Relaynet_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import keras\n",
    "from keras.layers import Activation\n",
    "from keras.layers import Conv2D, MaxPooling2D\n",
    "from keras.models import Model\n",
    "from keras.layers import Input\n",
    "from keras.layers import BatchNormalization\n",
    "from keras.layers import UpSampling2D\n",
    "from keras.layers import Concatenate\n",
    "from keras.layers import Lambda \n",
    "from keras.utils import to_categorical\n",
    "import tensorflow as tf\n",
    "\n",
    "from keras.layers import Reshape\n",
    "\n",
    "from keras import backend as K\n",
    "from keras import regularizers, optimizers\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.callbacks import ReduceLROnPlateau, CSVLogger,EarlyStopping,ModelCheckpoint\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import scipy.io as scio\n",
    "import numpy as np    \n",
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "import math\n",
    "import re\n",
    "from scipy.misc import imsave\n",
    "from scipy import ndimage, misc\n",
    "from numpy import unravel_index\n",
    "from operator import sub"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def atoi(text) : \n",
    "    return int(text) if text.isdigit() else text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def natural_keys(text) :\n",
    "    return [atoi(c) for c in re.split('(\\d+)', text)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "root_path = \"\"\n",
    "filenames = []\n",
    "for root, dirnames, filenames in os.walk(\"Dataset/DenoisedTrain/\"):\n",
    "    filenames.sort(key = natural_keys)\n",
    "    rootpath = root"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['denoised_1.png', 'denoised_2.png', 'denoised_3.png', 'denoised_4.png', 'denoised_5.png', 'denoised_6.png', 'denoised_7.png', 'denoised_8.png', 'denoised_9.png', 'denoised_10.png', 'denoised_11.png', 'denoised_12.png', 'denoised_13.png', 'denoised_14.png', 'denoised_15.png', 'denoised_16.png', 'denoised_17.png', 'denoised_18.png', 'denoised_19.png', 'denoised_20.png', 'denoised_21.png', 'denoised_22.png', 'denoised_23.png', 'denoised_24.png', 'denoised_25.png', 'denoised_26.png', 'denoised_27.png', 'denoised_28.png', 'denoised_29.png', 'denoised_30.png', 'denoised_31.png', 'denoised_32.png', 'denoised_33.png', 'denoised_34.png', 'denoised_35.png', 'denoised_36.png', 'denoised_37.png', 'denoised_38.png', 'denoised_39.png', 'denoised_40.png', 'denoised_41.png', 'denoised_42.png', 'denoised_43.png', 'denoised_44.png', 'denoised_45.png', 'denoised_46.png', 'denoised_47.png', 'denoised_48.png', 'denoised_49.png', 'denoised_50.png', 'denoised_51.png', 'denoised_52.png', 'denoised_53.png', 'denoised_54.png', 'denoised_55.png', 'denoised_56.png', 'denoised_57.png', 'denoised_58.png', 'denoised_59.png', 'denoised_60.png', 'denoised_61.png', 'denoised_62.png', 'denoised_63.png', 'denoised_64.png', 'denoised_65.png', 'denoised_66.png', 'denoised_67.png', 'denoised_68.png', 'denoised_69.png', 'denoised_70.png', 'denoised_71.png', 'denoised_72.png', 'denoised_73.png', 'denoised_74.png', 'denoised_75.png', 'denoised_76.png', 'denoised_77.png', 'denoised_78.png', 'denoised_79.png', 'denoised_80.png', 'denoised_81.png', 'denoised_82.png', 'denoised_83.png', 'denoised_84.png', 'denoised_85.png', 'denoised_86.png', 'denoised_87.png', 'denoised_88.png', 'denoised_89.png', 'denoised_90.png', 'denoised_91.png', 'denoised_92.png', 'denoised_93.png', 'denoised_94.png', 'denoised_95.png', 'denoised_96.png', 'denoised_97.png', 'denoised_98.png', 'denoised_99.png', 'denoised_100.png', 'denoised_101.png', 'denoised_102.png', 'denoised_103.png', 'denoised_104.png', 'denoised_105.png', 'denoised_106.png', 'denoised_107.png', 'denoised_108.png', 'denoised_109.png', 'denoised_110.png', 'denoised_111.png', 'denoised_112.png', 'denoised_113.png', 'denoised_114.png', 'denoised_115.png', 'denoised_116.png', 'denoised_117.png', 'denoised_118.png', 'denoised_119.png', 'denoised_120.png', 'denoised_121.png', 'denoised_122.png', 'denoised_123.png', 'denoised_124.png', 'denoised_125.png', 'denoised_126.png', 'denoised_127.png', 'denoised_128.png', 'denoised_129.png', 'denoised_130.png', 'denoised_131.png', 'denoised_132.png', 'denoised_133.png', 'denoised_134.png', 'denoised_135.png', 'denoised_136.png', 'denoised_137.png', 'denoised_138.png', 'denoised_139.png', 'denoised_140.png', 'denoised_141.png', 'denoised_142.png', 'denoised_143.png', 'denoised_144.png', 'denoised_145.png', 'denoised_146.png', 'denoised_147.png', 'denoised_148.png', 'denoised_149.png', 'denoised_150.png', 'denoised_151.png', 'denoised_152.png', 'denoised_153.png', 'denoised_154.png', 'denoised_155.png', 'denoised_156.png', 'denoised_157.png', 'denoised_158.png', 'denoised_159.png', 'denoised_160.png', 'denoised_161.png', 'denoised_162.png', 'denoised_163.png', 'denoised_164.png', 'denoised_165.png', 'denoised_166.png', 'denoised_167.png', 'denoised_168.png', 'denoised_169.png', 'denoised_170.png', 'denoised_171.png', 'denoised_172.png', 'denoised_173.png', 'denoised_174.png', 'denoised_175.png', 'denoised_176.png', 'denoised_177.png', 'denoised_178.png', 'denoised_179.png', 'denoised_180.png', 'denoised_181.png', 'denoised_182.png', 'denoised_183.png', 'denoised_184.png', 'denoised_185.png', 'denoised_186.png', 'denoised_187.png', 'denoised_188.png', 'denoised_189.png', 'denoised_190.png', 'denoised_191.png', 'denoised_192.png', 'denoised_193.png', 'denoised_194.png', 'denoised_195.png', 'denoised_196.png', 'denoised_197.png', 'denoised_198.png', 'denoised_199.png', 'denoised_200.png', 'denoised_201.png', 'denoised_202.png', 'denoised_203.png', 'denoised_204.png', 'denoised_205.png', 'denoised_206.png', 'denoised_207.png', 'denoised_208.png', 'denoised_209.png', 'denoised_210.png', 'denoised_211.png', 'denoised_212.png', 'denoised_213.png', 'denoised_214.png', 'denoised_215.png', 'denoised_216.png', 'denoised_217.png', 'denoised_218.png', 'denoised_219.png', 'denoised_220.png', 'denoised_221.png', 'denoised_222.png', 'denoised_223.png', 'denoised_224.png', 'denoised_225.png', 'denoised_226.png', 'denoised_227.png', 'denoised_228.png', 'denoised_229.png', 'denoised_230.png', 'denoised_231.png', 'denoised_232.png', 'denoised_233.png', 'denoised_234.png', 'denoised_235.png', 'denoised_236.png', 'denoised_237.png', 'denoised_238.png', 'denoised_239.png', 'denoised_240.png', 'denoised_241.png', 'denoised_242.png', 'denoised_243.png', 'denoised_244.png', 'denoised_245.png', 'denoised_246.png', 'denoised_247.png', 'denoised_248.png', 'denoised_249.png', 'denoised_250.png', 'denoised_251.png', 'denoised_252.png', 'denoised_253.png', 'denoised_254.png', 'denoised_255.png', 'denoised_256.png', 'denoised_257.png', 'denoised_258.png', 'denoised_259.png', 'denoised_260.png', 'denoised_261.png', 'denoised_262.png', 'denoised_263.png', 'denoised_264.png', 'denoised_265.png', 'denoised_266.png', 'denoised_267.png', 'denoised_268.png', 'denoised_269.png', 'denoised_270.png', 'denoised_271.png', 'denoised_272.png', 'denoised_273.png', 'denoised_274.png', 'denoised_275.png', 'denoised_276.png', 'denoised_277.png', 'denoised_278.png', 'denoised_279.png', 'denoised_280.png', 'denoised_281.png', 'denoised_282.png', 'denoised_283.png', 'denoised_284.png', 'denoised_285.png', 'denoised_286.png', 'denoised_287.png', 'denoised_288.png', 'denoised_289.png', 'denoised_290.png', 'denoised_291.png', 'denoised_292.png', 'denoised_293.png', 'denoised_294.png', 'denoised_295.png', 'denoised_296.png', 'denoised_297.png', 'denoised_298.png', 'denoised_299.png', 'denoised_300.png', 'denoised_301.png', 'denoised_302.png', 'denoised_303.png', 'denoised_304.png', 'denoised_305.png', 'denoised_306.png', 'denoised_307.png', 'denoised_308.png', 'denoised_309.png', 'denoised_310.png', 'denoised_311.png', 'denoised_312.png', 'denoised_313.png', 'denoised_314.png', 'denoised_315.png', 'denoised_316.png', 'denoised_317.png', 'denoised_318.png', 'denoised_319.png', 'denoised_320.png', 'denoised_321.png', 'denoised_322.png', 'denoised_323.png', 'denoised_324.png', 'denoised_325.png', 'denoised_326.png', 