Preprocessing - 15 June 2017.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "from scipy import ndimage, misc\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import glob, os\n",
    "import sys\n",
    "import time\n",
    "\n",
    "import urllib\n",
    "import requests\n",
    "import json\n",
    "import time\n",
    "import datetime\n",
    "\n",
    "# import the necessary packages\n",
    "import argparse\n",
    "import datetime\n",
    "import time\n",
    "import cv2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#########################################\n",
    "# PRE-STEP - GET NEST CAM ACCESS TOKEN  #\n",
    "#########################################\n",
    "\n",
    "import requests\n",
    "import json\n",
    "\n",
    "access_token = requests.post('https://api.home.nest.com/oauth2/access_token', data={'code':'xxxxxxx', 'client_id':'xxxxx', 'client_secret':'xxxxx', 'grant_type':'authorization_code'})\n",
    "print access_token.json()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#########################################\n",
    "# PRE-STEP - TEST THE TOKEN             #\n",
    "#########################################\n",
    "\n",
    "import requests\n",
    "import json\n",
    "\n",
    "access_token2 = \"c.xxxxxx\"\n",
    "\n",
    "result = requests.get('https://developer-api.nest.com/?auth='+access_token2)\n",
    "#print (result)\n",
    "\n",
    "#print(result.json())\n",
    "result.json()[\"devices\"][\"cameras\"][\"xxxxx\"][\"snapshot_url\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "#########################################\n",
    "# STEP A - CAPTURE FRAMES FROM NEST CAM #\n",
    "#########################################\n",
    "\n",
    "access_token2 = \"c.xxxxx\"\n",
    "image_count = 3600\n",
    "st=0\n",
    "\n",
    "for i in range(7000):\n",
    "    \n",
    "    try:\n",
    "        st = str(datetime.datetime.now())\n",
    "        \n",
    "        #retreive the image\n",
    "        result = requests.get('https://developer-api.nest.com/?auth='+access_token2)\n",
    "        snapshot_url = result.json()[\"devices\"][\"cameras\"][\"xxxxx\"][\"snapshot_url\"]\n",
    "\n",
    "        #save image\n",
    "        filenum = 900000+i\n",
    "        filename = \"camframes/5-31/\"+str(filenum)+\".jpg\"\n",
    "        urllib.urlretrieve(snapshot_url, filename)\n",
    "        \n",
    "    except IOError as e: \n",
    "        print (\"I/O error\")\n",
    "        print (st)\n",
    "        pass\n",
    "    \n",
    "    except KeyError as e2:\n",
    "        #print (\"KeyError (%d): %s\"%(e2.errno, e2.strerror))\n",
    "        print(e2)\n",
    "        print(\"Keyerror\")\n",
    "        print (st)\n",
    "        pass\n",
    "    \n",
    "    except:\n",
    "        print (\"other unexcepted error\")\n",
    "        print (st)\n",
    "        pass\n",
    "    \n",
    "    #time_delay\n",
    "    time.sleep(10)\n",
    "    \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "##########################################\n",
    "# STEP B - IMAGE SUBTRACTION, DOWNSAMPLE #\n",
    "##########################################\n",
    "\n",
    "#things to do before running this file:\n",
    "#  1: set input/output directory\n",
    "#  4: set filenum according to the date of directory\n",
    "\n",
    "dates = [\"5-17\",\"5-18\",\"5-25\",\"5-26\",\"5-27\",\"5-28\"]\n",
    "filenumidx = [517,518,525,526,527,528]\n",
    "\n",
    "#loop through directories\n",
    "for idx, date in enumerate(dates):\n",
    "\n",
    "    # Settings\n",
    "    dir = \"camframes/\"+str(date)+\"/*.jpg\"\n",
    "    dir_out=\"camframes/class/stageB/photos3/\"\n",
    "    thresh_level = 15 \n",
    "    filenum = filenumidx[idx]*100000 # 50900000 if may 9th, 531 if may 31st \n",
    "\n",
    "\n",
    "    # initialize variables\n",
    "    prevFrame = None\n",
    "    total_files = len(glob.glob(dir))\n",
    "    movement_graph = [0]* total_files\n",
    "    class_num = 0\n",
    "    print(total_files)\n",
    "\n",
    "    for i,file in enumerate(glob.glob(dir)):\n",
    "\n",
    "        #grab first image\n",
    "        try:\n",
    "            thisFrame = misc.imread(file)\n",
    "            thisFrame_gr = cv2.cvtColor(thisFrame, cv2.COLOR_BGR2GRAY)\n",
    "            thisFrame_gr_blur = cv2.GaussianBlur(thisFrame_gr, (21, 21), 0)\n",
    "            thisFrame_box = cv2.cvtColor(thisFrame, cv2.COLOR_BGR2GRAY)\n",
    "        except:\n",
    "            \"file error - skipping\"\n",
