CNN_for_Model_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CNN zur Modelerkennung\n",
    "@Martin Kraus\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 406,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "##eigentliches CNN\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.preprocessing import LabelBinarizer\n",
    "import os\n",
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format = 'retina'\n",
    "input_dir = 'C:/Users/Martin/Untitled Folder 1/'\n",
    "\n",
    "directories = os.listdir(input_dir)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 407,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#load preloaded data\n",
    "images = np.load('X_train.npy') #Saving train image arrays\n",
    "labels = np.load('Y_train.npy')#Saving train labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 408,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(995, 32, 32, 3)"
      ]
     },
     "execution_count": 408,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.shape(images)\n",
    "#plt.imshow(images[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 409,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(995, 32, 32, 3)"
      ]
     },
     "execution_count": 409,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.shape(images)\n",
    "#test_images= images\n",
    "#images =np.reshape(test_images, [1007,32,32,3])\n",
    "#np.shape(test_images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 410,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#normalise images\n",
    "def normalize(x):\n",
    "    \"\"\"\n",
    "    Normalize a list of sample image data in the range of 0 to 1\n",
    "    : x: List of image data.  The image shape is (32, 32, 3)\n",
    "    : return: Numpy array of normalize data\n",
    "    \"\"\"\n",
    " # TODO: Implement Function\n",
    "    x_min = np.min(x)\n",
    "    x_max = np.max(x)\n",
    "    return (x - x_min) / (x_max - x_min)  \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 411,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#normalize images\n",
    "images = normalize(images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 412,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import LabelBinarizer\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "\n",
    "isEncoded = False\n",
    "enc = OneHotEncoder()\n",
    "\n",
    "def one_hot_encode(x):\n",
    "    \"\"\"\n",
    "    One hot encode a list of sample labels. Return a one-hot encoded vector for each label.\n",
    "    : x: List of sample Labels\n",
    "    : return: Numpy array of one-hot encoded labels\n",
    "    \"\"\"\n",
    "    global isEncoded\n",
    "    \n",
    "    if not isEncoded:\n",
    "        enc.fit(x)  \n",
    "        \n",
    "        \n",
    "    return enc.transform(x).toarray()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 413,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import OneHotEncoder\n",
    "#enc_2 = OneHotEncoder()\n",
    "#enc_2.fit(labels)  \n",
    "#enc_2.transform(labels).toarray()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 414,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "labels = one_hot_encode(labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 415,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[  0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17\n",
      "  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35\n",
      "  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53\n",
      "  54  55  56  57  58  59  60  61  62  63  64  65  66  67  68  69  70  71\n",
      "  72  73  74  75  76  77  78  79  80  81  82  83  84  85  86  87  88  89\n",
      "  90  91  92  93  94  95  96  97  98  99 100 101 102 103 104 105 106 107\n",
      " 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125\n",
      " 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143\n",
      " 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161\n",
      " 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179\n",
      " 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197\n",
      " 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215\n",
      " 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233\n",
      " 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251\n",
      " 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269\n",
      " 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287\n",
      " 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305\n",
      " 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323\n",
      " 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341\n",
      " 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359\n",
      " 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377\n",
      " 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395\n",
      " 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413\n",
      " 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431\n",
      " 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449\n",
      " 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467\n",
      " 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485\n",
      " 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503\n",
      " 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521\n",
      " 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539\n",
      " 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557\n",
      " 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575\n",
      " 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593\n",
      " 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611\n",
