25Lights2.2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import the Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from PIL import Image\n",
    "\n",
    "#extract the bit pattern from the filename\n",
    "def decodeFileName(hexString):\n",
    "    number = np.zeros(25, dtype=np.bool)\n",
    "    bit = 0\n",
    "    for i in reversed(range(8)):\n",
    "        nibble = int(hexString[i], 16)\n",
    "        for j in range(4):\n",
    "            if (bit < 25):\n",
    "                number[bit] = (nibble & 0x1)        \n",
    "                nibble = nibble >> 1\n",
    "                bit = bit + 1\n",
    "                \n",
    "    return number\n",
    "\n",
    "#take an image and return an NP float arrray\n",
    "def img_to_float_array(img):\n",
    "    img_array = np.asarray(img, dtype='float32')\n",
    "    img_array = img_array.astype('float32') / 255.\n",
    "    img_array = img_array.reshape((img_array.shape[0], img_array.shape[1], 1))\n",
    "    return img_array\n",
    "\n",
    "#use some BASH commands to load files randomly\n",
    "def random_files(numMatches, directory):\n",
    "    files = !ls {directory} |sort -R |tail -{numMatches}\n",
    "    reutrnFiles = []\n",
    "    for i in range(numMatches):\n",
    "        reutrnFiles.append(files[i])\n",
    "    return files\n",
    "    \n",
    "#load the images and lables\n",
    "def loadImageBatch(batchSize=32, test=False):\n",
    "    images = np.zeros((batchSize, imageSize, imageSize, 1))\n",
    "    labels = np.zeros((batchSize, 25))\n",
    "    \n",
    "    directory = \"./images2/train/\"\n",
    "    \n",
    "    if (test):\n",
    "        directory = \"./images2/test/\"\n",
    "        \n",
    "    imageFiles = random_files(batchSize, directory)\n",
    "        \n",
    "    for i in range(batchSize):\n",
    "        fileName = imageFiles[i]\n",
    "        labels[i] = decodeFileName(fileName[0:8])\n",
    "        fileName = directory+imageFiles[i]\n",
    "        images[i] = img_to_float_array(Image.open(fileName))\n",
    "    \n",
    "    return (images, labels)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from keras.callbacks import TensorBoard\n",
    "import glob, os\n",
    "\n",
    "batch_size = 32\n",
    "\n",
    "#tensorboard callback\n",
    "tensorboard = TensorBoard(log_dir='/tmp/LEDS',\n",
    "                          histogram_freq=0,\n",
    "                          write_graph=True,\n",
    "                          write_images=False)\n",
    "\n",
    "for i in range(0, 30000):\n",
    "    training_batch = loadImageBatch(batch_size)\n",
    "    validation_batch = loadImageBatch(batch_size, test=True)\n",
    "    model.fit(training_batch[0],\n",
    "              training_batch[1],\n",
    "              epochs=1,\n",
    "              batch_size=batch_size,\n",
    "              validation_data=validation_batch,\n",
    "              callbacks=[tensorboard])\n",
    "    print(\"run {}\".format(i))\n",
    "\n",
    "model.save_weights(\"./averagePoolWeights.h5\")"
   ]
  }
 ],
 "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": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}