EW_transferlr_VGG16.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "小提醒 : 下面Model為了比對建立了很多個，但實際上只有寫一個就可以了(避免重複訓練)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "paths現在型態是: <class 'list'>\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['EWtransfer_lr_data\\\\cat.0.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.100.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1000.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10000.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10001.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10002.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10003.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10004.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10005.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10006.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10007.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10008.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10009.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1001.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10010.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10011.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10012.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10013.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10014.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10015.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10016.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10017.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10018.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10019.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1002.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10020.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10021.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10022.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10023.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10024.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10025.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10026.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10027.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10028.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10029.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1003.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10030.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10031.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10032.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10033.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10034.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10035.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10036.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10037.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10038.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10039.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1004.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10040.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10041.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10042.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10043.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10044.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10045.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10046.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10047.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10048.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10049.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1005.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10050.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10051.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10052.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10053.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10054.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10055.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10056.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10057.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10058.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10059.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1006.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10060.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10061.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10062.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10063.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10064.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10065.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10066.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10067.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10068.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10069.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1007.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10070.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10071.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10072.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10073.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10074.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10075.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10076.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10077.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10078.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10079.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1008.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10080.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10081.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10082.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10083.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10084.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10085.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10086.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10087.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10088.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10089.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1009.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10090.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10091.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10092.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10093.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10094.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10095.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10096.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10097.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10098.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10099.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.101.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1010.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10100.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10101.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10102.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10103.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10104.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10105.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10106.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10107.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10108.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10109.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1011.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10110.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10111.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10112.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10113.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10114.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10115.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10116.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10117.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10118.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10119.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1012.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10120.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10121.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10122.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10123.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10124.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10125.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10126.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10127.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10128.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10129.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1013.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10130.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10131.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10132.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10133.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10134.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10135.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10136.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10137.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10138.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10139.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1014.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10140.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10141.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10142.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10143.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10144.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10145.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10146.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10147.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10148.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10149.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1015.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10150.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10151.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10152.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10153.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10154.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10155.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10156.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10157.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10158.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10159.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1016.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10160.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10161.