      " 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629\n",
      " 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647\n",
      " 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665\n",
      " 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683\n",
      " 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701\n",
      " 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719\n",
      " 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737\n",
      " 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755\n",
      " 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773\n",
      " 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791\n",
      " 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809\n",
      " 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827\n",
      " 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845\n",
      " 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863\n",
      " 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881\n",
      " 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899\n",
      " 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917\n",
      " 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935\n",
      " 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953\n",
      " 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971\n",
      " 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989\n",
      " 990 991 992 993 994]\n",
      "[185 631 937 881 981 748 850 385 626 698  72 475  59 106 869 945 427 376\n",
      " 305  14 161 644 217 575  39 287 141 749 815 940 801  69 125 844 656 764\n",
      " 804 495 721 677 487 630 228  56 392  63 673 927 974 526 789  88 589 312\n",
      " 236 901 824 453 633 648 435  36 683 148 977 902 525 396 918 381 761 817\n",
      " 224 237 608 215 501 898 279 516 497 101 652   7 894 777 549 733 836 847\n",
      " 566 223  53 724 579 730 456 605 330 891  86 522 230  79 819 542 372 990\n",
      " 108 118 833  23 532 578 685 281 831 524 678 515 638 886 441 434 691 251\n",
      " 598 418 340 983   9 982 347 473 489 132 404 978 843 963 653 481 388 854\n",
      " 976 258 205 448 465 313 518   5 466 712  44 966 166 717 620 750 603 136\n",
      " 829 531 233 455 672 225 681 350  54 711 949  13 971 131 781 191 592 159\n",
      " 402 451 538 959 373 512 632 553 438 825 628 798 470 812 627 425 289  90\n",
      " 262  51 508 754  42  66 146 853 491 878 766 308 739 351 594 703 755 309\n",
      " 979 139 222  22 905 787 670 212 859 585  75 409 241  81 887 858 149 153\n",
      "  49 600 521  45 433 378  89 234 543   8 327 514 660 911 668  21 171 447\n",
      "  84 197 199 876 778 571 762 363 220 452  92  71 965 227 391 430 689 326\n",
      " 476 968 483 484  25 194 368 616  48 701 896 994 606 431 415  62 369  99\n",
      " 860 960 861 728 115 559 599 179  61 474 208 621 541 793 500 890 729 952\n",
      " 509 201 671 814 405  98 398 443 112 655 643 877 202 155 321 134 152 271\n",
      " 760 536 849 692 584 639  12 165 200 642 231 403 799 428 102 247 238 297\n",
      " 727 564 274 511 242 932 842 570   6 738 127 291 229 333 855 142 885 657\n",
      " 457 928 278 273 897 360 504 629 958  80  17 365 240 916 371 461 534 872\n",
      " 967 210 383 664 422 985 706   0 345 852 116 722 783 380 151 935 550 343\n",
      "  35 920 834 875   2 602 601 432 304 580 533 962 718 180 397  64 923 779\n",
      " 782 419 837  31 723 948 593  58  52  34 399  41 282 193  78 320 992 172\n",
      " 209 557  26 665 548 769 163 756 244 715 827 265  28 436 757 344 389 513\n",
      " 458 864 666 184  73 684 805 207 386 583 808 529 624 919 568 561 658 674\n",
      " 143 110 537 573 243 614 694 574  67 346 354  96 270  37 841 377 239 444\n",
      " 410 913  33 613 213 807 348 931 486 651 407 104 577 214 731 640 597 704\n",
      " 794 741 922 277 467 591 182 493 582  29 186 909 503  95 338 196 669 482\n",
      " 797 111 820 906 292 510 263 488 176 253 334 469 295 130 109 164 775  11\n",
      " 884 560 280 192 359 156 426 370 315 740 393 539 468 135 735 154 625 747\n",
      "  19 709 169 695  32 254 339 950 498 569 506 828 276 567   4 663 540 107\n",
      "  85 763 138  87 420  46 647 145 686 480 122 449 645 325 946 332 870 527\n",
      " 450 158 177 816 904 770 555 773 190 736 596 607 636 969 765 216 650 563\n",
      " 290 114 296  20 838  76 803 821 286 556 337 355 576 667 128 710 687 173\n",
      " 168  83 300 972 893 552 394 707 623 248 113  77 517 551 811 780 611 203\n",
      " 257 442 758 129 830 961 361 411 267 745 319 260 895 809 547 245 502 408\n",
      " 301 924 942 954 679 413 907 317 986 382  16 437 785 198 851  15 264 221\n",
      " 472 791 401 953 284 581 395 615 944 767 318 137 934 494  94 255  74 991\n",
      " 675 790 612 545 856  10 743  55 699  18 285 951  70 964 445 956 412  50\n",
      " 725 826 492 528 353 329 889 696 646 759 987 294 520 786 266 867 331 863\n",
      " 268 464 424 356 800 989  43 387 558 857 160 941 324 120 933 459 888 562\n",
      " 478 218  93 126 802 554 649 810 288 900 610 925 634 178 272 796 121 162\n",
      " 306 170 328 818 150 771 753 776 910 250  60 936 726 364 117 822 261 883\n",
      " 204 390 737  65 690 310 732 832 174 980 546 835 440 973 680 379 588   3\n",
      " 195 414 485 302 930 572 908 259 275 400 421 507 609 879 772 823 103 975\n",
      " 439 299 293 384 181  68 873 175 619 335 140 211 519 499 915 590  24 124\n",
      " 993 357 871 947 845 874  91 226 688 471 792 406 311 955 714 256 617 375\n",
      " 926 752 719 813 523  47 929 708 188 868 641 362 751 342 774 505 133 307\n",
      " 446 105 840 477 806 235 463 316  40 349 366 744 144 914 252 595 768 734\n",