    "            continue\n",
    "\n",
    "\n",
    "        #if prevFrame is empty, then fill prevFrame and go to next loop\n",
    "        if prevFrame == None:\n",
    "            prevFrame = thisFrame\n",
    "            prevFrame_gr_blur = thisFrame_gr_blur\n",
    "            continue\n",
    "\n",
    "\n",
    "        # compute diff prev & FirstFrame, max 0\n",
    "        frameDelta1 = cv2.subtract(prevFrame_gr_blur, thisFrame_gr_blur)\n",
    "        frameDelta2 = cv2.subtract(thisFrame_gr_blur, prevFrame_gr_blur)\n",
    "        frameDelta = cv2.max(0, frameDelta1)\n",
    "\n",
    "        # set threshold\n",
    "        thresh = cv2.threshold(frameDelta, thresh_level, 255, cv2.THRESH_BINARY)[1]\n",
    "\n",
    "        # dilate the thresholded image to fill in holes, then find contours on dilated image\n",
    "        thresh = cv2.dilate(thresh, None, iterations=2)\n",
    "        (cnts, _) = cv2.findContours(thresh.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)\n",
    "\n",
    "        #turn into scalar\n",
    "        thresh_ones = np.zeros(thresh.shape)\n",
    "        thresh_ones[thresh==255] = 1\n",
    "        movement_graph[i]= np.sum(thresh_ones)\n",
    "\n",
    "\n",
    "        # Write images to correct directory\n",
    "        if movement_graph[i] > 10000 and movement_graph[i]<200000: \n",
    "            print (file)\n",
    "            plotout = 1\n",
    "            misc.imsave(dir_out+str(filenum+i)+\".png\", thisFrame)\n",
    "            #misc.imsave(dir_out+str(int(filenum+i))+\"-prev.png\", prevFrame)\n",
    "            #misc.imsave(dir_out+str(int(filenum+i))+\"-gr.png\", thisFrame_gr)\n",
    "            #misc.imsave(dir_out+str(int(filenum+i))+\"-delta.png\", frameDelta)\n",
    "        else:\n",
    "            plotout = 0\n",
    "\n",
    "\n",
    "        #Print loop status\n",
    "        sys.stdout.write('\\r'+'Processing File: '+str(i)+ ' of '+ str(total_files))\n",
    "\n",
    "        # Get ready for next loop\n",
    "        prevFrame = thisFrame\n",
    "        prevFrame_gr_blur = thisFrame_gr_blur\n",
    "\n",
    "    #print(movement_graph)\n",
    "    plt.plot(movement_graph)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "##########################################\n",
    "# STEP C - MANUAL ASSIGN TO FIVE CLASSES #\n",
    "##########################################"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1092\n",
      "2070\n",
      "211\n",
      "1006\n",
      "6129\n"
     ]
    }
   ],
   "source": [
    "##########################################\n",
    "# STEP D - ASSIGN TO TRAIN, VAL, TEST    #\n",
    "##########################################\n",
    "\n",
    "from scipy import ndimage, misc\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import glob, os\n",
    "import sys\n",
    "import time\n",
    "\n",
    "# import the necessary packages\n",
    "import argparse\n",
    "import datetime\n",
    "import time\n",
    "import cv2\n",
    "%matplotlib inline\n",
    "\n",
    "classdirname = [\"sit\", \"stand\", \"care\", \"empty\", \"sleep\"]\n",
    "valratio = 2  # out of 10\n",
    "testratio = 2 # out of 10\n",
    "trainratio = 10-valratio-testratio\n",
    "\n",
    "for i, classname in enumerate(classdirname):\n",
    "\n",
    "    # Settings\n",
    "    dir = \"camframes/class/stageC/batch2/\"+classname+\"/*.png\"\n",
    "    dir_out_train=\"camframes/class/stageD/baby7/train/\"+classname+\"/\"\n",
    "    dir_out_val = \"camframes/class/stageD/baby7/val/\"+classname+\"/\"\n",
    "    dir_out_test = \"camframes/class/stageD/baby7/test/\"+classname+\"/\"\n",
    "    count = 0\n",
    "\n",
    "    # initialize variables\n",
    "    total_files = len(glob.glob(dir))\n",
    "    print (total_files)\n",
    "\n",
    "    for i,file in enumerate(glob.glob(dir)):\n",
    "\n",
    "        #test for filename, extract file name \n",
    "        filename = file[(file.rfind(\"/\")+1):]\n",
    "\n",
    "        if \"-\" in filename:\n",
    "            pass\n",
    "            #print (filename)\n",
    "        else:\n",
    "            count += 1\n",
    "            random = np.random.randint(1,10)\n",
    "            \n",
    "            if random < (trainratio+1):\n",
    "                dir_out = dir_out_train\n",
    "            elif random < (trainratio+valratio+1):\n",
    "                dir_out = dir_out_val\n",
    "            else:\n",
    "                dir_out = dir_out_test\n",