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10162.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10163.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10164.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10165.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10166.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10167.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10168.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10169.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1017.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10170.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10171.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10172.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10173.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10174.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10175.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10176.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10177.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10178.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10179.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1018.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10180.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10181.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10182.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10183.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10184.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10185.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10186.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10187.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10188.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10189.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1019.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10190.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10191.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10192.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10193.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10194.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10195.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10196.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10197.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10198.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10199.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.102.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1020.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10200.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10201.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10202.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10203.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10204.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10205.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10206.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10207.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10208.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10209.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1021.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10210.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10211.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10212.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10213.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10214.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10215.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10216.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10217.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10218.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10219.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1022.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10220.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10221.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10222.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10223.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10224.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10225.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10226.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10227.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10228.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10229.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1023.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10230.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10231.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10232.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10233.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10234.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10235.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10236.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10237.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10238.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10239.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1024.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10240.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10241.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10242.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10243.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10244.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10245.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10246.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10247.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10248.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10249.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1025.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10250.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10251.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10252.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10253.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10254.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10255.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10256.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10257.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10258.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10259.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1026.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10260.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10261.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10262.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10263.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10264.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10265.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10266.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10267.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10268.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10269.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1027.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10270.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10271.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10272.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10273.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10274.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10275.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10276.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10277.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10278.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10279.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1028.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10280.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10281.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10282.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10283.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10284.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10285.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10286.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10287.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10288.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10289.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1029.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10290.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10291.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10292.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10293.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10294.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10295.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10296.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10297.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10298.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10299.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.103.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1030.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10300.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10301.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10302.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10303.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10304.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10305.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10306.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10307.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10308.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10309.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1031.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10310.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10311.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10312.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10313.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10314.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10315.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10316.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10317.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10318.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10319.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1032.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10320.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10321.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10322.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10323.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10324.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10325.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10326.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10327.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10328.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10329.