76U8kOa9nuauSXLXB2AAAANhGFvkdVAAAAFgZCSoAAACjIEEFAABgFCSoAAAAjIIEFQAAgFGQ\noAIAADAKElQAAABGQYIKAADAKEhQAQAAGAUJKgAAAKMgQQUAAGAUJKgAAACMggQVAACAUZCgAgAA\nMAoSVAAAAEZBggoAAMAoSFABAAAYBQkqAAAAoyBBBQAAYBQkqAAAAIyCBBUAAIBRkKACAAAwChJU\nAAAARmFuglpVz66qm6rq1qq6rap+spt/SlXdUFV3df9Pnlrniqo6UlV3VtX5q6wAAAAAW8OQM6iP\nJnlNa+1lSV6e5IKqelWSA0lubK3tSXJj9zhVdXaS/UnOSXJBkrdW1QmrCB4AAICtY26C2ib+unv4\nzO6vJdmX5GA3/2CSi7rpfUmuba092lq7O8mRJOduatQAAABsOYO+g1pVJ1TVLUmOJrmhtfaBJLta\naw92izyUZFc3fVqS+6ZWv7+bBwAAAL0GJaittcdbay9PcnqSc6vqK9Y83zI5qzpYVV1WVYer6vAj\njzyyyKoAAABsQQvdxbe19qkkf5jJd0sfrqpTk6T7f7Rb7IEkZ0ytdno3b21Z17TW9rbW9u7cuXOZ\n2AEAANhChtzFd2dVvaCbfk6S1yb5aJLrk1zSLXZJkuu66euT7K+qE6vqrCR7kty02YEDAACwtewY\nsMypSQ52d+J9RpJDrbXfqao/SXKoqi5Ncm+Si5OktXZbVR1KcnuSx5Jc3lp7fDXhAwAAsFXMTVBb\nax9O8op15n8iyXk961yV5KoNRwcAAMC2sdB3UAEAAGBVJKgAAACMwpDvoHKc2H3gXRta/56rL9yk\nSAAAABbnDCoAAACjIEEFAABgFCSoAAAAjIIEFQAAgFGQoAIAADAKElQAAABGwc/M0MvP1gAAAF9I\nzqACAAAwChJUAAAARkGCCgAAwChIUAEAABgFCSoAAACjIEEFAABgFCSoAAAAjIIEFQAAgFGQoAIA\nADAKElQAAABGQYIKAADAKEhQAQAAGIW5CWpVnVFVf1hVt1fVbVX15m7+KVV1Q1Xd1f0/eWqdK6rq\nSFXdWVXnr7ICAAAAbA1DzqA+luSHWmtnJ3lVksur6uwkB5Lc2Frbk+TG7nG65/YnOSfJBUneWlUn\nrCJ4AAAAto65CWpr7cHW2ge76b9KckeS05LsS3KwW+xgkou66X1Jrm2tPdpauzvJkSTnbnbgAAAA\nbC0LfQe1qnYneUWSDyTZ1Vp7sHvqoSS7uunTktw3tdr93TwAAADoNThBrarnJfmtJG9prX16+rnW\nWkvSFnnhqrqsqg5X1eFHHnlkkVUBAADYggYlqFX1zEyS099srf3XbvbDVXVq9/ypSY528x9IcsbU\n6qd3856ktXZNa21va23vzp07l40fAACALWLIXXwrya8muaO19vNTT12f5JJu+pIk103N319VJ1bV\nWUn2JLlp80IGAABgK9oxYJmvS/L6JH9WVbd0834sydVJDlXVpUnuTXJxkrTWbquqQ0luz+QOwJe3\n1h7f9MgBAADYUuYmqK21P0pSPU+f17POVUmu2kBcAAAAbDML3cUXAAAAVkWCCgAAwChIUAEAABgF\nCSoAAACjIEEFAABgFIb8zAxsit0H3rXhMu65+sJNiAQAABgjZ1ABAAAYBWdQOa5t9KysM7IAADAe\nzqACAAAwChJUAAAARkGCCgAAwChIUAEAABgFCSoAAACjIEEFAABgFCSoAAAAjIIEFQAAgFHY8XQH\nAGOy+8C7NlzGPVdfuAmRAADA9uMMKgAAAKPgDCqs2EbPyjojCwDAduEMKgAAAKMgQQUAAGAUJKgA\nAACMwtwEtareVlVHq+ojU/NOqaobququ7v/JU89dUVVHqurOqjp/VYEDAACwtQw5g/qfklywZt6B\nJDe21vYkubF7nKo6O8n+JOd067y1qk7YtGgBAADYsuYmqK219yX55JrZ+5Ic7KYPJrloav61rbVH\nW2t3JzmS5NxNihUAAIAtbNmfmdnVWnuwm34oya5u+rQk759a7v5uHrBJ/GwNAABb1YZ/B7W11qqq\nLbpeVV2W5LIkOfPMMzcaBrCkjSa8iaQXAIDNsexdfB+uqlOTpPt/tJv/QJIzppY7vZv3FK21a1pr\ne1tre3fu3LlkGAAAAGwVy55BvT7JJUmu7v5fNzX/7VX180lenGRPkps2GiRwfHEZMgAAy5iboFbV\nO5K8OsmLqur+JP86k8T0UFVdmuTeJBcnSWvttqo6lOT2JI8luby19viKYgcAAGALmZugttZe1/PU\neT3LX5Xkqo0EBQAAwPaz4ZskAayaGzkBAGwPy94kCQAAADaVM6jAtuRGTgAA4yNBBdgEEl4AgI2T\noAKMkO/dAgDbkQQVYJtwlhcAGDsJKgBLcZYXANhsElQARsNZXgDY3iSoAGxZm53wOmsMAKslQQWA\np9HYkmgJNABPp2c83QEAAABA4gwqADCHs7IAfKFIUAGAL6ixXda8ijJdeg2wHJf4AgAAMAoSVAAA\nAEbBJb4AAMchlyEDW5EEFQCATee7wcAyXOILAADAKDiDCgAAm2AVZ41hu5GgAgDASI3tsuZVlCkp\nZ5oEFQAAGI3tmJQ7+/4E30EFAABgFFaWoFbVBVV1Z1UdqaoDq3odAAAAtoaVJKhVdUKSX0ryLUnO\nTvK6qjp7Fa8FAADA1rCqM6jnJjnSWvt4a+2zSa5Nsm9FrwUAAMAWsKoE9bQk9009vr+bBwAAAOuq\n1trmF1r1D5Jc0Fr7p93j1yd5ZWvtjVPLXJbksu7hlye5c9MD+cJ6UZK/HHF5qyhzO8a4Heu8ijLH\nXt4qytyOMW7HOq+izLGXt4oyx17eKsrcjjFuxzqvosyxl7eKMrdjjKuo8xfal7bWds5baFU/M/NA\nkjOmHp/ezfu81to1Sa5Z0et/wVXV4dba3rGWt4oyt2OM27HOqyhz7OWtosztGON2rPMqyhx7easo\nc+zlraLM7RjjdqzzKsoce3mrKHM7xriKOo/Vqi7x/dMke6rqrKp6VpL9Sa5f0WsBAACwBazkDGpr\n7bGqemOSdyc5IcnbWmu3reK1AAAA2BpWdYlvWmu/m+R3V1X+CG325cqruPxZjOMrbxVlbscYt2Od\nV1Hm2MtbRZnbMUZ1HmeZ2zHG7VjnVZQ59vJWUeZ2jHHLfDVynpXcJAkAAAAWtarvoAIAAMBiWmv+\n5vxlckfiP0xye5Lbkry5m39KkhuS3NX9P7mb/8wkB5P8WZI7klwxtMzuuTcl+Wg3/2e6ed+V5Jap\nv88lefnUOm9LcjTJR9Z5rR9K0pK8aGreFUmOZPLzPucvUOeXJfmTrm7/LcnzF6jzs5PclOTWrsyf\nnFPmC7sY/jrJf+hpm3Xr3bMNNxLjP+wefy7J3jXr9G7LRdt56rkzu3r/i556n5DkQ0l+Z05ffG2S\nm7s635zkNQPL+9kutg8n+e0kLxga34y+c2Umd/M+1oe/tZu/O8n/mZr/Hzdhf5nZ1jNiXGo7LlHe\nkL69bl9cQZ1/qmvnW5L8fpIXd/PPnWqTW5N8xwJtve7+kuRZSX6ti/HWJK8euP/1bceZMS7ab7r5\nX5nJeHRbF+ezB+wvfdtw5ri9TDsPGb8XHCOGxNjXzuuOEwPaZdG+PXccm9F3rsz6487c48EibT2v\n78yIb+mxdpnxe1aM3fNPOaam/xg9c3+eEV/f+LA7Sx4Lkrw8yfu79Q4nOXdFY87S+/SMGGceD2aU\n1zfuDHmvs0g7L/2eNv1jxNxj4IJ9ZxXj2LJ9p69d5u4vS5Q5aDsej39PewDHw1+SU5N8VTf9RUn+\nPMnZSX4myYFu/oEkP91N/+Mk13bTz01yT5LdA8v8piTvSXJi99yXrBPP30nysTXzvjHJV+WpidoZ\nmdys6t50CWr3OrcmOTHJWUk+luSEgfH9aZK/281/Q5KfWqDOleR53fQzk3wgyatmlHlSkq9P8n19\nO9569e7bhhuM8W9n8nu9782TB8WZ23LZdk7yziT/Jf0J6g8meXueGLT7+uIr8sRA9hVJHhhY3jcn\n2dFN//Sx8obEN6POV/YsvzvrfLCykf1lXlvPKG+p7bhEeUP6dl9f3Ow6P39qmX+e7k1hV8aOqXWP\nHns8oMy+/eXyJL92LO5Mko1nDKhz33acGeOM+Pq24Y5M3gS8rHv8wqwZG3v2l3W34Zp1njJuL9nO\nc8fvRcaIgTH2bcd1x4kB7bJoneeOYzPKvDLrjztzjweLtPW8vjMjvqXH2hkx9rXL3P6d9Y+pfcfo\nmfvzjPj6xofdWf5Y8PtJvqWb/61J3ruKMWeD+3RfjDOPBzPK6+uLQ97rLNLOG3lP29cX5x4DF+k7\nA9tl0RiX7Tt97TJ3f1mizEHb8Xj8c4nvAK21B1trH