'denoised_327.png', 'denoised_328.png', 'denoised_329.png', 'denoised_330.png', 'denoised_331.png', 'denoised_332.png', 'denoised_333.png', 'denoised_334.png', 'denoised_335.png', 'denoised_336.png', 'denoised_337.png', 'denoised_338.png', 'denoised_339.png', 'denoised_340.png', 'denoised_341.png', 'denoised_342.png', 'denoised_343.png', 'denoised_344.png', 'denoised_345.png', 'denoised_346.png', 'denoised_347.png', 'denoised_348.png', 'denoised_349.png', 'denoised_350.png', 'denoised_351.png', 'denoised_352.png', 'denoised_353.png', 'denoised_354.png', 'denoised_355.png', 'denoised_356.png', 'denoised_357.png', 'denoised_358.png', 'denoised_359.png', 'denoised_360.png', 'denoised_361.png', 'denoised_362.png', 'denoised_363.png', 'denoised_364.png', 'denoised_365.png', 'denoised_366.png', 'denoised_367.png', 'denoised_368.png', 'denoised_369.png', 'denoised_370.png', 'denoised_371.png', 'denoised_372.png', 'denoised_373.png', 'denoised_374.png', 'denoised_375.png', 'denoised_376.png', 'denoised_377.png', 'denoised_378.png', 'denoised_379.png', 'denoised_380.png', 'denoised_381.png', 'denoised_382.png', 'denoised_383.png', 'denoised_384.png', 'denoised_385.png', 'denoised_386.png', 'denoised_387.png', 'denoised_388.png', 'denoised_389.png', 'denoised_390.png', 'denoised_391.png', 'denoised_392.png', 'denoised_393.png', 'denoised_394.png', 'denoised_395.png', 'denoised_396.png', 'denoised_397.png', 'denoised_398.png', 'denoised_399.png', 'denoised_400.png', 'denoised_401.png', 'denoised_402.png', 'denoised_403.png', 'denoised_404.png', 'denoised_405.png', 'denoised_406.png', 'denoised_407.png', 'denoised_408.png', 'denoised_409.png', 'denoised_410.png', 'denoised_411.png', 'denoised_412.png', 'denoised_413.png', 'denoised_414.png', 'denoised_415.png', 'denoised_416.png', 'denoised_417.png', 'denoised_418.png', 'denoised_419.png', 'denoised_420.png', 'denoised_421.png', 'denoised_422.png', 'denoised_423.png', 'denoised_424.png', 'denoised_425.png', 'denoised_426.png', 'denoised_427.png', 'denoised_428.png', 'denoised_429.png', 'denoised_430.png', 'denoised_431.png', 'denoised_432.png', 'denoised_433.png', 'denoised_434.png', 'denoised_435.png', 'denoised_436.png', 'denoised_437.png', 'denoised_438.png', 'denoised_439.png', 'denoised_440.png', 'denoised_441.png', 'denoised_442.png', 'denoised_443.png', 'denoised_444.png', 'denoised_445.png', 'denoised_446.png', 'denoised_447.png', 'denoised_448.png', 'denoised_449.png', 'denoised_450.png', 'denoised_451.png', 'denoised_452.png', 'denoised_453.png', 'denoised_454.png', 'denoised_455.png', 'denoised_456.png', 'denoised_457.png', 'denoised_458.png', 'denoised_459.png', 'denoised_460.png', 'denoised_461.png', 'denoised_462.png', 'denoised_463.png', 'denoised_464.png', 'denoised_465.png', 'denoised_466.png', 'denoised_467.png', 'denoised_468.png', 'denoised_469.png', 'denoised_470.png', 'denoised_471.png', 'denoised_472.png', 'denoised_473.png', 'denoised_474.png', 'denoised_475.png', 'denoised_476.png', 'denoised_477.png', 'denoised_478.png', 'denoised_479.png', 'denoised_480.png', 'denoised_481.png', 'denoised_482.png', 'denoised_483.png', 'denoised_484.png', 'denoised_485.png', 'denoised_486.png', 'denoised_487.png', 'denoised_488.png', 'denoised_489.png', 'denoised_490.png', 'denoised_491.png', 'denoised_492.png', 'denoised_493.png', 'denoised_494.png', 'denoised_495.png', 'denoised_496.png', 'denoised_497.png', 'denoised_498.png', 'denoised_499.png', 'denoised_500.png', 'denoised_501.png', 'denoised_502.png', 'denoised_503.png', 'denoised_504.png', 'denoised_505.png', 'denoised_506.png', 'denoised_507.png', 'denoised_508.png', 'denoised_509.png', 'denoised_510.png', 'denoised_511.png', 'denoised_512.png', 'denoised_513.png', 'denoised_514.png', 'denoised_515.png', 'denoised_516.png', 'denoised_517.png', 'denoised_518.png', 'denoised_519.png', 'denoised_520.png', 'denoised_521.png', 'denoised_522.png', 'denoised_523.png', 'denoised_524.png', 'denoised_525.png', 'denoised_526.png', 'denoised_527.png', 'denoised_528.png', 'denoised_529.png', 'denoised_530.png', 'denoised_531.png', 'denoised_532.png', 'denoised_533.png', 'denoised_534.png', 'denoised_535.png', 'denoised_536.png', 'denoised_537.png', 'denoised_538.png', 'denoised_539.png', 'denoised_540.png', 'denoised_541.png', 'denoised_542.png', 'denoised_543.png', 'denoised_544.png', 'denoised_545.png', 'denoised_546.png', 'denoised_547.png', 'denoised_548.png', 'denoised_549.png', 'denoised_550.png', 'denoised_551.png', 'denoised_552.png', 'denoised_553.png', 'denoised_554.png', 'denoised_555.png', 'denoised_556.png', 'denoised_557.png', 'denoised_558.png', 'denoised_559.png', 'denoised_560.png', 'denoised_561.png', 'denoised_562.png', 'denoised_563.png', 'denoised_564.png', 'denoised_565.png', 'denoised_566.png', 'denoised_567.png', 'denoised_568.png', 'denoised_569.png', 'denoised_570.png', 'denoised_571.png', 'denoised_572.png', 'denoised_573.png', 'denoised_574.png', 'denoised_575.png', 'denoised_576.png', 'denoised_577.png', 'denoised_578.png', 'denoised_579.png', 'denoised_580.png', 'denoised_581.png', 'denoised_582.png', 'denoised_583.png', 'denoised_584.png', 'denoised_585.png', 'denoised_586.png', 'denoised_587.png', 'denoised_588.png', 'denoised_589.png', 'denoised_590.png', 'denoised_591.png', 'denoised_592.png', 'denoised_593.png', 'denoised_594.png', 'denoised_595.png', 'denoised_596.png', 'denoised_597.png', 'denoised_598.png', 'denoised_599.png', 'denoised_600.png', 'denoised_601.png', 'denoised_602.png', 'denoised_603.png', 'denoised_604.png', 'denoised_605.png', 'denoised_606.png', 'denoised_607.png', 'denoised_608.png', 'denoised_609.png', 'denoised_610.png', 'denoised_611.png', 'denoised_612.png', 'denoised_613.png', 'denoised_614.png', 'denoised_615.png', 'denoised_616.png', 'denoised_617.png', 'denoised_618.png', 'denoised_619.png', 'denoised_620.png', 'denoised_621.png', 'denoised_622.png', 'denoised_623.png', 'denoised_624.png', 'denoised_625.png', 'denoised_626.png', 'denoised_627.png', 'denoised_628.png', 'denoised_629.png', 'denoised_630.png', 'denoised_631.png', 'denoised_632.png', 'denoised_633.png', 'denoised_634.png', 'denoised_635.png', 'denoised_636.png', 'denoised_637.png', 'denoised_638.png', 'denoised_639.png', 'denoised_640.png', 'denoised_641.png', 'denoised_642.png', 'denoised_643.png', 'denoised_644.png', 'denoised_645.png', 'denoised_646.png', 'denoised_647.png', 'denoised_648.png', 'denoised_649.png', 'denoised_650.png', 'denoised_651.png', 'denoised_652.png', 'denoised_653.png', 'denoised_654.png', 'denoised_655.png', 'denoised_656.png', 'denoised_657.png', 'denoised_658.png', 'denoised_659.png', 'denoised_660.png', 'denoised_661.png', 'denoised_662.png', 'denoised_663.png', 'denoised_664.png', 'denoised_665.png', 'denoised_666.png', 'denoised_667.png', 'denoised_668.png', 'denoised_669.png', 'denoised_670.png', 'denoised_671.png', 'denoised_672.png', 'denoised_673.png', 'denoised_674.png', 'denoised_675.png', 'denoised_676.png', 'denoised_677.png', 'denoised_678.png', 'denoised_679.png', 'denoised_680.png', 'denoised_681.png', 'denoised_682.png', 'denoised_683.png', 'denoised_684.png', 'denoised_685.png', 'denoised_686.png', 'denoised_687.png', 'denoised_688.png', 'denoised_689.png', 'denoised_690.png', 'denoised_691.png', 'denoised_692.png', 'denoised_693.png', 'denoised_694.png', 'denoised_695.png', 'denoised_696.png', 'denoised_697.png', 'denoised_698.png', 'denoised_699.png', 'denoised_700.png', 'denoised_701.png', 'denoised_702.png', 'denoised_703.png', 'denoised_704.png', 'denoised_705.png', 'denoised_706.png', 'denoised_707.png', 'denoised_708.png', 'denoised_709.png', 'denoised_710.png', 'denoised_711.png', 'denoised_712.png', 'denoised_713.png', 'denoised_714.png', 'denoised_715.png', 'denoised_716.png', 'denoised_717.png', 'denoised_718.png', 'denoised_719.png', 'denoised_720.png', 'denoised_721.png', 'denoised_722.png', 'denoised_723.png', 'denoised_724.png', 'denoised_725.png', 'denoised_726.png', 'denoised_727.png', 'denoised_728.png', 'denoised_729.png', 'denoised_730.png', 'denoised_731.png', 'denoised_732.png', 'denoised_733.png', 'denoised_734.png', 'denoised_735.png', 'denoised_736.png', 'denoised_737.png', 'denoised_738.png', 'denoised_739.png', 'denoised_740.png', 'denoised_741.png', 'denoised_742.png', 'denoised_743.png', 'denoised_744.png', 'denoised_745.png', 'denoised_746.png', 'denoised_747.png', 'denoised_748.png', 'denoised_749.png', 'denoised_750.png', 'denoised_751.png', 'denoised_752.png', 'denoised_753.png', 'denoised_754.png', 'denoised_755.png', 'denoised_756.png', 'denoised_757.png', 'denoised_758.png', 'denoised_759.png', 'denoised_760.png', 'denoised_761.png', 'denoised_762.png', 'denoised_763.png', 'denoised_764.png', 'denoised_765.png', 'denoised_766.png', 'denoised_767.png', 'denoised_768.png', 'denoised_769.png', 'denoised_770.png']\n"