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1033.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10330.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10331.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10332.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10333.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10334.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10335.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10336.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10337.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10338.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10339.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1034.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10340.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10341.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10342.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10343.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10344.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10345.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10346.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10347.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10348.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10349.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1035.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10350.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10351.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10352.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10353.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10354.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10355.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10356.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10357.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10358.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10359.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1036.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10360.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10361.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10362.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10363.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10364.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10365.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10366.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10367.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10368.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10369.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1037.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10370.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10371.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10372.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10373.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10374.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10375.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10376.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10377.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10378.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10379.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1038.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10380.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10381.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10382.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10383.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10384.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10385.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10386.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10387.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10388.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10389.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1039.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10390.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10391.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10392.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10393.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10394.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10395.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10396.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10397.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10398.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10399.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.104.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1040.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10400.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10401.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10402.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10403.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10404.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10405.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10406.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10407.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10408.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10409.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1041.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10410.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10411.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10412.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10413.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10414.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10415.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10416.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10417.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10418.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10419.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1042.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10420.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10421.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10422.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10423.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10424.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10425.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10426.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10427.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10428.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10429.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1043.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10430.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10431.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10432.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10433.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10434.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10435.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10436.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10437.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10438.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10439.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1044.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10440.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10441.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10442.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10443.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10444.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10445.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10446.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10447.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10448.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10449.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1045.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10450.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10451.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10452.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10453.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10454.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10455.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10456.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10457.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10458.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10459.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1046.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10460.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10461.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10462.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10463.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10464.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10465.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10466.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10467.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10468.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10469.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1047.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10470.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10471.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10472.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10473.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10474.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10475.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10476.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10477.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10478.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10479.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1048.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10480.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10481.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10482.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10483.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10484.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10485.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10486.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10487.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10488.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10489.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1049.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10490.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10491.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10492.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10493.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10494.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10495.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10496.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10497.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10498.