      " 637  27 784 167  57 604 846 742 429 298 622 866 460 157 618 535 635 423\n",
      " 697 912 917 119 565 417 970 921 123 654 862 938 988 147 720 544 283 189\n",
      " 676 943 693 705 269  97 462 746 374 882 206 939 659 903 795 323 892 479\n",
      " 586 865 246 661 183 700 341 314  82 303  38 682 880 187 788 232 416 848\n",
      " 957 367 587 716 249 219 358 899 713 322 984 496 530   1 662 839 336 702\n",
      " 490 454 100 352  30]\n"
     ]
    }
   ],
   "source": [
    "#randomize data\n",
    "order_vec = np.array(range(len(images)))\n",
    "#print(order_vec)\n",
    "np.random.shuffle(order_vec)\n",
    "#print(order_vec)\n",
    "\n",
    "images = images[order_vec]\n",
    "labels = labels[order_vec]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 416,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "884\n",
      "111\n"
     ]
    }
   ],
   "source": [
    "#wähle Daten aus um nicht zu viele bilder mit model zu haben\n",
    "count = 0\n",
    "\n",
    "to_check = labels\n",
    "for i in range(len(to_check)):\n",
    "    if(to_check[i,0]==1):\n",
    "        count +=1\n",
    "print(count)\n",
    "count_0=0\n",
    "for i in range(len(to_check)):\n",
    "    if(to_check[i,0]==0):\n",
    "        count_0 +=1\n",
    "print(count_0)\n",
    "\n",
    "labels_1_index = labels[0:,0]==1\n",
    "labels_0_index = labels[0:,0]==0\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "fifty_fifty_bool = True\n",
    "if fifty_fifty_bool == True:\n",
    "    test_images_0 = images[labels_0_index]\n",
    "    test_images_1 = images[labels_1_index][range(count_0)]\n",
    "    images = np.concatenate([test_images_0, test_images_1], axis=0)\n",
    "    test_labels_0 = labels[labels_0_index]\n",
    "    test_labels_1 = labels[labels_1_index][range(count_0)]\n",
    "    labels = np.concatenate([test_labels_0, test_labels_1], axis=0)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 417,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[  0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17\n",
      "  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35\n",
      "  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53\n",
      "  54  55  56  57  58  59  60  61  62  63  64  65  66  67  68  69  70  71\n",
      "  72  73  74  75  76  77  78  79  80  81  82  83  84  85  86  87  88  89\n",
      "  90  91  92  93  94  95  96  97  98  99 100 101 102 103 104 105 106 107\n",
      " 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125\n",
      " 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143\n",
      " 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161\n",
      " 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179\n",
      " 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197\n",
      " 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215\n",
      " 216 217 218 219 220 221]\n",
      "[220 100  88  26 198  45 106  67 113  58 135 148 173 202  21 133   5 170\n",
      " 116 121 186 207  17  80 213  19  77  83  66  33 187 216 204  53   0 105\n",
      "  81  79  82   2  36  78 152 179 192  11 188  87  49  16 172 112  74 118\n",
      " 158 126  54 156 214 174 142  48 144 176  43  51  10 132  52  60  94 134\n",
      " 129   6 177  63 199  15  57 205 123 221  55   1 193  23  38  35 122 160\n",
      "  20 130  41  99  71 209  47  70 102 131 168 154 103 201  13 166 169 114\n",
      " 219  27 196  50 111 191  25 171 217  61  85 109  95  76  86  92 215 101\n",
      " 110 147 181  24 140 117 137 125 194 212 143 153 164 149   9  93  31  96\n",
      "  75  56 108  44  39  12 208 124 138 167 161  84 175 127  42  65 189 107\n",
      " 162 200  73   8  69  32 136  46 190  62 115 211 203  40 141  72  29 139\n",
      "  34   4  64 163  30  22  89 150 210  18 184 157 195 218   3 178 145 128\n",
      " 155  97 151  59  14 119 180 104 146 182 197 120 159 165 206   7  98 185\n",
      "  90  37  28  91  68 183]\n"
     ]
    }
   ],
   "source": [
    "#randomize data\n",
    "order_vec = np.array(range(len(images)))\n",
    "#print(order_vec)\n",
    "np.random.shuffle(order_vec)\n",
    "#print(order_vec)\n",
    "\n",
    "images = images[order_vec]\n",
    "labels = labels[order_vec]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 418,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#split data in validation und train set\n",
    "train_proportion = 0.8\n",
    "train_indexes = range(int(train_proportion*len(images)))\n",
    "val_indexes = range(int(train_proportion*len(images))+1, len(images))\n",
    "train_images = images[train_indexes]\n",
    "train_labels = labels[train_indexes]\n",
    "valid_features = images[val_indexes]\n",
    "valid_labels = labels[val_indexes]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 419,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#baue batches\n",
    "def get_batches(batch_size):\n",
    "    batches_images = train_images[range(batch_size)]\n",
    "    batches_labels = train_labels[range(batch_size)]\n",
    "    return batches_images, batches_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 420,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def neural_net_image_input(image_shape):\n",
    "    \"\"\"\n",
    "    Return a Tensor for a bach of image input\n",
    "    : image_shape: Shape of the images\n",
    "    : return: Tensor for image input.\n",
    "    \"\"\"\n",
    "    # TODO: Implement Function\n",
    "    input = tf.placeholder(tf.float32, shape = [None, image_shape[0],image_shape[1],image_shape[2]], name = \"x\")\n",
    "    \n",
    "    return input\n",
    "\n",
    "\n",
    "def neural_net_label_input(n_classes):\n",