+ym/yqTT5BOS7Ivk0+V0v2/6NgqSU6qqh1J\nnpPks0k+PbDM709ydWvt0e65o+uE9Lok164p731JPrnOsr+Q5Ee6mI7Zl8lg82hr7e5MPok/d2B8\nL0nyvm6xG5J85wJ1bq21v+4ePrP7a31lttY+01r7oyT/d516zap33zZcOsbW2h2ttTvXCWHmtlym\nnavqoiR3Z/Lp2VNU1elJLkzyK2vieEpfbK19qLX2v7r5tyV5TlWdOK+81trvt9Ye6x6+P5PfMh4U\n34w6L22J7TizrRfdp+dtxyXKG9K3+/aXTa1za216HzipKyettb+Z6gPPPjZ/TYx9ZfbtL2cn+YOp\nuD+VZO9UeX117tuOM2Ncot98c5IPt9Zu7eZ/orX2+HSZPfvLuttwjaeM23Pq3Bfj3PF7kTFiYIx9\n7bzuODGgXRaq85BxbEaZfeYeDxZs65l9Z8axZemxdkaMfWXO7d89x9S+4/7M/XlGfH3jw1wz9umW\n5PndYl+c5Fh/2dQxZ41F9+l1Y5x3PJjRd/r64pD3Oou089LvaWeMEXOPgRvoO5syjmXJvjOjXebu\nL4uWOWQ7Hq8kqAuqqt2ZfJr7gSS7WmsPdk89lGRXN/3OJJ9J8mCSv0jyc6219ZLH9cp8SZJvqKoP\nVNX/qKqvWWeVf5TkHQNi3ZfJJ823rnnqtCT3TT2+PzMSiDXx3ZbJwJ1MLrU4o5seVOeqOqGqbsnk\nE/UbWmuzylxW3zbcSIx9Bm/LIe1cVc9L8qNJfnLGa/5iJh86fG5qXl9fnPadST547M3fnPKmvSHJ\nf18gvs9bU+ckeVNVfbiq3lZVJ08telZV3dJti29YoMwNtfU65W1kOy5bXq+evrjZdU5VXVVV92Vy\nmdRPTC33yqo6ding900dxOeW2ePWJN9eVTuq6qwkX501+3tPnXu349AYB/ablyRpVfXuqvpgVf3I\nOkWtu7/0bcMpveP2gu08ZMxZdoyYe2yZ0c6fHye65Wa2y4J1nta7/80Yu9cbd4bsL4u09dy+M+DY\nssxYO3j8HhJjj75j9Nz9eUB8ay17LHhLkp/t2uXnMrkMflCMi445Uxbdp/tinKuv7/T0xYXeg07Z\n0Pu7qVh3Z8AYMcCifeeYzRrHlu07fe0yZH9ZtMwtS4K6gO6A8VtJ3rLm04y01o59OpZMPs1+PMmL\nM7kE64eq6ssGlrkjk+8+vCrJDyc5VFU1tfwrk/xNa+0jc2J9bpIfywY78TrxvSHJP6uqmzO5ROKz\n3aKD6txae7y19vJMPqE6t6q+YkaZy+rbhhuJcUMWaOcrk/zC1Cdoa8v5tiRHW2s3973Wmr54bL1z\nMrl05XsXKa+qfjzJY0l+s5s1M741666t8y8n+bJMvofzYJJ/1y36YJIzu23+g0neXlXPX6fIRbbj\noLZeYJ8+tvy623HZ8ubp6YubXufW2o+31s7IpJ3fOPX6H2itnZPka5JcUVXPXrTea7wtk4TqcCZv\nPv64i3lenaeff9J2HBLjAv1mRyaXS31X9/87quq8qXJ695e+bditN3PcXrCdZ9rAGDH32NLXzuuM\nE3PbZZk6z9v/esrsG3dm7i9LtPXMvjMjvr5teGXmjLVLjN9zY+zRd4yeuT8P6YtrbORY8P1JfqBr\nlx9I8qtDYkwWH3O6119mn+6Lca6+GHv64uD3oGts6P1dstgYMcsSfefYeps5ji3dd3raZW55S5S5\ndbURXGd8PPxlcpr93Ul+cGrenUlO7aZPTXJnN/1LSV4/tdzbklw8sMzfS/JNU48/lmTn1ONfSPJj\nPTHuTvd9gkyuwT+ayXcF7slkp/uLJH8rk0/trpha791JvnZIfGuef0mSmxap85r1fyJrvlczXebU\nvO/J7O8ofL7es7bhZsSYp35nZu62XKSdk/zPqTb7VCaX4Lxxarl/m8kAd08mn+r+TZLf6OuL3ePT\nM/muxdetU791y5va7n+S5LlTy8+Mb4G+86Q2W/Pck7bxkttxblv3lLfUdlymvCF9e72+uNl1XvP8\nmTPa5Q+Gtsu8tpx6/o+TnD2gzjO346wYF+w3+5McnJr/r5L88JD9ZdY2zIxxe4l2njnm9MU4oC/O\njLGvnbPOODG07wyt85D9r6/MNfN254lj5Mz9ZdG2ntd3ZsW33jbMgLF2Vow9ZQ6KMbPH56cco/v2\n53nbMPPHh3WfX68vJvnfeeKnEyvJp4fEOKMvbmh/6SlzZowZeDzo6dvTfXHoe9BB7bxAeQuPEX11\nXrbvzGuXZWJctO/0tcui5S1S5tC+czz9Pe0BHA9/3UDy60l+cc38n82Tv0R/7I6DP5onvgh9UiZ3\nDPvKgWV+X5J/002/JJNLuY4NaM/I5G6EX9YT56zB5p48cZOkc/Lkm2x8PE+9UUJffF8yFcuvJ3nD\nAnXemSfujvacTA7A39ZX5tR6M3e8tfXu24YbiXHq+ffmyQnqzG25TDtPLXNlZt8Y49V58l0b1+uL\nL+ji+/sD+vl0eRd022fnjOXXjW9GnU+dmv6BPHHThZ3HtlkmZzoeSHLKRvaXeW09o7yltuOi5Q3p\n2319cQV13jM1/aYk7+ymz8oTN4740ky+L/WiIfWesb88N8lJ3fRrk7xvYJ372mVmjEv0m5OTfLCL\nc0cmN+y5cMD+su427B7PG7cXbee54/ciY8TAGPu247rjxIB2WbTOc8exGWX2jTtzjweLtPW8vjMj\nvqXH2hkx9rXLoP6dpx5T+477M/fnvvhmjA8bORbcke6uqEnOS3LzKsacDe7T68Y4td73ZP1kra+8\nvr44qG8v0M4beU87s3/31XnRvjOwXRYdx5btO33tMnd/WbTMRbbj8fb3tAdwPPxlcilMyxO3eL4l\nkzuwvTDJjZnchvw96QbSJM/L5K57t3Wdfr1PKPvKfFYmn3Z/JJMDyWum1nl1kvf3xPiOTC6P+X+Z\nfOp06Zrn78mT3yD8eCafTt+Z7q5yA+N7cyafYv95kqvzRPI8pM5fmcntwj/c1e8nuvnrljkV9ycz\nuYX2/VnzadN69e7bhhuM8Tu68h9N8nCSdw/Zlsu089S6V2Z4gtrXF/9lJt8dmb79+lPuDL1OeUcy\neXM463b/68Y3o87/OZPvo304yfV54tPp7+za5JZuW/y9je4v89p6RnlLbcdFyxvYt/v64mbX+be6\nsj6cyU8LnNYt//o17XLRAu2y7v6SyRuiOzN5o/aeJF86sM597TIzxhnxzRpnv7sr8yNZ84HCjP1l\n3W04b9xepp2HjN+LjBEDY+zbjuuOEwPaZdG+PXccm1Fm37gz93iwRFv39p0Z8S091i4zfs+KsXt+\nvWNq33F/d2bszzPi6xsfNnIs+PpM7op6aybfK/zqVYw5G9yn142xzTkezCivb+we8l5nkXbeyHva\nWX2xt86L9p0VjWPL9p2+dplZ3jJlLrIdj7e/Y50PAAAAnlZukgQAAMAoSFABAAAYBQkqAAAAoyBB\nBQAAYBQkqAAAAIyCBBUAAIBRkKACAAAwChJUAAAARuH/Ay31+xwxgzT6AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x13b448198>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed4b\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6gAAAD8CAYAAAB6tolUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x13aed40f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed4c\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6gAAAD8CAYAAAB6tolUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUAEAAJgECSoAAACT8P8Bn3HdDIC0/nEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11439b518>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed4d\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6gAAAD8CAYAAAB6tolUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11452ca90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed4e\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6gAAAD8CAYAAAB6tolUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSZJ6wQRVkiRJktQLJqiSJEmSpF74/1MhcYKgug0HAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11d928780>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed5a\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6gAAAD8CAYAAAB6tolUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYBQkqAAAAIyCBBUAAIBR+P+njwnjgwFLsgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11d74cbe0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mixed5b\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6IAAAD8CAYAAABtlBmdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nFAAAgKb+PzcVuYFwg+5OAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1362efcc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "k = 4\n",
    "end_idx = 40\n",
    "\n",
    "top_channels = {}\n",
    "\n",
    "for layer in list(layers.keys()):\n",
    "    print(layer)\n",
    "    top_channels[layer] = {\n",
    "        'channels': layers[layer],\n",
    "        'channels_used': 0,\n",
    "        'channels_not_used': 0,\n",
    "        'percent_channels_used': 0\n",
    "    }\n",
    "    all_channels = []\n",
    "    \n",
    "    for c in imagenet:\n",
    "        for ch in c['topChannels'][layer][0:k]:\n",
    "            all_channels.append(ch['channel'])\n",
    "            \n",
    "    (channels, counts) = np.unique(all_channels, return_counts=True)\n",
    "    sorted_counts, sorted_channels = (list(t) for t in zip(*sorted(zip(counts, channels), reverse=True)))\n",
    "\n",
    "    plt.figure(figsize=(16,4))\n",
    "    plt.bar(\n",
    "        range(len(sorted_channels[0:end_idx])),\n",
    "        np.array(sorted_counts[0:end_idx]),\n",
    "        align='center'\n",
    "    )\n",
    "    plt.xticks(range(len(sorted_channels[0:end_idx])),[str(x) for x in sorted_channels[0:end_idx]])\n",
    "    plt.show()\n",
    "    \n",
    "    top_channels[layer]['channels_used'] = int(len(np.unique(sorted_channels)))\n",
    "    top_channels[layer]['channels_not_used'] = layers[layer] - top_channels[layer]['channels_used']\n",
    "    top_channels[layer]['percent_channels_used'] = top_channels[layer]['channels_used'] / layers[layer]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}