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10499.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.105.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1050.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10500.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10501.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10502.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10503.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10504.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10505.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10506.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10507.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10508.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10509.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1051.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10510.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10511.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10512.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10513.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10514.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10515.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10516.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10517.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10518.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10519.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1052.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10520.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10521.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10522.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10523.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10524.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10525.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10526.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10527.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10528.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10529.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1053.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10530.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10531.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10532.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10533.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10534.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10535.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10536.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10537.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10538.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10539.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1054.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10540.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10541.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10542.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10543.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10544.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10545.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10546.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10547.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10548.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10549.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1055.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10550.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10551.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10552.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10553.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10554.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10555.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10556.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10557.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10558.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10559.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1056.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10560.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10561.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10562.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10563.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10564.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10565.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10566.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10567.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10568.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10569.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1057.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10570.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10571.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10572.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10573.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10574.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10575.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10576.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10577.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10578.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10579.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1058.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10580.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10581.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10582.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10583.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10584.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10585.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10586.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10587.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10588.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10589.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1059.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10590.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10591.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10592.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10593.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10594.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10595.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10596.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10597.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10598.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10599.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.106.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1060.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10600.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10601.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10602.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10603.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10604.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10605.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10606.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10607.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10608.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10609.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1061.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10610.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10611.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10612.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10613.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10614.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10615.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10616.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10617.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10618.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10619.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1062.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10620.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10621.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10622.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10623.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10624.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10625.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10626.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10627.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10628.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10629.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1063.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10630.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10631.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10632.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10633.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10634.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10635.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10636.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10637.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10638.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10639.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1064.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10640.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10641.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10642.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10643.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10644.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10645.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10646.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10647.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10648.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10649.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1065.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10650.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10651.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10652.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10653.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10654.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10655.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10656.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10657.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10658.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10659.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1066.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10660.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10661.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10662.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10663.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10664.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10665.