    "    \"\"\"\n",
    "    Return a Tensor for a batch of label input\n",
    "    : n_classes: Number of classes\n",
    "    : return: Tensor for label input.\n",
    "    \"\"\"\n",
    "    input = tf.placeholder(tf.float32, shape = [None, n_classes], name = \"y\")\n",
    "    #(None, 32, 32, 3))\n",
    "    return input\n",
    "\n",
    "def neural_net_keep_prob_input():\n",
    "    \"\"\"\n",
    "    Return a Tensor for keep probability\n",
    "    : return: Tensor for keep probability.\n",
    "    \"\"\"\n",
    "    input = tf.placeholder(tf.float32,  name = \"keep_prob\")\n",
    "    #(None, 32, 32, 3))\n",
    "    return input\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 421,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def conv2d_maxpool(x_tensor, conv_num_outputs, conv_ksize, conv_strides, pool_ksize, pool_strides):\n",
    "    \"\"\"\n",
    "    Apply convolution then max pooling to x_tensor\n",
    "    :param x_tensor: TensorFlow Tensor\n",
    "    :param conv_num_outputs: Number of outputs for the convolutional layer\n",
    "    :param conv_ksize: kernal size 2-D Tuple for the convolutional layer\n",
    "    :param conv_strides: Stride 2-D Tuple for convolution\n",
    "    :param pool_ksize: kernal size 2-D Tuple for pool\n",
    "    :param pool_strides: Stride 2-D Tuple for pool\n",
    "    : return: A tensor that represents convolution and max pooling of x_tensor\n",
    "    \"\"\"\n",
    "    weight = tf.Variable(tf.truncated_normal([conv_ksize[0], conv_ksize[1],x_tensor.get_shape().as_list()[3], conv_num_outputs], \n",
    "                                             mean =0.0, stddev=0.1))\n",
    "    #out_height = int(np.ceil(float(x_tensor.get_shape().as_list()[1] - conv_ksize[0] + 1) / float(conv_strides[0]))) \n",
    "    #out_width  = int(np.ceil(float(x_tensor.get_shape().as_list()[2] - conv_ksize[1] + 1) / float(conv_strides[1])))\n",
    "    #print(out_width)\n",
    "    # weight = tf.Variable(tf.truncated_normal([out_height, out_width,x_tensor.get_shape().as_list()[3], conv_num_outputs], \n",
    "    #                                          mean =0.0, stddev=0.1))\n",
    "    #bias = tf.Variable(tf.random_normal([conv_num_outputs]))\n",
    "    bias = tf.Variable(tf.zeros([conv_num_outputs]))\n",
    "    conv_layer = tf.nn.conv2d(x_tensor, weight, strides=[1, conv_strides[0], conv_strides[1], 1], padding='SAME')\n",
    "    # Add bias\n",
    "    conv_layer = tf.nn.bias_add(conv_layer, bias)\n",
    "    # Apply activation function\n",
    "    conv_layer = tf.nn.relu(conv_layer)\n",
    "    conv_layer = tf.nn.max_pool(conv_layer, ksize=[1, pool_ksize[0], pool_ksize[1], 1],\n",
    "                                strides=[1, pool_strides[0], pool_strides[1], 1], padding='SAME')\n",
    "    \n",
    "    return conv_layer\n",
    "\n",
    "def flatten(x_tensor):\n",
    "    \"\"\"\n",
    "    Flatten x_tensor to (Batch Size, Flattened Image Size)\n",
    "    : x_tensor: A tensor of size (Batch Size, ...), where ... are the image dimensions.\n",
    "    : return: A tensor of size (Batch Size, Flattened Image Size).\n",
    "    \"\"\"\n",
    "    new_dimension = x_tensor.get_shape().as_list()[1]*x_tensor.get_shape().as_list()[2]*x_tensor.get_shape().as_list()[3]\n",
    "    flattened_tensor = tf.reshape(x_tensor,[-1, new_dimension ])\n",
    "    #print(flattened_tensor.shape)\n",
    "    return flattened_tensor\n",
    "\n",
    "def fully_conn(x_tensor, num_outputs):\n",
    "    \"\"\"\n",
    "    Apply a fully connected layer to x_tensor using weight and bias\n",
    "    : x_tensor: A 2-D tensor where the first dimension is batch size.\n",
    "    : num_outputs: The number of output that the new tensor should be.\n",
    "    : return: A 2-D tensor where the second dimension is num_outputs.\n",
    "    \"\"\"\n",
    "    weight = tf.Variable(tf.truncated_normal([x_tensor.get_shape().as_list()[1], num_outputs], mean =0.0,stddev=0.1))\n",
    "    \n",
    "    #bias = tf.Variable(tf.random_normal([num_outputs]))\n",
    "    bias = tf.Variable(tf.zeros([num_outputs]))\n",
    "    fully_layer = tf.add(tf.matmul(x_tensor, weight), bias)\n",
    "    fully_layer = tf.nn.relu(fully_layer)\n",
    "    \n",
    "    #return tf.contrib.layers.fully_connected(x_tensor, num_outputs, activation_fn=tf.nn.relu)\n",
    "    return fully_layer\n",
    "\n",
    "def output(x_tensor, num_outputs):\n",
    "    \"\"\"\n",
    "    Apply a output layer to x_tensor using weight and bias\n",
    "    : x_tensor: A 2-D tensor where the first dimension is batch size.\n",
    "    : num_outputs: The number of output that the new tensor should be.\n",
    "    : return: A 2-D tensor where the second dimension is num_outputs.\n",
    "    \"\"\"\n",
    "    weight = tf.Variable(tf.truncated_normal([x_tensor.get_shape().as_list()[1], num_outputs], mean =0.0,stddev=0.1))\n",
    "    \n",
    "    #bias = tf.Variable(tf.random_normal([num_outputs]))\n",
    "    bias = tf.Variable(tf.zeros([num_outputs]))\n",
    "    out_layer = tf.add(tf.matmul(x_tensor, weight), bias)\n",
    "    #out_layer = tf.nn.relu(out_layer)\n",
    "    return out_layer\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 422,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def conv_net(x, keep_prob):\n",
    "    \"\"\"\n",
    "    Create a convolutional neural network model\n",
    "    : x: Placeholder tensor that holds image data.\n",
    "    : keep_prob: Placeholder tensor that hold dropout keep probability.\n",
    "    : return: Tensor that represents logits\n",
    "    \"\"\"\n",