    "            \n",
    "            thisFrame = misc.imread(file)\n",
    "            misc.imsave(dir_out+str(filename), thisFrame)\n",
    "        \n",
    "       "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "##########################################\n",
    "# STEP E - AUGMENT IMAGES IN TRAIN    #\n",
    "##########################################\n",
    "\n",
    "from scipy import ndimage, misc\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import glob, os\n",
    "import sys\n",
    "import time\n",
    "\n",
    "# import the necessary packages\n",
    "import argparse\n",
    "import datetime\n",
    "import time\n",
    "import cv2\n",
    "%matplotlib inline\n",
    "\n",
    "classdirname = [\"sit\", \"stand\", \"care\", \"empty\", \"sleep\"]\n",
    "#classdirname = [\"care\"]\n",
    "dir_in_train = \"camframes/class/stageD/baby7/train/\"\n",
    "dir_out_train = \"camframes/class/stageD/baby7aug/train/\"\n",
    "\n",
    "\n",
    "for j, classname in enumerate(classdirname):\n",
    "\n",
    "    # Settings\n",
    "    dir_in_train_class=dir_in_train+classname+\"/\"\n",
    "    dir_out_train_class=dir_out_train+classname+\"/\"\n",
    "    print(dir_in_train_class)\n",
    "\n",
    "    # initialize variables\n",
    "    count = 0\n",
    "    total_files = len(glob.glob(dir_in_train_class+\"*.png\"))\n",
    "    print (total_files)\n",
    "\n",
    "    for i,file in enumerate(glob.glob(dir_in_train_class+\"*.png\")):\n",
    "\n",
    "        try:\n",
    "            img = misc.imread(file)\n",
    "            hflip_img = cv2.flip(img,1)\n",
    "            vflip_img = cv2.flip(img,0)\n",
    "            bflip_img = cv2.flip(img,-1)\n",
    "        except:\n",
    "            \"file error - skipping\"\n",
    "            continue\n",
    "        \n",
    "        print(file.split(\".\")[0].replace(\"baby7\",\"baby7aug\"))\n",
    "        filepath = file.split(\".\")[0].replace(\"baby7\",\"baby7aug\")\n",
    "        \n",
    "        misc.imsave(filepath+\"-hflip.png\", hflip_img)\n",
    "        misc.imsave(filepath+\"-vflip.png\", vflip_img)\n",
    "        misc.imsave(filepath+\"-bflip.png\", bflip_img)\n",
    "\n",
    "\n",
    "\n",
    "        #Print loop status\n",
    "        sys.stdout.write('\\r'+'Processing File: '+str(i)+ ' of '+ str(total_files))\n",
    "\n",
    "            \n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "##########################################\n",
    "# CONVERT DIFFBABY TO GREYSCALE .        #\n",
    "##########################################\n",
    "\n",
    "from scipy import ndimage, misc\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import glob, os\n",
    "import sys\n",
    "import time\n",
    "\n",
    "# import the necessary packages\n",
    "import argparse\n",
    "import datetime\n",
    "import time\n",
    "import cv2\n",
    "%matplotlib inline\n",
    "\n",
    "classdirname = [\"sit\", \"stand\", \"care\", \"empty\", \"sleep\"]\n",
    "#classdirname = [\"care\"]\n",
    "dir_in = \"camframes/class/stageD/diffbaby/\"\n",
    "dir_out = \"camframes/class/stageD/diffbabygrey/\"\n",
    "\n",
    "\n",
    "for j, classname in enumerate(classdirname):\n",
    "\n",
    "    # Settings\n",
    "    dir_in_class=dir_in+classname+\"/\"\n",
    "    dir_out_class=dir_out+classname+\"/\"\n",
    "    print(dir_in_class)\n",
    "\n",
    "    # initialize variables\n",
    "    count = 0\n",
    "    total_files = len(glob.glob(dir_in_class+\"*.png\"))\n",
    "    print (total_files)\n",
    "\n",
    "    for i,file in enumerate(glob.glob(dir_in_class+\"*.png\")):\n",
    "\n",
    "        try:\n",
    "            img = misc.imread(file)\n",
    "            img_gr = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n",
    "            \n",
    "        except:\n",
    "            \"file error - skipping\"\n",
    "            continue\n",
    "        \n",
    "        #print(file.split(\".\")[0].replace(\"diffbaby\",\"diffbabygrey\"))\n",
    "        #filepath = file.split(\".\")[0].replace(\"diffbaby\",\"diffbabygrey\")\n",
    "        \n",
    "        print(dir_out+classname+\"/\"+str(i)+\"-gr.png\")\n",
    "        \n",
    "        misc.imsave(dir_out+classname+\"/\"+str(i)+\"-gr.png\", img_gr)\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "        #Print loop status\n",
    "        sys.stdout.write('\\r'+'Processing File: '+str(i)+ ' of '+ str(total_files))\n",
    "\n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}