     ]
    }
   ],
   "source": [
    "print(filenames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "denoised_1.png\n",
      "denoised_2.png\n",
      "denoised_3.png\n",
      "denoised_4.png\n",
      "denoised_5.png\n",
      "denoised_6.png\n",
      "denoised_7.png\n",
      "denoised_8.png\n",
      "denoised_9.png\n",
      "denoised_10.png\n",
      "denoised_11.png\n",
      "denoised_12.png\n",
      "denoised_13.png\n",
      "denoised_14.png\n",
      "denoised_15.png\n",
      "denoised_16.png\n",
      "denoised_17.png\n",
      "denoised_18.png\n",
      "denoised_19.png\n",
      "denoised_20.png\n",
      "denoised_21.png\n",
      "denoised_22.png\n",
      "denoised_23.png\n",
      "denoised_24.png\n",
      "denoised_25.png\n",
      "denoised_26.png\n",
      "denoised_27.png\n",
      "denoised_28.png\n",
      "denoised_29.png\n",
      "denoised_30.png\n",
      "denoised_31.png\n",
      "denoised_32.png\n",
      "denoised_33.png\n",
      "denoised_34.png\n",
      "denoised_35.png\n",
      "denoised_36.png\n",
      "denoised_37.png\n",
      "denoised_38.png\n",
      "denoised_39.png\n",
      "denoised_40.png\n",
      "denoised_41.png\n",
      "denoised_42.png\n",
      "denoised_43.png\n",
      "denoised_44.png\n",
      "denoised_45.png\n",
      "denoised_46.png\n",
      "denoised_47.png\n",
      "denoised_48.png\n",
      "denoised_49.png\n",
      "denoised_50.png\n",
      "denoised_51.png\n",
      "denoised_52.png\n",
      "denoised_53.png\n",
      "denoised_54.png\n",
      "denoised_55.png\n",
      "denoised_56.png\n",
      "denoised_57.png\n",
      "denoised_58.png\n",
      "denoised_59.png\n",
      "denoised_60.png\n",
      "denoised_61.png\n",
      "denoised_62.png\n",
      "denoised_63.png\n",
      "denoised_64.png\n",
      "denoised_65.png\n",
      "denoised_66.png\n",
      "denoised_67.png\n",
      "denoised_68.png\n",
      "denoised_69.png\n",
      "denoised_70.png\n",
      "denoised_71.png\n",
      "denoised_72.png\n",
      "denoised_73.png\n",
      "denoised_74.png\n",
      "denoised_75.png\n",
      "denoised_76.png\n",
      "denoised_77.png\n",
      "denoised_78.png\n",
      "denoised_79.png\n",
      "denoised_80.png\n",
      "denoised_81.png\n",
      "denoised_82.png\n",
      "denoised_83.png\n",
      "denoised_84.png\n",
      "denoised_85.png\n",
      "denoised_86.png\n",
      "denoised_87.png\n",
      "denoised_88.png\n",
      "denoised_89.png\n",
      "denoised_90.png\n",
      "denoised_91.png\n",
      "denoised_92.png\n",
      "denoised_93.png\n",
      "denoised_94.png\n",
      "denoised_95.png\n",
      "denoised_96.png\n",
      "denoised_97.png\n",
      "denoised_98.png\n",
      "denoised_99.png\n",
      "denoised_100.png\n",
      "denoised_101.png\n",
      "denoised_102.png\n",
      "denoised_103.png\n",
      "denoised_104.png\n",
      "denoised_105.png\n",
      "denoised_106.png\n",
      "denoised_107.png\n",
      "denoised_108.png\n",
      "denoised_109.png\n",
      "denoised_110.png\n",
      "denoised_111.png\n",
      "denoised_112.png\n",
      "denoised_113.png\n",
      "denoised_114.png\n",
      "denoised_115.png\n",
      "denoised_116.png\n",
      "denoised_117.png\n",
      "denoised_118.png\n",
      "denoised_119.png\n",
      "denoised_120.png\n",
      "denoised_121.png\n",
      "denoised_122.png\n",
      "denoised_123.png\n",
      "denoised_124.png\n",
      "denoised_125.png\n",
      "denoised_126.png\n",
      "denoised_127.png\n",
      "denoised_128.png\n",
      "denoised_129.png\n",
      "denoised_130.png\n",
      "denoised_131.png\n",
      "denoised_132.png\n",
      "denoised_133.png\n",
      "denoised_134.png\n",
      "denoised_135.png\n",
      "denoised_136.png\n",
      "denoised_137.png\n",
      "denoised_138.png\n",
      "denoised_139.png\n",
      "denoised_140.png\n",
      "denoised_141.png\n",
      "denoised_142.png\n",
      "denoised_143.png\n",
      "denoised_144.png\n",
      "denoised_145.png\n",
      "denoised_146.png\n",
      "denoised_147.png\n",
      "denoised_148.png\n",
      "denoised_149.png\n",
      "denoised_150.png\n",
      "denoised_151.png\n",
      "denoised_152.png\n",
      "denoised_153.png\n",
      "denoised_154.png\n",
      "denoised_155.png\n",
      "denoised_156.png\n",
      "denoised_157.png\n",
      "denoised_158.png\n",
      "denoised_159.png\n",
      "denoised_160.png\n",
      "denoised_161.png\n",
      "denoised_162.png\n",
      "denoised_163.png\n",
      "denoised_164.png\n",
      "denoised_165.png\n",
      "denoised_166.png\n",
      "denoised_167.png\n",
      "denoised_168.png\n",
      "denoised_169.png\n",
      "denoised_170.png\n",
      "denoised_171.png\n",
      "denoised_172.png\n",
      "denoised_173.png\n",
      "denoised_174.png\n",
      "denoised_175.png\n",
      "denoised_176.png\n",
      "denoised_177.png\n",
      "denoised_178.png\n",
      "denoised_179.png\n",
      "denoised_180.png\n",
      "denoised_181.png\n",
      "denoised_182.png\n",
      "denoised_183.png\n",
      "denoised_184.png\n",
      "denoised_185.png\n",
      "denoised_186.png\n",
      "denoised_187.png\n",
      "denoised_188.png\n",
      "denoised_189.png\n",
      "denoised_190.png\n",
      "denoised_191.png\n",
      "denoised_192.png\n",
      "denoised_193.png\n",
      "denoised_194.png\n",
      "denoised_195.png\n",
      "denoised_196.png\n",
      "denoised_197.png\n",
      "denoised_198.png\n",
      "denoised_199.png\n",
      "denoised_200.png\n",
      "denoised_201.png\n",
      "denoised_202.png\n",
      "denoised_203.png\n",
      "denoised_204.png\n",
      "denoised_205.png\n",
      "denoised_206.png\n",
      "denoised_207.png\n",
      "denoised_208.png\n",
      "denoised_209.png\n",
      "denoised_210.png\n",
      "denoised_211.png\n",
      "denoised_212.png\n",
      "denoised_213.png\n",
      "denoised_214.png\n",
      "denoised_215.png\n",
      "denoised_216.png\n",
      "denoised_217.png\n",
      "denoised_218.png\n",
      "denoised_219.png\n",
      "denoised_220.png\n",
      "denoised_221.png\n",
      "denoised_222.png\n",
      "denoised_223.png\n",
      "denoised_224.png\n",
      "denoised_225.png\n",
      "denoised_226.png\n",
      "denoised_227.png\n",
      "denoised_228.png\n",
      "denoised_229.png\n",
      "denoised_230.png\n",
      "denoised_231.png\n",
      "denoised_232.png\n",
      "denoised_233.png\n",
      "denoised_234.png\n",
      "denoised_235.png\n",
      "denoised_236.png\n",
      "denoised_237.png\n",
      "denoised_238.png\n",
      "denoised_239.png\n",
      "denoised_240.png\n",
      "denoised_241.png\n",
      "denoised_242.png\n",
      "denoised_243.png\n",
      "denoised_244.png\n",
      "denoised_245.png\n",
      "denoised_246.png\n",
      "denoised_247.png\n",
      "denoised_248.png\n",
      "denoised_249.png\n",
      "denoised_250.png\n",
      "denoised_251.png\n",
      "denoised_252.png\n",
      "denoised_253.png\n",
      "denoised_254.png\n",
      "denoised_255.png\n",
      "denoised_256.png\n",
      "denoised_257.png\n",
      "denoised_258.png\n",
      "denoised_259.png\n",
      "denoised_260.png\n",
      "denoised_261.png\n",
      "denoised_262.png\n",
      "denoised_263.png\n",
      "denoised_264.png\n",
      "denoised_265.png\n",