    "    # TODO: Apply 1, 2, or 3 Convolution and Max Pool layers\n",
    "    #    Play around with different number of outputs, kernel size and stride\n",
    "    # Function Definition from Above:\n",
    "    #    conv2d_maxpool(x_tensor, conv_num_outputs, conv_ksize, conv_strides, pool_ksize, pool_strides)\n",
    "    conv_layer_1 = conv2d_maxpool(x, 64, [4,4], [2,2], [2,2], [2,2])\n",
    "    #shape 16,16,36\n",
    "    conv_layer_2 = conv2d_maxpool(conv_layer_1, 128, [4,4], [1,1], [2,2], [2,2])\n",
    "    #shape  8,8,128\n",
    "    #conv_layer_3 = conv2d_maxpool(conv_layer_2, 256, [3,3], [1,1], [1,1], [1,1])\n",
    "    # TODO: Apply a Flatten Layer\n",
    "    # Function Definition from Above:\n",
    "    #   flatten(x_tensor)\n",
    "    flatten_layer = flatten(conv_layer_2)\n",
    "    #flatten_layer = tf.nn.dropout(flatten_layer, keep_prob)\n",
    "    # TODO: Apply 1, 2, or 3 Fully Connected Layers\n",
    "    #    Play around with different number of outputs\n",
    "    # Function Definition from Above:\n",
    "    #   fully_conn(x_tensor, num_outputs)\n",
    "    fully_con_layer_1 = fully_conn(flatten_layer, 100)\n",
    "    fully_con_layer_1 = tf.nn.dropout(fully_con_layer_1, keep_prob)\n",
    "    fully_con_layer_2 = fully_conn(fully_con_layer_1, 50)\n",
    "    fully_con_layer_2 = tf.nn.dropout(fully_con_layer_2, keep_prob)\n",
    "    #fully_con_layer_3 = fully_conn(fully_con_layer_2, 50)\n",
    "    # TODO: Apply an Output Layer\n",
    "    #    Set this to the number of classes\n",
    "    # Function Definition from Above:\n",
    "    #   output(x_tensor, num_outputs)\n",
    "    output_layer = output(fully_con_layer_2, nr_of_labels)\n",
    "    \n",
    "    # TODO: return output\n",
    "    return output_layer\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 423,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "nr_of_labels =2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 424,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "tf.reset_default_graph()\n",
    "\n",
    "# Inputs\n",
    "x = neural_net_image_input((32, 32, 3))\n",
    "y = neural_net_label_input(2)\n",
    "keep_prob = neural_net_keep_prob_input()\n",
    "\n",
    "# Model\n",
    "logits = conv_net(x, keep_prob)\n",
    "\n",
    "# Name logits Tensor, so that is can be loaded from disk after training\n",
    "logits = tf.identity(logits, name='logits')\n",
    "\n",
    "# Loss and Optimizer\n",
    "cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits, labels=y))\n",
    "optimizer = tf.train.AdamOptimizer().minimize(cost)\n",
    "\n",
    "# Accuracy\n",
    "correct_pred = tf.equal(tf.argmax(logits, 1), tf.argmax(y, 1))\n",
    "accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32), name='accuracy')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 425,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def train_neural_network(session, optimizer, keep_probability, feature_batch, label_batch):\n",
    "    \"\"\"\n",
    "    Optimize the session on a batch of images and labels\n",
    "    : session: Current TensorFlow session\n",
    "    : optimizer: TensorFlow optimizer function\n",
    "    : keep_probability: keep probability\n",
    "    : feature_batch: Batch of Numpy image data\n",
    "    : label_batch: Batch of Numpy label data\n",
    "    \"\"\"\n",
    "    session.run(optimizer, feed_dict={x: feature_batch, y: label_batch, keep_prob: keep_probability})\n",
    "   \n",
    "    pass\n",
    "def print_stats(session, feature_batch, label_batch, cost, accuracy):\n",
    "    \"\"\"\n",
    "    Print information about loss and validation accuracy\n",
    "    : session: Current TensorFlow session\n",
    "    : feature_batch: Batch of Numpy image data\n",
    "    : label_batch: Batch of Numpy label data\n",
    "    : cost: TensorFlow cost function\n",
    "    : accuracy: TensorFlow accuracy function\n",
    "    \"\"\"\n",
    "    loss = session.run(cost, feed_dict={x: feature_batch, y: label_batch, keep_prob: 1.0})\n",
    "    valid_acc = session.run(accuracy, feed_dict={x: valid_features, y: valid_labels, keep_prob: 1.0})\n",
    "    print('Loss at {}'.format(loss), 'Validation Accuracy at {}'.format(valid_acc))\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 426,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# TODO: Tune Parameters\n",
    "epochs = 50\n",
    "batch_size = 128\n",
    "keep_probability = 0.6"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 427,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5454545454545454\n"
     ]
    }
   ],
   "source": [
    "count = 0\n",
    "\n",
    "to_check = valid_labels\n",
    "for i in range(len(to_check)):\n",
    "    if(to_check[i,0]==1):\n",
    "        count +=1\n",
    "        \n",
    "print(count/len(to_check))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 428,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checking the Training on a Single Batch...\n",
      "Epoch  1, CIFAR-10 Batch 1:  Loss at 1.078268051147461 Validation Accuracy at 0.4545454680919647\n",
      "Epoch  2, CIFAR-10 Batch 1:  Loss at 0.6226668357849121 Validation Accuracy at 0.47727271914482117\n",
      "Epoch  3, CIFAR-10 Batch 1:  Loss at 0.6590913534164429 Validation Accuracy at 0.5909090638160706\n",
      "Epoch  4, CIFAR-10 Batch 1:  Loss at 0.6290079355239868 Validation Accuracy at 0.6136363744735718\n",
      "Epoch  5, CIFAR-10 Batch 1:  Loss at 0.612542986869812 Validation Accuracy at 0.9545454382896423\n",
      "Epoch  6, CIFAR-10 Batch 1:  Loss at 0.6091745495796204 Validation Accuracy at 0.6363636255264282\n",
      "Epoch  7, CIFAR-10 Batch 1:  Loss at 0.596767008304596 Validation Accuracy at 0.8863636255264282\n",