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10666.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10667.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10668.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10669.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1067.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10670.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10671.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10672.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10673.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10674.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10675.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10676.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10677.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10678.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10679.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1068.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10680.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10681.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10682.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10683.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10684.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10685.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10686.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10687.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10688.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10689.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1069.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10690.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10691.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10692.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10693.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10694.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10695.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10696.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10697.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10698.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10699.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.107.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1070.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10700.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10701.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10702.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10703.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10704.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10705.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10706.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10707.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10708.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10709.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1071.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10710.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10711.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10712.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10713.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10714.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10715.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10716.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10717.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10718.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10719.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1072.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10720.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10721.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10722.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10723.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10724.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10725.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10726.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10727.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10728.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10729.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1073.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10730.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10731.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10732.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10733.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10734.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10735.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10736.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10737.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10738.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10739.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1074.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10740.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10741.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10742.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10743.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10744.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10745.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10746.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10747.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10748.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10749.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1075.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10750.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10751.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10752.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10753.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10754.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10755.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10756.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10757.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10758.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10759.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1076.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10760.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10761.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10762.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10763.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10764.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10765.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10766.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10767.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10768.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10769.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1077.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10770.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10771.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10772.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10773.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10774.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10775.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10776.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10777.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10778.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10779.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1078.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10780.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10781.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10782.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10783.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10784.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10785.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10786.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10787.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10788.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10789.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1079.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10790.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10791.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10792.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10793.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10794.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10795.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10796.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10797.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10798.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10799.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.108.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1080.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10800.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10801.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10802.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10803.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10804.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10805.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10806.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10807.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10808.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10809.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1081.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10810.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10811.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10812.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10813.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10814.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10815.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10816.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10817.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10818.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10819.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1082.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10820.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10821.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10822.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10823.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10824.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10825.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10826.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10827.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10828.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10829.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1083.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10830.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10831.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10832.