      "denoised_266.png\n",
      "denoised_267.png\n",
      "denoised_268.png\n",
      "denoised_269.png\n",
      "denoised_270.png\n",
      "denoised_271.png\n",
      "denoised_272.png\n",
      "denoised_273.png\n",
      "denoised_274.png\n",
      "denoised_275.png\n",
      "denoised_276.png\n",
      "denoised_277.png\n",
      "denoised_278.png\n",
      "denoised_279.png\n",
      "denoised_280.png\n",
      "denoised_281.png\n",
      "denoised_282.png\n",
      "denoised_283.png\n",
      "denoised_284.png\n",
      "denoised_285.png\n",
      "denoised_286.png\n",
      "denoised_287.png\n",
      "denoised_288.png\n",
      "denoised_289.png\n",
      "denoised_290.png\n",
      "denoised_291.png\n",
      "denoised_292.png\n",
      "denoised_293.png\n",
      "denoised_294.png\n",
      "denoised_295.png\n",
      "denoised_296.png\n",
      "denoised_297.png\n",
      "denoised_298.png\n",
      "denoised_299.png\n",
      "denoised_300.png\n",
      "denoised_301.png\n",
      "denoised_302.png\n",
      "denoised_303.png\n",
      "denoised_304.png\n",
      "denoised_305.png\n",
      "denoised_306.png\n",
      "denoised_307.png\n",
      "denoised_308.png\n",
      "denoised_309.png\n",
      "denoised_310.png\n",
      "denoised_311.png\n",
      "denoised_312.png\n",
      "denoised_313.png\n",
      "denoised_314.png\n",
      "denoised_315.png\n",
      "denoised_316.png\n",
      "denoised_317.png\n",
      "denoised_318.png\n",
      "denoised_319.png\n",
      "denoised_320.png\n",
      "denoised_321.png\n",
      "denoised_322.png\n",
      "denoised_323.png\n",
      "denoised_324.png\n",
      "denoised_325.png\n",
      "denoised_326.png\n",
      "denoised_327.png\n",
      "denoised_328.png\n",
      "denoised_329.png\n",
      "denoised_330.png\n",
      "denoised_331.png\n",
      "denoised_332.png\n",
      "denoised_333.png\n",
      "denoised_334.png\n",
      "denoised_335.png\n",
      "denoised_336.png\n",
      "denoised_337.png\n",
      "denoised_338.png\n",
      "denoised_339.png\n",
      "denoised_340.png\n",
      "denoised_341.png\n",
      "denoised_342.png\n",
      "denoised_343.png\n",
      "denoised_344.png\n",
      "denoised_345.png\n",
      "denoised_346.png\n",
      "denoised_347.png\n",
      "denoised_348.png\n",
      "denoised_349.png\n",
      "denoised_350.png\n",
      "denoised_351.png\n",
      "denoised_352.png\n",
      "denoised_353.png\n",
      "denoised_354.png\n",
      "denoised_355.png\n",
      "denoised_356.png\n",
      "denoised_357.png\n",
      "denoised_358.png\n",
      "denoised_359.png\n",
      "denoised_360.png\n",
      "denoised_361.png\n",
      "denoised_362.png\n",
      "denoised_363.png\n",
      "denoised_364.png\n",
      "denoised_365.png\n",
      "denoised_366.png\n",
      "denoised_367.png\n",
      "denoised_368.png\n",
      "denoised_369.png\n",
      "denoised_370.png\n",
      "denoised_371.png\n",
      "denoised_372.png\n",
      "denoised_373.png\n",
      "denoised_374.png\n",
      "denoised_375.png\n",
      "denoised_376.png\n",
      "denoised_377.png\n",
      "denoised_378.png\n",
      "denoised_379.png\n",
      "denoised_380.png\n",
      "denoised_381.png\n",
      "denoised_382.png\n",
      "denoised_383.png\n",
      "denoised_384.png\n",
      "denoised_385.png\n",
      "denoised_386.png\n",
      "denoised_387.png\n",
      "denoised_388.png\n",
      "denoised_389.png\n",
      "denoised_390.png\n",
      "denoised_391.png\n",
      "denoised_392.png\n",
      "denoised_393.png\n",
      "denoised_394.png\n",
      "denoised_395.png\n",
      "denoised_396.png\n",
      "denoised_397.png\n",
      "denoised_398.png\n",
      "denoised_399.png\n",
      "denoised_400.png\n",
      "denoised_401.png\n",
      "denoised_402.png\n",
      "denoised_403.png\n",
      "denoised_404.png\n",
      "denoised_405.png\n",
      "denoised_406.png\n",
      "denoised_407.png\n",
      "denoised_408.png\n",
      "denoised_409.png\n",
      "denoised_410.png\n",
      "denoised_411.png\n",
      "denoised_412.png\n",
      "denoised_413.png\n",
      "denoised_414.png\n",
      "denoised_415.png\n",
      "denoised_416.png\n",
      "denoised_417.png\n",
      "denoised_418.png\n",
      "denoised_419.png\n",
      "denoised_420.png\n",
      "denoised_421.png\n",
      "denoised_422.png\n",
      "denoised_423.png\n",
      "denoised_424.png\n",
      "denoised_425.png\n",
      "denoised_426.png\n",
      "denoised_427.png\n",
      "denoised_428.png\n",
      "denoised_429.png\n",
      "denoised_430.png\n",
      "denoised_431.png\n",
      "denoised_432.png\n",
      "denoised_433.png\n",
      "denoised_434.png\n",
      "denoised_435.png\n",
      "denoised_436.png\n",
      "denoised_437.png\n",
      "denoised_438.png\n",
      "denoised_439.png\n",
      "denoised_440.png\n",
      "denoised_441.png\n",
      "denoised_442.png\n",
      "denoised_443.png\n",
      "denoised_444.png\n",
      "denoised_445.png\n",
      "denoised_446.png\n",
      "denoised_447.png\n",
      "denoised_448.png\n",
      "denoised_449.png\n",
      "denoised_450.png\n",
      "denoised_451.png\n",
      "denoised_452.png\n",
      "denoised_453.png\n",
      "denoised_454.png\n",
      "denoised_455.png\n",
      "denoised_456.png\n",
      "denoised_457.png\n",
      "denoised_458.png\n",
      "denoised_459.png\n",
      "denoised_460.png\n",
      "denoised_461.png\n",
      "denoised_462.png\n",
      "denoised_463.png\n",
      "denoised_464.png\n",
      "denoised_465.png\n",
      "denoised_466.png\n",
      "denoised_467.png\n",
      "denoised_468.png\n",
      "denoised_469.png\n",
      "denoised_470.png\n",
      "denoised_471.png\n",
      "denoised_472.png\n",
      "denoised_473.png\n",
      "denoised_474.png\n",
      "denoised_475.png\n",
      "denoised_476.png\n",
      "denoised_477.png\n",
      "denoised_478.png\n",
      "denoised_479.png\n",
      "denoised_480.png\n",
      "denoised_481.png\n",
      "denoised_482.png\n",
      "denoised_483.png\n",
      "denoised_484.png\n",
      "denoised_485.png\n",
      "denoised_486.png\n",
      "denoised_487.png\n",
      "denoised_488.png\n",
      "denoised_489.png\n",
      "denoised_490.png\n",
      "denoised_491.png\n",
      "denoised_492.png\n",
      "denoised_493.png\n",
      "denoised_494.png\n",
      "denoised_495.png\n",
      "denoised_496.png\n",
      "denoised_497.png\n",
      "denoised_498.png\n",
      "denoised_499.png\n",
      "denoised_500.png\n",
      "denoised_501.png\n",
      "denoised_502.png\n",
      "denoised_503.png\n",
      "denoised_504.png\n",
      "denoised_505.png\n",
      "denoised_506.png\n",
      "denoised_507.png\n",
      "denoised_508.png\n",
      "denoised_509.png\n",
      "denoised_510.png\n",
      "denoised_511.png\n",
      "denoised_512.png\n",
      "denoised_513.png\n",
      "denoised_514.png\n",
      "denoised_515.png\n",
      "denoised_516.png\n",
      "denoised_517.png\n",
      "denoised_518.png\n",
      "denoised_519.png\n",
      "denoised_520.png\n",
      "denoised_521.png\n",
      "denoised_522.png\n",
      "denoised_523.png\n",
      "denoised_524.png\n",
      "denoised_525.png\n",
      "denoised_526.png\n",
      "denoised_527.png\n",
      "denoised_528.png\n",
      "denoised_529.png\n",
      "denoised_530.png\n",
      "denoised_531.png\n",
      "denoised_532.png\n",
      "denoised_533.png\n",
      "denoised_534.png\n",
      "denoised_535.png\n",
      "denoised_536.png\n",
      "denoised_537.png\n",
      "denoised_538.png\n",
      "denoised_539.png\n",
      "denoised_540.png\n",
      "denoised_541.png\n",
      "denoised_542.png\n",
      "denoised_543.png\n",
      "denoised_544.png\n",
      "denoised_545.png\n",
      "denoised_546.png\n",
      "denoised_547.png\n",
      "denoised_548.png\n",
      "denoised_549.png\n",
      "denoised_550.png\n",
      "denoised_551.png\n",
      "denoised_552.png\n",
      "denoised_553.png\n",
      "denoised_554.png\n",
      "denoised_555.png\n",
      "denoised_556.png\n",