      "Epoch  8, CIFAR-10 Batch 1:  Loss at 0.5900323987007141 Validation Accuracy at 0.9318181872367859\n",
      "Epoch  9, CIFAR-10 Batch 1:  Loss at 0.5845392942428589 Validation Accuracy at 0.9090909361839294\n",
      "Epoch 10, CIFAR-10 Batch 1:  Loss at 0.5782239437103271 Validation Accuracy at 0.9090909361839294\n",
      "Epoch 11, CIFAR-10 Batch 1:  Loss at 0.5608611702919006 Validation Accuracy at 0.9318181872367859\n",
      "Epoch 12, CIFAR-10 Batch 1:  Loss at 0.5460914373397827 Validation Accuracy at 0.8863636255264282\n",
      "Epoch 13, CIFAR-10 Batch 1:  Loss at 0.5312604308128357 Validation Accuracy at 0.8409090638160706\n",
      "Epoch 14, CIFAR-10 Batch 1:  Loss at 0.5102517604827881 Validation Accuracy at 0.8863636255264282\n",
      "Epoch 15, CIFAR-10 Batch 1:  Loss at 0.4836655855178833 Validation Accuracy at 0.9318181872367859\n",
      "Epoch 16, CIFAR-10 Batch 1:  Loss at 0.4556095004081726 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 17, CIFAR-10 Batch 1:  Loss at 0.43908289074897766 Validation Accuracy at 0.9090909361839294\n",
      "Epoch 18, CIFAR-10 Batch 1:  Loss at 0.4127803444862366 Validation Accuracy at 0.9090909361839294\n",
      "Epoch 19, CIFAR-10 Batch 1:  Loss at 0.37220126390457153 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 20, CIFAR-10 Batch 1:  Loss at 0.34596914052963257 Validation Accuracy at 0.9318181872367859\n",
      "Epoch 21, CIFAR-10 Batch 1:  Loss at 0.3183683753013611 Validation Accuracy at 0.9318181872367859\n",
      "Epoch 22, CIFAR-10 Batch 1:  Loss at 0.28366684913635254 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 23, CIFAR-10 Batch 1:  Loss at 0.26824283599853516 Validation Accuracy at 0.9318181872367859\n",
      "Epoch 24, CIFAR-10 Batch 1:  Loss at 0.24678927659988403 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 25, CIFAR-10 Batch 1:  Loss at 0.2257637083530426 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 26, CIFAR-10 Batch 1:  Loss at 0.20741595327854156 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 27, CIFAR-10 Batch 1:  Loss at 0.17429374158382416 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 28, CIFAR-10 Batch 1:  Loss at 0.15256060659885406 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 29, CIFAR-10 Batch 1:  Loss at 0.13347676396369934 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 30, CIFAR-10 Batch 1:  Loss at 0.11784488707780838 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 31, CIFAR-10 Batch 1:  Loss at 0.10198447108268738 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 32, CIFAR-10 Batch 1:  Loss at 0.0861387550830841 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 33, CIFAR-10 Batch 1:  Loss at 0.07337772846221924 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 34, CIFAR-10 Batch 1:  Loss at 0.062178369611501694 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 35, CIFAR-10 Batch 1:  Loss at 0.05467458441853523 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 36, CIFAR-10 Batch 1:  Loss at 0.04914160445332527 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 37, CIFAR-10 Batch 1:  Loss at 0.03992551565170288 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 38, CIFAR-10 Batch 1:  Loss at 0.035072412341833115 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 39, CIFAR-10 Batch 1:  Loss at 0.028078429400920868 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 40, CIFAR-10 Batch 1:  Loss at 0.025075364857912064 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 41, CIFAR-10 Batch 1:  Loss at 0.021370606496930122 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 42, CIFAR-10 Batch 1:  Loss at 0.017660144716501236 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 43, CIFAR-10 Batch 1:  Loss at 0.018171902745962143 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 44, CIFAR-10 Batch 1:  Loss at 0.015724724158644676 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 45, CIFAR-10 Batch 1:  Loss at 0.013995343819260597 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 46, CIFAR-10 Batch 1:  Loss at 0.009966823272407055 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 47, CIFAR-10 Batch 1:  Loss at 0.00815819762647152 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 48, CIFAR-10 Batch 1:  Loss at 0.00822435598820448 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 49, CIFAR-10 Batch 1:  Loss at 0.006202083081007004 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 50, CIFAR-10 Batch 1:  Loss at 0.005292184185236692 Validation Accuracy at 0.9545454382896423\n"
     ]
    }
   ],
   "source": [
    "print('Checking the Training on a Single Batch...')\n",
    "with tf.Session() as sess:\n",
    "    # Initializing the variables\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    \n",
    "    # Training cycle\n",
    "    for epoch in range(epochs):\n",
    "        batch_i = 1\n",
    "        batch_features, batch_labels = get_batches(batch_size)\n",
    "        train_neural_network(sess, optimizer, keep_probability, batch_features, batch_labels)\n",
    "        print('Epoch {:>2}, Witt-Model-classifier Batch {}:  '.format(epoch + 1, batch_i), end='')\n",
    "        print_stats(sess, batch_features, batch_labels, cost, accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 429,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training...\n",
      "Epoch  1, CIFAR-10 Batch 1:  Loss at 0.6586146354675293 Validation Accuracy at 0.5454545617103577\n",
      "Epoch  2, CIFAR-10 Batch 1:  Loss at 0.6274292469024658 Validation Accuracy at 0.7045454382896423\n",