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10833.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10834.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10835.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10836.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10837.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10838.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10839.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1084.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10840.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10841.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10842.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10843.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10844.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10845.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10846.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10847.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10848.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10849.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1085.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10850.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10851.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10852.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10853.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10854.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10855.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10856.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10857.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10858.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10859.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1086.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10860.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10861.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10862.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10863.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10864.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10865.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10866.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10867.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10868.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10869.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1087.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10870.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10871.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10872.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10873.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10874.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10875.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10876.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10877.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10878.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10879.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1088.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10880.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10881.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10882.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10883.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10884.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10885.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10886.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10887.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10888.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10889.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.1089.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10890.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10891.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10892.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10893.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10894.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10895.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10896.jpg',\n",
       " 'EWtransfer_lr_data\\\\cat.10897.jpg',\n",
       " ...]"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import glob\n",
    "import pandas as pd\n",
    "\n",
    "#先把貓圖片路徑存在變數paths中\n",
    "paths = glob.glob(\"EWtransfer_lr_data/cat.*.jpg\")\n",
    "print('paths現在型態是:', type(paths))\n",
    "paths #有一堆貓圖片的路徑"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>path</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>EWtransfer_lr_data\\cat.0.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>EWtransfer_lr_data\\cat.1.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>EWtransfer_lr_data\\cat.10.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>EWtransfer_lr_data\\cat.100.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>EWtransfer_lr_data\\cat.1000.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12495</th>\n",
       "      <td>EWtransfer_lr_data\\cat.9995.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12496</th>\n",
       "      <td>EWtransfer_lr_data\\cat.9996.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12497</th>\n",
       "      <td>EWtransfer_lr_data\\cat.9997.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12498</th>\n",
       "      <td>EWtransfer_lr_data\\cat.9998.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12499</th>\n",
       "      <td>EWtransfer_lr_data\\cat.9999.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>12500 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  path  target\n",
       "0         EWtransfer_lr_data\\cat.0.jpg       0\n",
       "1         EWtransfer_lr_data\\cat.1.jpg       0\n",
       "2        EWtransfer_lr_data\\cat.10.jpg       0\n",
       "3       EWtransfer_lr_data\\cat.100.jpg       0\n",
       "4      EWtransfer_lr_data\\cat.1000.jpg       0\n",
       "...                                ...     ...\n",
       "12495  EWtransfer_lr_data\\cat.9995.jpg       0\n",
       "12496  EWtransfer_lr_data\\cat.9996.jpg       0\n",
       "12497  EWtransfer_lr_data\\cat.9997.jpg       0\n",
       "12498  EWtransfer_lr_data\\cat.9998.jpg       0\n",
       "12499  EWtransfer_lr_data\\cat.9999.jpg       0\n",
       "\n",
       "[12500 rows x 2 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#設一個空 DataFrame等一下接剛抓下來的貓圖片路徑\n",
    "catdf = pd.DataFrame(columns=[\"path\", \"target\"])\n",
    "catdf[\"path\"] = paths\n",
    "catdf[\"target\"] = 0\n",
    "# 標記貓的圖片\n",
    "catdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>path</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>EWtransfer_lr_data\\dog.0.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>EWtransfer_lr_data\\dog.1.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>EWtransfer_lr_data\\dog.10.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>EWtransfer_lr_data\\dog.100.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>EWtransfer_lr_data\\dog.1000.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12495</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9995.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12496</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9996.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12497</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9997.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12498</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9998.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12499</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9999.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>12500 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  path  target\n",
       "0         EWtransfer_lr_data\\dog.0.jpg       1\n",
       "1         EWtransfer_lr_data\\dog.1.jpg       1\n",
       "2        EWtransfer_lr_data\\dog.10.jpg       1\n",
       "3       EWtransfer_lr_data\\dog.100.jpg       1\n",
       "4      EWtransfer_lr_data\\dog.1000.jpg       1\n",
       "...                                ...     ...\n",
       "12495  EWtransfer_lr_data\\dog.9995.jpg       1\n",
       "12496  EWtransfer_lr_data\\dog.9996.jpg       1\n",
       "12497  EWtransfer_lr_data\\dog.9997.jpg       1\n",
       "12498  EWtransfer_lr_data\\dog.9998.jpg       1\n",
       "12499  EWtransfer_lr_data\\dog.9999.jpg       1\n",
       "\n",
       "[12500 rows x 2 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 一樣的動作標記狗\n",
    "paths = glob.glob(\"EWtransfer_lr_data/dog.*.jpg\")\n",
    "dogdf = pd.DataFrame(columns=[\"path\", \"target\"])\n",
    "dogdf[\"path\"] = paths\n",
    "dogdf[\"target\"] = 1\n",
    "dogdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>path</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>EWtransfer_lr_data\\cat.0.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>EWtransfer_lr_data\\cat.1.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>EWtransfer_lr_data\\cat.10.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>EWtransfer_lr_data\\cat.100.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>EWtransfer_lr_data\\cat.1000.jpg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24995</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9995.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24996</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9996.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24997</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9997.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24998</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9998.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24999</th>\n",
       "      <td>EWtransfer_lr_data\\dog.9999.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>25000 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  path  target\n",
       "0         EWtransfer_lr_data\\cat.0.jpg       0\n",
       "1         EWtransfer_lr_data\\cat.1.jpg       0\n",
       "2        EWtransfer_lr_data\\cat.10.jpg       0\n",
       "3       EWtransfer_lr_data\\cat.100.jpg       0\n",
       "4      EWtransfer_lr_data\\cat.1000.jpg       0\n",
       "...                                ...     ...\n",
       "24995  EWtransfer_lr_data\\dog.9995.jpg       1\n",
       "24996  EWtransfer_lr_data\\dog.9996.jpg       1\n",
       "24997  EWtransfer_lr_data\\dog.9997.jpg       1\n",
       "24998  EWtransfer_lr_data\\dog.9998.jpg       1\n",
       "24999  EWtransfer_lr_data\\dog.9999.jpg       1\n",
       "\n",
       "[25000 rows x 2 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# pd.concat把DataFrame合體\n",
    "# 把要合體的DataFrame組成一個大list\n",
    "# ignore_index設定True，丟掉原本的列標籤生成新的列標籤索引\n",
    "traindf = pd.concat([catdf, dogdf], ignore_index=True)\n",