      "denoised_557.png\n",
      "denoised_558.png\n",
      "denoised_559.png\n",
      "denoised_560.png\n",
      "denoised_561.png\n",
      "denoised_562.png\n",
      "denoised_563.png\n",
      "denoised_564.png\n",
      "denoised_565.png\n",
      "denoised_566.png\n",
      "denoised_567.png\n",
      "denoised_568.png\n",
      "denoised_569.png\n",
      "denoised_570.png\n",
      "denoised_571.png\n",
      "denoised_572.png\n",
      "denoised_573.png\n",
      "denoised_574.png\n",
      "denoised_575.png\n",
      "denoised_576.png\n",
      "denoised_577.png\n",
      "denoised_578.png\n",
      "denoised_579.png\n",
      "denoised_580.png\n",
      "denoised_581.png\n",
      "denoised_582.png\n",
      "denoised_583.png\n",
      "denoised_584.png\n",
      "denoised_585.png\n",
      "denoised_586.png\n",
      "denoised_587.png\n",
      "denoised_588.png\n",
      "denoised_589.png\n",
      "denoised_590.png\n",
      "denoised_591.png\n",
      "denoised_592.png\n",
      "denoised_593.png\n",
      "denoised_594.png\n",
      "denoised_595.png\n",
      "denoised_596.png\n",
      "denoised_597.png\n",
      "denoised_598.png\n",
      "denoised_599.png\n",
      "denoised_600.png\n",
      "denoised_601.png\n",
      "denoised_602.png\n",
      "denoised_603.png\n",
      "denoised_604.png\n",
      "denoised_605.png\n",
      "denoised_606.png\n",
      "denoised_607.png\n",
      "denoised_608.png\n",
      "denoised_609.png\n",
      "denoised_610.png\n",
      "denoised_611.png\n",
      "denoised_612.png\n",
      "denoised_613.png\n",
      "denoised_614.png\n",
      "denoised_615.png\n",
      "denoised_616.png\n",
      "denoised_617.png\n",
      "denoised_618.png\n",
      "denoised_619.png\n",
      "denoised_620.png\n",
      "denoised_621.png\n",
      "denoised_622.png\n",
      "denoised_623.png\n",
      "denoised_624.png\n",
      "denoised_625.png\n",
      "denoised_626.png\n",
      "denoised_627.png\n",
      "denoised_628.png\n",
      "denoised_629.png\n",
      "denoised_630.png\n",
      "denoised_631.png\n",
      "denoised_632.png\n",
      "denoised_633.png\n",
      "denoised_634.png\n",
      "denoised_635.png\n",
      "denoised_636.png\n",
      "denoised_637.png\n",
      "denoised_638.png\n",
      "denoised_639.png\n",
      "denoised_640.png\n",
      "denoised_641.png\n",
      "denoised_642.png\n",
      "denoised_643.png\n",
      "denoised_644.png\n",
      "denoised_645.png\n",
      "denoised_646.png\n",
      "denoised_647.png\n",
      "denoised_648.png\n",
      "denoised_649.png\n",
      "denoised_650.png\n",
      "denoised_651.png\n",
      "denoised_652.png\n",
      "denoised_653.png\n",
      "denoised_654.png\n",
      "denoised_655.png\n",
      "denoised_656.png\n",
      "denoised_657.png\n",
      "denoised_658.png\n",
      "denoised_659.png\n",
      "denoised_660.png\n",
      "denoised_661.png\n",
      "denoised_662.png\n",
      "denoised_663.png\n",
      "denoised_664.png\n",
      "denoised_665.png\n",
      "denoised_666.png\n",
      "denoised_667.png\n",
      "denoised_668.png\n",
      "denoised_669.png\n",
      "denoised_670.png\n",
      "denoised_671.png\n",
      "denoised_672.png\n",
      "denoised_673.png\n",
      "denoised_674.png\n",
      "denoised_675.png\n",
      "denoised_676.png\n",
      "denoised_677.png\n",
      "denoised_678.png\n",
      "denoised_679.png\n",
      "denoised_680.png\n",
      "denoised_681.png\n",
      "denoised_682.png\n",
      "denoised_683.png\n",
      "denoised_684.png\n",
      "denoised_685.png\n",
      "denoised_686.png\n",
      "denoised_687.png\n",
      "denoised_688.png\n",
      "denoised_689.png\n",
      "denoised_690.png\n",
      "denoised_691.png\n",
      "denoised_692.png\n",
      "denoised_693.png\n",
      "denoised_694.png\n",
      "denoised_695.png\n",
      "denoised_696.png\n",
      "denoised_697.png\n",
      "denoised_698.png\n",
      "denoised_699.png\n",
      "denoised_700.png\n",
      "denoised_701.png\n",
      "denoised_702.png\n",
      "denoised_703.png\n",
      "denoised_704.png\n",
      "denoised_705.png\n",
      "denoised_706.png\n",
      "denoised_707.png\n",
      "denoised_708.png\n",
      "denoised_709.png\n",
      "denoised_710.png\n",
      "denoised_711.png\n",
      "denoised_712.png\n",
      "denoised_713.png\n",
      "denoised_714.png\n",
      "denoised_715.png\n",
      "denoised_716.png\n",
      "denoised_717.png\n",
      "denoised_718.png\n",
      "denoised_719.png\n",
      "denoised_720.png\n",
      "denoised_721.png\n",
      "denoised_722.png\n",
      "denoised_723.png\n",
      "denoised_724.png\n",
      "denoised_725.png\n",
      "denoised_726.png\n",
      "denoised_727.png\n",
      "denoised_728.png\n",
      "denoised_729.png\n",
      "denoised_730.png\n",
      "denoised_731.png\n",
      "denoised_732.png\n",
      "denoised_733.png\n",
      "denoised_734.png\n",
      "denoised_735.png\n",
      "denoised_736.png\n",
      "denoised_737.png\n",
      "denoised_738.png\n",
      "denoised_739.png\n",
      "denoised_740.png\n",
      "denoised_741.png\n",
      "denoised_742.png\n",
      "denoised_743.png\n",
      "denoised_744.png\n",
      "denoised_745.png\n",
      "denoised_746.png\n",
      "denoised_747.png\n",
      "denoised_748.png\n",
      "denoised_749.png\n",
      "denoised_750.png\n",
      "denoised_751.png\n",
      "denoised_752.png\n",
      "denoised_753.png\n",
      "denoised_754.png\n",
      "denoised_755.png\n",
      "denoised_756.png\n",
      "denoised_757.png\n",
      "denoised_758.png\n",
      "denoised_759.png\n",
      "denoised_760.png\n",
      "denoised_761.png\n",
      "denoised_762.png\n",
      "denoised_763.png\n",
      "denoised_764.png\n",
      "denoised_765.png\n",
      "denoised_766.png\n",
      "denoised_767.png\n",
      "denoised_768.png\n",
      "denoised_769.png\n",
      "denoised_770.png\n"
     ]
    }
   ],
   "source": [
    "images = []\n",
    "for filename in filenames :\n",
    "    filepath = os.path.join(root,filename)\n",
    "    image = ndimage.imread(filepath, mode = \"L\")\n",
    "    images.append(image)\n",
    "    print(filename)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f5ccea02210>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAGoAAAEACAYAAABbOeMfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngFoTuQJqTeT/A/7GoCZcGwLRAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5ccedac190>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(images[0],cmap = \"gray\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(216, 64)\n"
     ]
    }
   ],
   "source": [
    "print (images[0].shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "labels = np.load('resized_cropped_labeledimages.npy')\n",
    "labels_list = []\n",
    "for i in range(len(labels)):\n",
    "    labels_list.append(labels[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(216, 64)\n"
     ]
    }
   ],
   "source": [
    "print(labels_list[0].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels_list[0][0][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_labels = np.zeros((770,216,64,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for i in range(len(labels_list)) :\n",
    "    for j in range(216) :\n",
    "        for k in range(64):\n",
    "            if(labels_list[i][j][k] == 0):\n",
    "                train_labels[i][j][k][0] = 1\n",
    "            if(labels_list[i][j][k] == 1):\n",
    "                train_labels[i][j][k][1] = 1\n",
    "            if(labels_list[i][j][k] == 2):\n",
    "                train_labels[i][j][k][2] = 1\n",
    "            if(labels_list[i][j][k] == 3):\n",
    "                train_labels[i][j][k][3] = 1\n",
    "            if(labels_list[i][j][k] == 4):\n",
    "                train_labels[i][j][k][4] = 1\n",
    "            if(labels_list[i][j][k] == 5):\n",
    "                train_labels[i][j][k][5] = 1\n",
    "            if(labels_list[i][j][k] == 6):\n",
    "                train_labels[i][j][k][6] = 1\n",