      "Epoch  3, CIFAR-10 Batch 1:  Loss at 0.6202279329299927 Validation Accuracy at 0.8636363744735718\n",
      "Epoch  4, CIFAR-10 Batch 1:  Loss at 0.6039412021636963 Validation Accuracy at 0.7272727489471436\n",
      "Epoch  5, CIFAR-10 Batch 1:  Loss at 0.5702357292175293 Validation Accuracy at 0.9090909361839294\n",
      "Epoch  6, CIFAR-10 Batch 1:  Loss at 0.5655685663223267 Validation Accuracy at 0.6818181872367859\n",
      "Epoch  7, CIFAR-10 Batch 1:  Loss at 0.526997447013855 Validation Accuracy at 0.75\n",
      "Epoch  8, CIFAR-10 Batch 1:  Loss at 0.4871102571487427 Validation Accuracy at 0.9545454382896423\n",
      "Epoch  9, CIFAR-10 Batch 1:  Loss at 0.4516737461090088 Validation Accuracy at 0.9090909361839294\n",
      "Epoch 10, CIFAR-10 Batch 1:  Loss at 0.43324145674705505 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 11, CIFAR-10 Batch 1:  Loss at 0.4232521057128906 Validation Accuracy at 0.9090909361839294\n",
      "Epoch 12, CIFAR-10 Batch 1:  Loss at 0.3970862925052643 Validation Accuracy at 0.8863636255264282\n",
      "Epoch 13, CIFAR-10 Batch 1:  Loss at 0.3547390401363373 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 14, CIFAR-10 Batch 1:  Loss at 0.3241971433162689 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 15, CIFAR-10 Batch 1:  Loss at 0.2933815121650696 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 16, CIFAR-10 Batch 1:  Loss at 0.2755570411682129 Validation Accuracy at 0.9318181872367859\n",
      "Epoch 17, CIFAR-10 Batch 1:  Loss at 0.2525132894515991 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 18, CIFAR-10 Batch 1:  Loss at 0.2248390018939972 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 19, CIFAR-10 Batch 1:  Loss at 0.21513935923576355 Validation Accuracy at 0.9318181872367859\n",
      "Epoch 20, CIFAR-10 Batch 1:  Loss at 0.18299388885498047 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 21, CIFAR-10 Batch 1:  Loss at 0.17003965377807617 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 22, CIFAR-10 Batch 1:  Loss at 0.16697487235069275 Validation Accuracy at 0.9318181872367859\n",
      "Epoch 23, CIFAR-10 Batch 1:  Loss at 0.12026223540306091 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 24, CIFAR-10 Batch 1:  Loss at 0.09785829484462738 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 25, CIFAR-10 Batch 1:  Loss at 0.08315597474575043 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 26, CIFAR-10 Batch 1:  Loss at 0.07099586725234985 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 27, CIFAR-10 Batch 1:  Loss at 0.06043460592627525 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 28, CIFAR-10 Batch 1:  Loss at 0.050678402185440063 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 29, CIFAR-10 Batch 1:  Loss at 0.04406125470995903 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 30, CIFAR-10 Batch 1:  Loss at 0.03886505588889122 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 31, CIFAR-10 Batch 1:  Loss at 0.036348797380924225 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 32, CIFAR-10 Batch 1:  Loss at 0.03516396880149841 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 33, CIFAR-10 Batch 1:  Loss at 0.034291598945856094 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 34, CIFAR-10 Batch 1:  Loss at 0.032471589744091034 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 35, CIFAR-10 Batch 1:  Loss at 0.025267725810408592 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 36, CIFAR-10 Batch 1:  Loss at 0.01834295131266117 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 37, CIFAR-10 Batch 1:  Loss at 0.014303067699074745 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 38, CIFAR-10 Batch 1:  Loss at 0.01123939361423254 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 39, CIFAR-10 Batch 1:  Loss at 0.010605301707983017 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 40, CIFAR-10 Batch 1:  Loss at 0.007478250190615654 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 41, CIFAR-10 Batch 1:  Loss at 0.00559915229678154 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 42, CIFAR-10 Batch 1:  Loss at 0.005956882610917091 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 43, CIFAR-10 Batch 1:  Loss at 0.0045290738344192505 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 44, CIFAR-10 Batch 1:  Loss at 0.0031299814581871033 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 45, CIFAR-10 Batch 1:  Loss at 0.003926997538655996 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 46, CIFAR-10 Batch 1:  Loss at 0.004085836932063103 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 47, CIFAR-10 Batch 1:  Loss at 0.002360214479267597 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 48, CIFAR-10 Batch 1:  Loss at 0.002259391825646162 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 49, CIFAR-10 Batch 1:  Loss at 0.002944720908999443 Validation Accuracy at 0.9545454382896423\n",
      "Epoch 50, CIFAR-10 Batch 1:  Loss at 0.0025436747819185257 Validation Accuracy at 0.9545454382896423\n"
     ]
    }
   ],
   "source": [
    "save_model_path = './image_classification'\n",
    "\n",
    "print('Training...')\n",
    "with tf.Session() as sess:\n",
    "    # Initializing the variables\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    \n",
    "    # Training cycle\n",
    "    for epoch in range(epochs):\n",
    "        # Loop over all batches\n",
    "        n_batches = 5\n",
    "        batch_i = 1\n",
    "        batch_features, batch_labels = get_batches(batch_size)\n",
    "        train_neural_network(sess, optimizer, keep_probability, batch_features, batch_labels)\n",