    "traindf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 來試著移植VGG16"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 224, 224, 3)       0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 224, 224, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 224, 224, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 112, 112, 64)      0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 112, 112, 128)     73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 112, 112, 128)     147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 56, 56, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 56, 56, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 28, 28, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 28, 28, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 14, 14, 512)       0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 7, 7, 512)         0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 25088)             0         \n",
      "=================================================================\n",
      "Total params: 14,714,688\n",
      "Trainable params: 14,714,688\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from keras.applications.vgg16 import VGG16\n",
    "from keras.models import Model #Model比較彈性\n",
    "#另一種from keras.models import Sequential的Sequential是一層一層疊\n",
    "\n",
    "from keras.layers import Dense, Flatten, Dropout\n",
    "from keras.layers import BatchNormalization\n",
    "                                          #weights = \"imagenet\"是預設 不寫也沒關係\n",
    "vgg = VGG16(include_top=False, weights = \"imagenet\", input_shape=(224, 224, 3))\n",
    "\n",
    "x = Flatten()(vgg.output) #這行語法看起來很怪 也是一個tensor 用一個變x數接攤平後vgg的output\n",
    "\n",
    "cnn = Model(inputs = vgg.input, outputs=x) # vgg的output之後(flatten之後)就是接全連階層\n",
    "cnn.summary() #看到最後攤平後25088有點大，希望等一下每層Dense都慢慢減少神經元"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "vgg.input #vgg.input 是一個tensor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "vgg.output #vgg.input 也是一個tensor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "vgg.layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_2 (InputLayer)         (None, 224, 224, 3)       0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 224, 224, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 224, 224, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 112, 112, 64)      0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 112, 112, 128)     73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 112, 112, 128)     147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 56, 56, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 56, 56, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 28, 28, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 28, 28, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 14, 14, 512)       0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 7, 7, 512)         0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 7, 7, 512)         2048      \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 25088)             0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 2048)              51382272  \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 256)               524544    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 2)                 514       \n",
      "=================================================================\n",
      "Total params: 66,624,066\n",
      "Trainable params: 66,623,042\n",
      "Non-trainable params: 1,024\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "vgg = VGG16(include_top=False, input_shape=(224, 224, 3))\n",
    "\n",
    "# 跟Sequential()慢慢疊的比起來 ，這個Model()寫法要熟悉一下\n",
    "# BatchNormalization這步很重要，進MLP之前批標準化(重新標準化)\n",
    "x = BatchNormalization()(vgg.output) #他是移植vgg後的第一個動作 所以要接(vgg.output)\n",
    "x = Flatten()(x)  #要進MLP了 所以攤平\n",
    "x = Dense(2048, activation=\"relu\")(x) #這幾行都是 某tensor丟給x，用x自己接 變成新的tensor \n",
    "x = Dropout(0.25)(x)\n",
    "x = Dense(256, activation=\"relu\")(x)\n",
    "x = Dropout(0.25)(x)\n",
    "x = Dense(units = 2, activation=\"softmax\")(x)  #設activation=\"softmax\" 目前是2選1 所以units = 2\n",
    "# 也可以寫activation=\"sigmoid\" , units = 1\n",
    "cnn = Model(inputs = vgg.input, outputs=x)\n",
    "cnn.summary()\n",
    "\n",
    "# 下一塊有訓練更快的方法"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_3\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_3 (InputLayer)         (None, 224, 224, 3)       0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 224, 224, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 224, 224, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 112, 112, 64)      0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 112, 112, 128)     73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 112, 112, 128)     147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 56, 56, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 56, 56, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 28, 28, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 28, 28, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 14, 14, 512)       0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 7, 7, 512)         0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 7, 7, 512)         2048      \n",
      "_________________________________________________________________\n",
      "flatten_3 (Flatten)          (None, 25088)             0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 2048)              51382272  \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 256)               524544    \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 2)                 514       \n",
      "=================================================================\n",
      "Total params: 66,624,066\n",
      "Trainable params: 51,908,354\n",
      "Non-trainable params: 14,715,712\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "vgg = VGG16(include_top=False, input_shape=(224, 224, 3))\n",
    "\n",
    "# 這步一定要在模型確定前，就是做(模型.compile)之前就寫 \n",
    "# 滑到最上面回憶一下 當初移植VGG的時候有設定include_top=False 沒有移植VGG的全連接層 只有移植VGG的CNN部分\n",
    "for l in vgg.layers:  #vgg.layers就是把每一層(vgg只有CNN部分)當成一個list\n",
    "    l.trainable = False # trainable這個屬性會決定這層要不要訓練，都固定False\n",
    "        \n",
    "# BatchNormalization這步很重要，進MLP之前批標準化(重新標準化)\n",
    "x = BatchNormalization()(vgg.output)  #他是移植vgg後的第一個動作 所以要接(vgg.output)\n",
    "x = Flatten()(x)   #要進MLP了 所以攤平\n",
    "x = Dense(2048, activation=\"relu\")(x)\n",
    "x = Dropout(0.25)(x)\n",
    "x = Dense(256, activation=\"relu\")(x)\n",
    "x = Dropout(0.25)(x)\n",
    "x = Dense(2, activation=\"softmax\")(x)\n",
    "cnn = Model(inputs=vgg.input, outputs=x)\n",
    "cnn.summary()\n",
    "#跟樓上比對一下 Non-trainable params ，減少了前面CNN卷積的訓練"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#確定模型怎麼訓練\n",
    "cnn.compile(loss=\"categorical_crossentropy\",\n",
    "            optimizer=\"adam\",\n",
    "            metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 把要傳進去的資料 題目和答案都轉成numpy array\n",
    "import numpy as np\n",
    "x = np.array(traindf[\"path\"]) # x 目前還只是路徑 \n",
    "y = np.array(traindf[\"target\"])\n",
    "\n",
    "# scikit-learn 要做資料切分\n",
    "from sklearn.model_selection import train_test_split #很常用\n",
    "\n",
    "# [90%x, 10%x, 90%y, 10%y] 記得那4個變數接 接完可以一個一個檢查一下\n",
    "x_train, x_test, y_train, y_test = train_test_split(x, \n",
    "                                                    y, \n",
    "                                                    test_size=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(22500,)\n",
      "(2500,)\n",
      "(22500,)\n",
      "(2500,)\n"
     ]
    }
   ],
   "source": [
    "print(x_train.shape)\n",
    "print(x_test.shape) # 此時的x_train，x_test還只是圖片的路徑\n",
    "print(y_train.shape)\n",
    "print(y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 答案 Y 記得轉成onehot-encoding\n",
    "from keras.utils.np_utils import to_categorical\n",
    "y_train_cat = to_categorical(y_train)\n",
    "y_test_cat = to_categorical(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<PIL.Image.Image image mode=RGB size=224x224 at 0x26A6BAD98C8>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# PIL: Pillow \n",
    "from keras.preprocessing.image import load_img   #會把路徑轉成圖片\n",
    "\n",
    "# 抓第一張圖來實驗\n",
    "img = load_img(x_train[0], target_size=(224, 224, 3))\n",
    "img #因為上面有做資料切分 所以每個人的第一張圖片應該都不一樣 跑出來的圖也不一樣"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[[135, 155, 106],\n",
       "        [129, 149,  96],\n",
       "        [139, 160, 104],\n",
       "        ...,\n",
       "        [126, 173,  79],\n",
       "        [122, 170,  70],\n",
       "        [129, 177,  75]],\n",
       "\n",
       "       [[129, 148, 102],\n",
       "        [123, 143,  94],\n",
       "        [119, 139,  86],\n",
       "        ...,\n",
       "        [ 93, 139,  49],\n",
       "        [ 86, 133,  39],\n",
       "        [103, 150,  54]],\n",
       "\n",
       "       [[110, 126,  89],\n",
       "        [117, 134,  92],\n",
       "        [120, 137,  92],\n",
       "        ...,\n",
       "        [ 92, 137,  54],\n",
       "        [ 81, 126,  43],\n",
       "        [ 96, 141,  56]],\n",
       "\n",
       "       ...,\n",
       "\n",
       "       [[122, 126, 127],\n",
       "        [122, 126, 127],\n",
       "        [131, 135, 136],\n",
       "        ...,\n",
       "        [106, 110, 121],\n",
       "        [104, 111, 121],\n",
       "        [102, 109, 119]],\n",
       "\n",
       "       [[132, 136, 137],\n",
       "        [127, 131, 132],\n",
       "        [126, 130, 131],\n",
       "        ...,\n",
       "        [106, 110, 121],\n",
       "        [106, 113, 123],\n",
       "        [104, 111, 121]],\n",
       "\n",
       "       [[128, 132, 133],\n",
       "        [122, 126, 127],\n",
       "        [141, 145, 146],\n",
       "        ...,\n",
       "        [106, 110, 121],\n",
       "        [107, 114, 124],\n",
       "        [105, 112, 122]]], dtype=uint8)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array(img) # 原本圖片的矩陣"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[[ 2.0609970e+00,  3.8221001e+01,  1.1320000e+01],\n",
       "        [-7.9390030e+00,  3.2221001e+01,  5.3199997e+00],\n",
       "        [ 6.0997009e-02,  4.3221001e+01,  1.5320000e+01],\n",
       "        ...,\n",
       "        [-2.4939003e+01,  5.6221001e+01,  2.3199997e+00],\n",
       "        [-3.3939003e+01,  5.3221001e+01, -1.6800003e+00],\n",
       "        [-2.8939003e+01,  6.0221001e+01,  5.3199997e+00]],\n",
       "\n",
       "       [[-1.9390030e+00,  3.1221001e+01,  5.3199997e+00],\n",
       "        [-9.9390030e+00,  2.6221001e+01, -6.8000031e-01],\n",
       "        [-1.7939003e+01,  2.2221001e+01, -4.6800003e+00],\n",
       "        ...,\n",
       "        [-5.4939003e+01,  2.2221001e+01, -3.0680000e+01],\n",
       "        [-6.4939003e+01,  1.6221001e+01, -3.7680000e+01],\n",