    "            if(labels_list[i][j][k] == 7):\n",
    "                train_labels[i][j][k][7] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(770, 216, 64, 8)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 1.,  0.,  0.,  0.,  0.,  0.,  0.,  0.])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_labels[0][0][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "770\n",
      "(216, 64, 1)\n",
      "[2, 4, 5, 6, 8, 10, 11, 12, 14, 15, 16, 17, 19, 20, 21, 22, 23, 26, 28, 29, 30, 31, 32, 33, 34, 35, 37, 38, 40, 41, 42, 43, 45, 47, 49, 50, 51, 52, 53, 54, 56, 57, 59, 61, 62, 63, 64, 65, 66, 68, 69, 70, 72, 73, 76, 77, 80, 81, 82, 83, 86, 87, 89, 91, 96, 97, 98, 99, 101, 103, 104, 105, 106, 107, 108, 110, 111, 113, 114, 116, 117, 118, 119, 120, 122, 124, 126, 128, 129, 130, 131, 132, 133, 138, 139, 140, 144, 147, 149, 152, 154, 155, 156, 157, 158, 159, 161, 164, 165, 166, 170, 172, 174, 175, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 195, 196, 197, 198, 199, 200, 201, 202, 205, 207, 208, 210, 211, 213, 215, 218, 219, 220, 221, 222, 223, 224, 226, 227, 228, 231, 232, 233, 234, 237, 240, 242, 243, 244, 245, 246, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 261, 262, 263, 264, 265, 266, 267, 269, 270, 272, 273, 274, 275, 276, 277, 280, 281, 285, 286, 288, 289, 292, 295, 296, 297, 298, 301, 303, 305, 306, 309, 310, 311, 312, 313, 315, 316, 317, 318, 319, 320, 322, 323, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 348, 349, 350, 351, 352, 355, 361, 362, 363, 365, 366, 367, 369, 371, 374, 376, 379, 381, 382, 384, 385, 386, 388, 390, 392, 393, 394, 397, 398, 399, 402, 405, 406, 407, 410, 411, 414, 415, 417, 418, 419, 420, 422, 423, 425, 426, 429, 430, 431, 432, 435, 438, 439, 440, 442, 443, 444, 445, 446, 448, 450, 451, 454, 455, 457, 459, 463, 464, 465, 466, 468, 470, 471, 476, 478, 481, 482, 483, 484, 486, 487, 488, 491, 492, 494, 495, 497, 498, 499, 500, 501, 502, 503, 504, 505, 506, 507, 508, 509, 510, 513, 515, 516, 517, 520, 521, 522, 525, 526, 529, 530, 531, 533, 534, 538, 539, 542, 543, 544, 545, 546, 547, 549, 552, 554, 555, 556, 559, 561, 562, 564, 566, 567, 568, 571, 573, 574, 575, 576, 578, 579, 583, 588, 591, 592, 594, 595, 597, 600, 601, 609, 610, 613, 616, 620, 622, 623, 624, 627, 628, 629, 630, 633, 634, 636, 640, 641, 642, 645, 646, 647, 648, 649, 650, 651, 652, 654, 659, 661, 663, 664, 665, 666, 667, 668, 669, 671, 672, 673, 675, 678, 679, 680, 681, 682, 685, 686, 688, 690, 691, 694, 695, 696, 697, 701, 702, 703, 705, 706, 707, 708, 709, 710, 711, 712, 714, 715, 717, 718, 719, 721, 722, 723, 724, 725, 726, 727, 729, 731, 732, 734, 736, 737, 738, 739, 740, 741, 742, 743, 744, 745, 746, 748, 749, 750, 751, 752, 753, 756, 757, 758, 759, 762, 763, 764, 765, 766, 768, 769]\n",
      "[0, 1, 3, 7, 9, 13, 18, 24, 25, 27, 36, 39, 44, 46, 48, 55, 58, 60, 67, 71, 74, 75, 78, 79, 84, 85, 88, 90, 92, 93, 94, 95, 100, 102, 109, 112, 115, 121, 123, 125, 127, 134, 135, 136, 137, 141, 142, 143, 145, 146, 148, 150, 151, 153, 160, 162, 163, 167, 168, 169, 171, 173, 176, 193, 194, 203, 204, 206, 209, 212, 214, 216, 217, 225, 229, 230, 235, 236, 238, 239, 241, 247, 249, 260, 268, 271, 278, 279, 282, 283, 284, 287, 290, 291, 293, 294, 299, 300, 302, 304, 307, 308, 314, 321, 324, 330, 347, 353, 354, 356, 357, 358, 359, 360, 364, 368, 370, 372, 373, 375, 377, 378, 380, 383, 387, 389, 391, 395, 396, 400, 401, 403, 404, 408, 409, 412, 413, 416, 421, 424, 427, 428, 433, 434, 436, 437, 441, 447, 449, 452, 453, 456, 458, 460, 461, 462, 467, 469, 472, 473, 474, 475, 477, 479, 480, 485, 489, 490, 493, 496, 511, 512, 514, 518, 519, 523, 524, 527, 528, 532, 535, 536, 537, 540, 541, 548, 550, 551, 553, 557, 558, 560, 563, 565, 569, 570, 572, 577, 580, 581, 582, 584, 585, 586, 587, 589, 590, 593, 596, 598, 599, 602, 603, 604, 605, 606, 607, 608, 611, 612, 614, 615, 617, 618, 619, 621, 625, 626, 631, 632, 635, 637, 638, 639, 643, 644, 653, 655, 656, 657, 658, 660, 662, 670, 674, 676, 677, 683, 684, 687, 689, 692, 693, 698, 699, 700, 704, 713, 716, 720, 728, 730, 733, 735, 747, 754, 755, 760, 761, 767]\n"
     ]
    }
   ],
   "source": [
    "images=np.array(images)\n",
    "print(images.shape[0])\n",
    "images = images.reshape(images.shape[0],216,64,1)\n",
    "\n",
    "print(images[0].shape)\n",
    "train_indices = np.random.choice(770,500,replace = False)\n",
    "print(sorted(train_indices))\n",
    "train_images_random = []\n",
    "train_labels_random = []\n",
    "\n",
    "for i in train_indices:\n",
    "    train_images_random.append(images[i])\n",
    "    train_labels_random.append(train_labels[i])\n",
    "\n",
    "test_indices = [x for x in range(770) if x not in train_indices]\n",
    "print(test_indices)\n",
    "test_images = []\n",
    "test_labels = []\n",
    "for i in test_indices:\n",
    "    test_images.append(images[i])\n",
    "    test_labels.append(train_labels[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_images = np.array(train_images_random)\n",
    "train_labels = np.array(train_labels_random)\n",
    "test_images = np.array(test_images)\n",
    "test_labels = np.array(test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_images = train_images.astype('float32')\n",
    "train_labels = train_labels.astype('float32')\n",
    "test_images = test_images.astype('float32')\n",
    "test_labels = test_labels.astype('float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(500, 216, 64, 1)\n",
      "(270, 216, 64, 1)\n",
      "(500, 216, 64, 8)\n"
     ]
    }
   ],
   "source": [
    "print(train_images.shape)\n",
    "print(test_images.shape)\n",
    "print(train_labels.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[[  2.]\n",
      "  [  0.]\n",
      "  [  0.]\n",
      "  ..., \n",
      "  [ 11.]\n",
      "  [ 18.]\n",
      "  [ 13.]]\n",
      "\n",
      " [[  0.]\n",
      "  [  0.]\n",
      "  [  0.]\n",
      "  ..., \n",
      "  [ 10.]\n",
      "  [ 16.]\n",
      "  [ 17.]]\n",
      "\n",
      " [[  2.]\n",
      "  [  4.]\n",
      "  [  1.]\n",
      "  ..., \n",
      "  [ 10.]\n",
      "  [ 17.]\n",
      "  [ 19.]]\n",
      "\n",
      " ..., \n",
      " [[ 24.]\n",
      "  [ 23.]\n",
      "  [ 22.]\n",
      "  ..., \n",
      "  [ 13.]\n",
      "  [ 12.]\n",
      "  [ 12.]]\n",
      "\n",
      " [[ 24.]\n",
      "  [ 24.]\n",
      "  [ 23.]\n",
      "  ..., \n",
      "  [ 12.]\n",
      "  [ 13.]\n",
      "  [ 14.]]\n",
      "\n",
      " [[ 25.]\n",
      "  [ 24.]\n",
      "  [ 23.]\n",
      "  ..., \n",
      "  [ 12.]\n",
      "  [ 13.]\n",
      "  [ 14.]]]\n"
     ]
    }
   ],
   "source": [
    "print(train_images[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "500\n",
      "500\n",
      "(500, 216, 64, 8)\n"
     ]
    }
   ],
   "source": [
    "print (len(train_images))\n",
    "print (len(train_labels))\n",
    "print (np.array(train_labels).shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(500, 216, 64, 1)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_images.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "500"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train_images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "data_shape = 216*64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "weight_decay = 0.0001"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 216, 64, 1)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 216, 64, 64)       640       \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 216, 64, 64)       256       \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 216, 64, 64)       0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 108, 32, 64)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 108, 32, 64)       36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 108, 32, 64)       256       \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 108, 32, 64)       0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 54, 16, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 54, 16, 64)        36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 54, 16, 64)        256       \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 54, 16, 64)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 27, 8, 64)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 27, 8, 64)         36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_4 (Batch (None, 27, 8, 64)         256       \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 27, 8, 64)         0         \n",