    "        print('Epoch {:>2}, witt-model-classifier Batch {}:  '.format(epoch + 1, batch_i), end='')\n",
    "        print_stats(sess, batch_features, batch_labels, cost, accuracy)\n",
    "            \n",
    "    # Save Model\n",
    "    saver = tf.train.Saver()\n",
    "    save_path = saver.save(sess, save_model_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 463,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def display_image_predictions(features, labels, predictions):\n",
    "    n_classes = nr_of_labels\n",
    "    label_names = [0,1]\n",
    "    label_binarizer = LabelBinarizer()\n",
    "    label_binarizer.fit(range(n_classes))\n",
    "    label_ids = label_binarizer.inverse_transform(np.array(labels))\n",
    "\n",
    "    nr_of_test_images = len(labels)\n",
    "    fig, axies = plt.subplots(nrows=nr_of_test_images, ncols=2)\n",
    "    fig.tight_layout()\n",
    "    fig.suptitle('Softmax Predictions', fontsize=20, y=1.1)\n",
    "\n",
    "    n_predictions = 2\n",
    "    margin = 0.05\n",
    "    ind = np.arange(n_predictions)\n",
    "    width = (1. - 2. * margin) / n_predictions\n",
    "\n",
    "    for image_i, (feature, label_id, pred_indicies, pred_values) in enumerate(zip(features, label_ids, predictions.indices, predictions.values)):\n",
    "        pred_names = [label_names[pred_i] for pred_i in pred_indicies]\n",
    "        correct_name = label_names[label_id]\n",
    "\n",
    "        axies[image_i][0].imshow(feature)\n",
    "        axies[image_i][0].set_title(correct_name)\n",
    "        axies[image_i][0].set_axis_off()\n",
    "\n",
    "        axies[image_i][1].barh(ind + margin, pred_values[::-1], width)\n",
    "        axies[image_i][1].set_yticks(ind + margin)\n",
    "        axies[image_i][1].set_yticklabels(pred_names[::-1])\n",
    "        axies[image_i][1].set_xticks([0, 0.5, 1.0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 476,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./image_classification\n",
      "Testing Accuracy: 0.9545454382896423\n",
      "\n",
      "44\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAArsAAAJ/CAYAAAB4GhsgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1f3288cea20>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 319,
       "width": 349
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format = 'retina'\n",
    "\n",
    "import tensorflow as tf\n",
    "import pickle\n",
    "#import helper\n",
    "import random\n",
    "\n",
    "# Set batch size if not already set\n",
    "try:\n",
    "    if batch_size:\n",
    "        pass\n",
    "except NameError:\n",
    "    batch_size = 64\n",
    "\n",
    "save_model_path = './image_classification'\n",
    "n_samples = 4\n",
    "top_n_predictions = 2\n",
    "\n",
    "def test_model():\n",
    "    \"\"\"\n",
    "    Test the saved model against the test dataset\n",
    "    \"\"\"\n",
    "\n",
    "    test_features, test_labels = valid_features, valid_labels\n",
    "    loaded_graph = tf.Graph()\n",
    "\n",
    "    with tf.Session(graph=loaded_graph) as sess:\n",
    "        # Load model\n",
    "        loader = tf.train.import_meta_graph(save_model_path + '.meta')\n",
    "        loader.restore(sess, save_model_path)\n",
    "\n",
    "        # Get Tensors from loaded model\n",
    "        loaded_x = loaded_graph.get_tensor_by_name('x:0')\n",
    "        loaded_y = loaded_graph.get_tensor_by_name('y:0')\n",
    "        loaded_keep_prob = loaded_graph.get_tensor_by_name('keep_prob:0')\n",
    "        loaded_logits = loaded_graph.get_tensor_by_name('logits:0')\n",
    "        loaded_acc = loaded_graph.get_tensor_by_name('accuracy:0')\n",
    "        \n",
    "        # Get accuracy in batches for memory limitations\n",
    "        test_batch_acc_total = 0\n",
    "        test_batch_count = 0\n",
    "        \n",
    "        train_feature_batch, train_label_batch = test_features, test_labels\n",
    "        test_batch_acc_total += sess.run(\n",
    "            loaded_acc,\n",
    "            feed_dict={loaded_x: train_feature_batch, loaded_y: train_label_batch, loaded_keep_prob: 1.0})\n",
    "        test_batch_count += 1\n",
    "\n",
    "        print('Testing Accuracy: {}\\n'.format(test_batch_acc_total/test_batch_count))\n",
    "        print(len(test_features))\n",
    "        # Print Random Samples\n",
    "        random_test_features, random_test_labels = tuple(zip(*random.sample(list(zip(test_features, test_labels)), n_samples)))\n",
    "        random_test_predictions = sess.run(\n",
    "            tf.nn.top_k(tf.nn.softmax(loaded_logits), top_n_predictions),\n",
    "            feed_dict={loaded_x: random_test_features, loaded_y: random_test_labels, loaded_keep_prob: 1.0})\n",
    "        display_image_predictions(random_test_features, random_test_labels, random_test_predictions)\n",
    "\n",
    "\n",
    "test_model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 403,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x1f331549f28>"
      ]
     },
     "execution_count": 403,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfoAAAH0CAYAAADVH+85AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1f331075438>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 250,
       "width": 253
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format = 'retina'\n",
    "\n",
    "plt.imshow(batch_features[3,]) \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1007, 1)"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.shape(labels)"
   ]
  }
 ],
 "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": "3.5.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}