      "_________________________________________________________________\n",
      "up_sampling2d_1 (UpSampling2 (None, 54, 16, 64)        0         \n",
      "_________________________________________________________________\n",
      "concatenate_1 (Concatenate)  (None, 54, 16, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (None, 54, 16, 64)        73792     \n",
      "_________________________________________________________________\n",
      "batch_normalization_5 (Batch (None, 54, 16, 64)        256       \n",
      "_________________________________________________________________\n",
      "activation_5 (Activation)    (None, 54, 16, 64)        0         \n",
      "_________________________________________________________________\n",
      "up_sampling2d_2 (UpSampling2 (None, 108, 32, 64)       0         \n",
      "_________________________________________________________________\n",
      "concatenate_2 (Concatenate)  (None, 108, 32, 128)      0         \n",
      "_________________________________________________________________\n",
      "conv2d_6 (Conv2D)            (None, 108, 32, 64)       73792     \n",
      "_________________________________________________________________\n",
      "batch_normalization_6 (Batch (None, 108, 32, 64)       256       \n",
      "_________________________________________________________________\n",
      "activation_6 (Activation)    (None, 108, 32, 64)       0         \n",
      "_________________________________________________________________\n",
      "Layer19 (UpSampling2D)       (None, 216, 64, 64)       0         \n",
      "_________________________________________________________________\n",
      "concatenate_3 (Concatenate)  (None, 216, 64, 128)      0         \n",
      "_________________________________________________________________\n",
      "conv2d_7 (Conv2D)            (None, 216, 64, 64)       73792     \n",
      "_________________________________________________________________\n",
      "batch_normalization_7 (Batch (None, 216, 64, 64)       256       \n",
      "_________________________________________________________________\n",
      "activation_7 (Activation)    (None, 216, 64, 64)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_8 (Conv2D)            (None, 216, 64, 8)        520       \n",
      "_________________________________________________________________\n",
      "reshape_1 (Reshape)          (None, 13824, 8)          0         \n",
      "_________________________________________________________________\n",
      "activation_8 (Activation)    (None, 13824, 8)          0         \n",
      "=================================================================\n",
      "Total params: 335,112.0\n",
      "Trainable params: 334,216.0\n",
      "Non-trainable params: 896.0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Defines the input tensor\n",
    "inputs = Input(shape=(216,64,1))\n",
    "\n",
    "L1 = Conv2D(64,kernel_size=(3,3),padding = \"same\",kernel_regularizer=regularizers.l2(weight_decay))(inputs)\n",
    "L2 = BatchNormalization()(L1)\n",
    "L2 = Activation('relu')(L2)\n",
    "#L3 = Lambda(maxpool_1,output_shape = shape)(L2)\n",
    "L3 = MaxPooling2D(pool_size=(2,2))(L2)\n",
    "L4 = Conv2D(64,kernel_size=(3,3),padding = \"same\",kernel_regularizer=regularizers.l2(weight_decay))(L3)\n",
    "L5 = BatchNormalization()(L4)\n",
    "L5 = Activation('relu')(L5)\n",
    "#L6 = Lambda(maxpool_2,output_shape = shape)(L5)\n",
    "L6 = MaxPooling2D(pool_size=(2,2))(L5)\n",
    "L7 = Conv2D(64,kernel_size=(3,3),padding = \"same\",kernel_regularizer=regularizers.l2(weight_decay))(L6)\n",
    "L8 = BatchNormalization()(L7)\n",
    "L8 = Activation('relu')(L8)\n",
    "#L9 = Lambda(maxpool_3,output_shape = shape)(L8)\n",
    "L9 = MaxPooling2D(pool_size=(2,2))(L8)\n",
    "L10 = Conv2D(64,kernel_size=(3,3),padding = \"same\",kernel_regularizer=regularizers.l2(weight_decay))(L9)\n",
    "L11 = BatchNormalization()(L10)\n",
    "L11 = Activation('relu')(L11)\n",
    "L12 = UpSampling2D(size = (2,2))(L11)\n",
    "#L12 = Lambda(unpool_3,output_shape = unpool_shape)(L11)\n",
    "L13 = Concatenate(axis = 3)([L8,L12])\n",
    "L14 = Conv2D(64,kernel_size=(3,3),padding = \"same\",kernel_regularizer=regularizers.l2(weight_decay))(L13)\n",
    "L15 = BatchNormalization()(L14)\n",
    "L15 = Activation('relu')(L15)\n",
    "L16 = UpSampling2D(size= (2,2))(L15)\n",
    "#L16 = Lambda(unpool_2,output_shape=unpool_shape)(L15)\n",
    "L17 = Concatenate(axis = 3)([L16,L5])\n",
    "L18 = Conv2D(64,kernel_size=(3,3),padding = \"same\",kernel_regularizer=regularizers.l2(weight_decay))(L17)\n",
    "L19 = BatchNormalization()(L18)\n",
    "L19 = Activation('relu')(L19)\n",
    "#L20 = Lambda(unpool_1,output_shape=unpool_shape)(L19)\n",
    "L20 = UpSampling2D(size=(2,2),name = \"Layer19\")(L19)\n",
    "L21 = Concatenate(axis=3)([L20,L2])\n",
    "L22 = Conv2D(64,kernel_size=(3,3),padding = \"same\",kernel_regularizer=regularizers.l2(weight_decay))(L21)\n",
    "L23 = BatchNormalization()(L22)\n",
    "L23 = Activation('relu')(L23)\n",
    "L24 = Conv2D(8,kernel_size=(1,1),padding = \"same\",kernel_regularizer=regularizers.l2(weight_decay))(L23)\n",
    "L = Reshape((data_shape,8),input_shape = (216,64,8))(L24)\n",
    "L = Activation('softmax')(L)\n",
    "model = Model(inputs = inputs, outputs = L)\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "weights = np.load('weighted_cropped_images.npy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(770, 216, 64)"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weights.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weights[0][0][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  1.   ,   4.336,   6.63 ,  12.007,  12.459,  14.647,  15.368,\n",
       "        17.978,  19.336,  21.63 ,  27.007,  27.459,  29.647,  30.368,\n",
       "        32.978])"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.unique(weights)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "weights_matrix = []\n",
    "for i in train_indices:\n",
    "    weights_matrix.append(weights[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sample_weights = np.array(weights_matrix)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sample_weights = np.reshape(sample_weights,(500,data_shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(500, 13824)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_weights.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "500"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train_images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(500, 216, 64, 8)"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_labels = np.reshape(train_labels,(500,data_shape,8))\n",
    "test_labels = np.reshape(test_labels,(270,data_shape,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(500, 13824, 8)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "count = np.zeros(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for i in range(500):\n",
    "    for j in range(13824):\n",
    "        for k in range(8):\n",
    "            if(train_labels[i][j][k]==1):\n",
    "                count[k] = count[k]+1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 408,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3972411.0\n",
      "312782.0\n",
      "605075.0\n",
      "341224.0\n",
      "263099.0\n",
      "920815.0\n",
      "274885.0\n",
      "221709.0\n"
     ]
    }
   ],
   "source": [
    "for i in range(8):\n",
    "    print(count[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 409,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 3972411.,   312782.,   605075.,   341224.,   263099.,   920815.,\n",
       "         274885.,   221709.])"
      ]
     },
     "execution_count": 409,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 410,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "median = np.median(count)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 411,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "327003.0"
      ]
     },
     "execution_count": 411,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "median"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 364,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "scale = np.zeros(8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 367,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for i in range(8):\n",
    "    scale[i] = (median/count[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 368,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.08176891,  1.01882538,  0.54272503,  0.98185769,  1.25668549,\n",
       "        0.354605  ,  1.19773269,  1.47007454])"
      ]
     },
     "execution_count": 368,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scale"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 369,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "weights = np.zeros(8)\n",
    "for i in range(8):\n",
    "    weights[i] = scale[i]/scale[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 370,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  1.        ,  12.45981304,   6.63730261,  12.00771353,\n",
       "        15.36874386,   4.33667245,  14.6477755 ,  17.97840371])"
      ]
     },
     "execution_count": 370,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "smooth = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def dice_coef(y_true, y_pred):\n",
    "    y_true_f = K.flatten(y_true)\n",
    "    y_pred_f = K.flatten(y_pred)\n",
    "    intersection = K.sum(y_true_f * y_pred_f)\n",
    "    return (2. * intersection + smooth) / (K.sum(y_true_f) + K.sum(y_pred_f) + smooth)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def dice_coef_loss(y_true, y_pred):\n",
    "    return -dice_coef(y_true, y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def customized_loss(y_true,y_pred):\n",
    "    return (1*K.categorical_crossentropy(y_true, y_pred))+(0.5*dice_coef_loss(y_true, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "optimiser = optimizers.Adam(lr = 0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.compile(optimizer=optimiser,loss=customized_loss,metrics=['accuracy',dice_coef],sample_weight_mode='temporal')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(216, 64, 1)"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_images[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#Defining Callback functions which will be called by model during runtime when specified condition satisfies\n",
    "lr_reducer = ReduceLROnPlateau(factor=0.5, cooldown=0, patience=6, min_lr=0.5e-6)\n",
    "csv_logger = CSVLogger('Relaynet_sample_weights_denoised_lr_e2_testing_bs_20.csv')\n",
    "model_chekpoint = ModelCheckpoint(\"Relaynet_sample_weights_denoised_lr_e2_testing_bs_20.hdf5\",monitor = 'val_loss',verbose = 1,save_best_only=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 500 samples, validate on 270 samples\n",
      "Epoch 1/200\n",
      "380/500 [=====================>........] - ETA: 15s - loss: 45.2449 - acc: 0.6268 - dice_coef: 0.5852"
     ]
    }
   ],
   "source": [
    "\n",
    "model.fit(train_images,train_labels,batch_size=20,epochs=200,validation_data=(test_images,test_labels),sample_weight=sample_weights,callbacks=[lr_reducer, csv_logger,model_chekpoint])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.load_weights(\"Relaynet_sample_weights_denoised_lr_e2_testing_bs_20.hdf5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "testing_image = train_images[28]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(216, 64, 1)"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "testing_image.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "testing_image = np.squeeze(testing_image,axis = 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(216, 64)"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "testing_image.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f5c7c7f4ed0>"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAGoAAAEACAYAAABbOeMfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5c84468650>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(testing_image,cmap = \"gray\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "testing_image = testing_image.reshape((1,216,64,1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "prediction = model.predict(testing_image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "prediction = np.squeeze(prediction,axis = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(13824, 8)\n"
     ]
    }
   ],
   "source": [
    "print(prediction.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.argmax(prediction[6999])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "prediction = np.reshape(prediction,(216,64,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "output = np.zeros((216,64))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ground = np.zeros((216,64))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for i in range(216):\n",
    "    for j in range(64):\n",
    "        index = np.argmax(prediction[i][j])\n",
    "        output[i][j] = index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(13824, 8)"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_labels[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.,  0.,  0.,  1.,  0.,  0.,  0.,  0.], dtype=float32)"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_labels[250][6999]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_ground_truth = np.reshape(train_labels[28],(216,64,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for i in range(216):\n",
    "    for j in range(64):\n",
    "        index = np.argmax(test_ground_truth[i][j])\n",
    "        ground[i][j] = index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for i in range(216):\n",
    "    for j in range(64):\n",
    "        index = np.argmax(prediction[i][j])\n",
    "        output[i][j] = index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(216, 64)\n"
     ]
    }
   ],
   "source": [
    "print (output.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "color= np.zeros((216,64,3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "c0 = 0\n",
    "c1 = 0\n",
    "c2 = 0\n",
    "c3 = 0\n",
    "c4 = 0\n",
    "c5 = 0\n",
    "c6 = 0\n",
    "c7 = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for j in range(216):\n",
    "    for k in range(64):\n",
    "        if(output[j][k]==0):\n",
    "            c0 = c0 + 1\n",
    "            color[j][k] = [0,0,0]\n",
    "        if(output[j][k]==1):\n",
    "            c1 = c1 + 1\n",
    "            color[j][k] = [128,0,0]\n",
    "        if(output[j][k]==2):\n",
    "            c2 = c2 + 1\n",
    "            color[j][k] = [0,128,0]\n",
    "        if(output[j][k]==3):\n",
    "            c3 = c3 + 1\n",
    "            color[j][k] = [128,128,0] \n",
    "        if(output[j][k]==4):\n",
    "            c4 = c4 + 1\n",
    "            color[j][k] = [0,128,128]\n",
    "        if(output[j][k]==5):\n",
    "            c5 = c5 + 1\n",
    "            color[j][k] = [64,0,0]\n",
    "        if(output[j][k]==6):\n",
    "            c6 = c6 + 1\n",
    "            color[j][k] = [192,0,0]\n",
    "        if(output[j][k]==7):\n",
    "            c7 = c7 + 1\n",
    "            color[j][k] = [64,128,0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8088\n",
      "789\n",
      "993\n",
      "675\n",
      "816\n",
      "1438\n",
      "513\n",
      "512\n"
     ]
    }
   ],
   "source": [
    "print(c0)\n",
    "print(c1)\n",
    "print(c2)\n",
    "print(c3)\n",
    "print(c4)\n",
    "print(c5)\n",
    "print(c6)\n",
    "print(c7)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f5c7c736350>"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAGoAAAEACAYAAABbOeMfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRCnBRCnBRCnBRCnBRCnBRCnh/wGRo4ArNWVHfAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5c7c79a150>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(color)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "imsave(\"results2.png\",color)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "color= np.zeros((216,64,3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "c0 = 0\n",
    "c1 = 0\n",
    "c2 = 0\n",
    "c3 = 0\n",
    "c4 = 0\n",
    "c5 = 0\n",
    "c6 = 0\n",
    "c7 = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for j in range(216):\n",
    "    for k in range(64):\n",
    "        if(ground[j][k]==0):\n",
    "            c0 = c0 + 1\n",
    "            color[j][k] = [0,0,0]\n",
    "        if(ground[j][k]==1):\n",
    "            c1 = c1 + 1\n",
    "            color[j][k] = [128,0,0]\n",
    "        if(ground[j][k]==2):\n",
    "            c2 = c2 + 1\n",
    "            color[j][k] = [0,128,0]\n",
    "        if(ground[j][k]==3):\n",
    "            c3 = c3 + 1\n",
    "            color[j][k] = [128,128,0] \n",
    "        if(ground[j][k]==4):\n",
    "            c4 = c4 + 1\n",
    "            color[j][k] = [0,128,128]\n",
    "        if(ground[j][k]==5):\n",
    "            c5 = c5 + 1\n",
    "            color[j][k] = [64,0,0]\n",
    "        if(ground[j][k]==6):\n",
    "            c6 = c6 + 1\n",
    "            color[j][k] = [192,0,0]\n",
    "        if(ground[j][k]==7):\n",
    "            c7 = c7 + 1\n",
    "            color[j][k] = [64,128,0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8273\n",
      "592\n",
      "1035\n",
      "703\n",
      "763\n",
      "1501\n",
      "548\n",
      "409\n"
     ]
    }
   ],
   "source": [
    "print(c0)\n",
    "print(c1)\n",
    "print(c2)\n",
    "print(c3)\n",
    "print(c4)\n",
    "print(c5)\n",
    "print(c6)\n",
    "print(c7)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f5c7c66e190>"
      ]
     },
     "execution_count": 121,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAGoAAAEACAYAAABbOeMfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nF2UEF2UEF2UEF2UEF2UEF2WEPwGpL+KTqkyeBQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5c7c6d27d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(color)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 322,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "imsave(\"results.png\",color)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [Root]",
   "language": "python",
   "name": "Python [Root]"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}