       "        [-4.9939003e+01,  3.3221001e+01, -2.0680000e+01]],\n",
       "\n",
       "       [[-1.4939003e+01,  9.2210007e+00, -1.3680000e+01],\n",
       "        [-1.1939003e+01,  1.7221001e+01, -6.6800003e+00],\n",
       "        [-1.1939003e+01,  2.0221001e+01, -3.6800003e+00],\n",
       "        ...,\n",
       "        [-4.9939003e+01,  2.0221001e+01, -3.1680000e+01],\n",
       "        [-6.0939003e+01,  9.2210007e+00, -4.2680000e+01],\n",
       "        [-4.7939003e+01,  2.4221001e+01, -2.7680000e+01]],\n",
       "\n",
       "       ...,\n",
       "\n",
       "       [[ 2.3060997e+01,  9.2210007e+00, -1.6800003e+00],\n",
       "        [ 2.3060997e+01,  9.2210007e+00, -1.6800003e+00],\n",
       "        [ 3.2060997e+01,  1.8221001e+01,  7.3199997e+00],\n",
       "        ...,\n",
       "        [ 1.7060997e+01, -6.7789993e+00, -1.7680000e+01],\n",
       "        [ 1.7060997e+01, -5.7789993e+00, -1.9680000e+01],\n",
       "        [ 1.5060997e+01, -7.7789993e+00, -2.1680000e+01]],\n",
       "\n",
       "       [[ 3.3060997e+01,  1.9221001e+01,  8.3199997e+00],\n",
       "        [ 2.8060997e+01,  1.4221001e+01,  3.3199997e+00],\n",
       "        [ 2.7060997e+01,  1.3221001e+01,  2.3199997e+00],\n",
       "        ...,\n",
       "        [ 1.7060997e+01, -6.7789993e+00, -1.7680000e+01],\n",
       "        [ 1.9060997e+01, -3.7789993e+00, -1.7680000e+01],\n",
       "        [ 1.7060997e+01, -5.7789993e+00, -1.9680000e+01]],\n",
       "\n",
       "       [[ 2.9060997e+01,  1.5221001e+01,  4.3199997e+00],\n",
       "        [ 2.3060997e+01,  9.2210007e+00, -1.6800003e+00],\n",
       "        [ 4.2060997e+01,  2.8221001e+01,  1.7320000e+01],\n",
       "        ...,\n",
       "        [ 1.7060997e+01, -6.7789993e+00, -1.7680000e+01],\n",
       "        [ 2.0060997e+01, -2.7789993e+00, -1.6680000e+01],\n",
       "        [ 1.8060997e+01, -4.7789993e+00, -1.8680000e+01]]], dtype=float32)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# VGG16的訓練方式 mode是caffe\n",
    "# caffe:先轉成BGR 沒做縮放 只有平移 (B-mean(B),G-mean(G),R-mean(R))數字通常會在 -120~120之間\n",
    "from keras.applications.vgg16 import preprocess_input \n",
    "#不管移植什麼模型都要跟著移植那個模型處理圖片的方式，不然權重會亂掉(很重要!!)\n",
    "\n",
    "preprocess_input(np.array(img))\n",
    "# preprocess_input處理後 和樓上比對"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 不能跟平常一樣 直接 .fit   ，因為圖片沒有整個讀取進來"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--------------- 第 1 輪 ---------------\n",
      "[1.3726157, 0.84]\n",
      "--------------- 第 2 輪 ---------------\n",
      "[0.9267696, 0.92]\n",
      "--------------- 第 3 輪 ---------------\n",
      "[1.3026068, 0.9]\n",
      "--------------- 第 4 輪 ---------------\n",
      "[1.8665092, 0.86]\n",
      "--------------- 第 5 輪 ---------------\n",
      "[0.63346267, 0.88]\n",
      "--------------- 第 6 輪 ---------------\n",
      "[0.5891408, 0.94]\n",
      "--------------- 第 7 輪 ---------------\n",
      "[0.1107411, 0.98]\n",
      "--------------- 第 8 輪 ---------------\n",
      "[0.29732585, 0.96]\n",
      "--------------- 第 9 輪 ---------------\n",
      "[1.4008483, 0.86]\n",
      "--------------- 第 10 輪 ---------------\n",
      "[2.053667, 0.86]\n"
     ]
    }
   ],
   "source": [
    "batch = 50    # 每看20筆資料做一次梯度下降(調整參數)\n",
    "train = 10\n",
    "\n",
    "# 下面的動作做10次\n",
    "for i in range(train):\n",
    "    img_list = [] # 叫一個空list等一下接處理後的圖片(矩陣表示)\n",
    "    \n",
    "    # 抽20個數字，也就是所有的x_train中 等一下會抽20張圖\n",
    "    idx = np.random.randint(0, x_train.shape[0], batch) #x_train.shape[0]是x_train有幾'筆'資料 \n",
    "    for fp in x_train[idx]: # idx是20個亂數，x_train[idx]是用這20個亂數一張一張打圖抓出來，fp是20個路徑\n",
    "        img = load_img(fp, target_size=(224, 224, 3)) #把路徑讀成圖片(矩陣表示) 順便設定大小\n",
    "        img_pre = preprocess_input(np.array(img))  # 經過preprocess_input處理後\n",
    "        img_list.append(img_pre)   # 再慢慢加到剛剛叫的空list，此時矩陣已經是處理後的樣子\n",
    "    x_train_np = np.array(img_list) # 把處理後的圖片矩陣(題目) 轉成numpy\n",
    "    #用這200個亂數當索引 直接拿到各自的答案(y_train_cat[idx]都是onehot-encoding) 也轉成numpy\n",
    "    y_train_np = np.array(y_train_cat[idx]) \n",
    "    \n",
    "    # XX.train_on_batch就是一個批次的訓練，注意batch不要設太大 會當機\n",
    "    result = cnn.train_on_batch(x_train_np, y_train_np)\n",
    "    print(\"-\" * 15,'第', i+1, \"輪\",\"-\" * 15) #左邊15個- (第幾次) 右邊也15個-\n",
    "    print(result)\n",
    "# (很上面的complie可以改設定)，報表的第一個是loss，第二個是準確度\n",
    "# 發現都表現得可能不太好的話 拉去最上面看有沒有加BatchNormalization()，另外在執行這塊一兩次也可以"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50, 224, 224, 3)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train_np.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50, 2)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_np.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 3s 17ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[15.99377155303955, 0.7850000262260437]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batch = 200    # 每看200筆資料做一次梯度下降(調整參數)\n",
    "test_img_list_ori = [] #先設一個空list等一下放圖原本的矩陣\n",
    "test_img_list = [] #先設一個空list等一下放圖處理後的矩陣\n",
    "\n",
    "\n",
    "# 抽200個數字，也就是所有的x_test中 等一下會抽200張圖\n",
    "idx = np.random.randint(0, x_test.shape[0], batch) #x_test.shape[0]是x_test有幾'筆'資料 \n",
    "\n",
    "for fp in x_test[idx]: # idx是200個亂數，x_test[idx]是用這200個亂數一張一張打圖抓出來\n",
    "    img = load_img(fp, target_size=(224, 224, 3)) #把路徑讀成圖片(矩陣表示) 順便設定大小\n",
    "    test_img_list_ori.append(np.array(img)) # 慢慢加到剛剛叫的空list，此時矩陣還是原本的樣子\n",
    "    img_pre = preprocess_input(np.array(img)) # 經過preprocess_input處理後\n",
    "    test_img_list.append(img_pre) # 再慢慢加到剛剛叫的空list，此時矩陣已經是處理後的樣子\n",
    "x_test_np = np.array(test_img_list)   # test題目\n",
    "y_test_np = np.array(y_test_cat[idx]) # test真正的答案(onehot-encoding)\n",
    "y_test_ori = y_test[idx] # test真正的答案(沒有做onehot-encoding)\n",
    "\n",
    "cnn.evaluate(x_test_np, y_test_np) \n",
    "# (很上面的complie可以改設定)，報表的第一個是loss，第二個是準確度"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 4.45671467e-14],\n",
       "       [1.53504447e-13, 1.00000000e+00],\n",
       "       [1.00000000e+00, 4.64583427e-18],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [7.13665358e-23, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.43180159e-13],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [9.96151865e-01, 3.84816783e-03],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [6.65578161e-13, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [9.99999166e-01, 8.48826517e-07],\n",
       "       [9.97120500e-01, 2.87954137e-03],\n",
       "       [1.00000000e+00, 1.34420439e-24],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 2.33640058e-08],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [1.00000000e+00, 1.01374919e-32],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [2.28751943e-01, 7.71248102e-01],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 8.73433894e-24],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 4.74489177e-21],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 6.42507714e-11],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 7.27857591e-31],\n",
       "       [1.04306722e-34, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 8.91566404e-36],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [4.45385424e-24, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.04248411e-31],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.04782175e-25],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.77540177e-10],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [9.59723234e-01, 4.02766839e-02],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [4.48006483e-15, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 3.10823698e-12],\n",
       "       [2.21338410e-20, 1.00000000e+00],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [1.00000000e+00, 3.65095950e-16],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [2.21338410e-20, 1.00000000e+00],\n",
       "       [1.00000000e+00, 8.80738640e-37],\n",
       "       [1.00000000e+00, 1.33749164e-12],\n",
       "       [4.06619061e-29, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.54171812e-34],\n",
       "       [1.00000000e+00, 3.60085914e-25],\n",
       "       [1.00000000e+00, 3.98910022e-19],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [2.96885915e-13, 1.00000000e+00],\n",
       "       [3.79760107e-15, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.46497802e-29],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 3.11537375e-29],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 5.40480204e-12],\n",
       "       [2.94407897e-37, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [9.97644603e-01, 2.35532038e-03],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 8.32546654e-09],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.46497802e-29],\n",
       "       [3.40813422e-09, 1.00000000e+00],\n",
       "       [1.00000000e+00, 2.24191299e-18],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [1.00000000e+00, 1.41096204e-24],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [2.28751943e-01, 7.71248102e-01],\n",
       "       [1.00000000e+00, 3.08512190e-29],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [1.00000000e+00, 1.53944756e-29],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [5.83637139e-10, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.22849504e-18],\n",
       "       [1.00000000e+00, 1.77540177e-10],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 3.44155977e-28],\n",
       "       [5.39066591e-16, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [1.00000000e+00, 1.54171812e-34],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 6.68692676e-22],\n",
       "       [1.00000000e+00, 6.11810385e-38],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.27863697e-18],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.52972682e-18, 1.00000000e+00],\n",
       "       [3.84812310e-34, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 7.83231038e-16],\n",
       "       [1.00000000e+00, 9.49142547e-23],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.84788770e-32],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.41474896e-12, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 3.51029325e-14],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [9.99923468e-01, 7.65723817e-05],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [3.72078534e-14, 1.00000000e+00],\n",
       "       [1.00000000e+00, 1.34420439e-24],\n",
       "       [4.17222002e-14, 1.00000000e+00],\n",
       "       [4.35459879e-06, 9.99995589e-01],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [3.18746643e-05, 9.99968171e-01],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 4.40460423e-23],\n",
       "       [9.77367206e-25, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [2.23444845e-35, 1.00000000e+00],\n",
       "       [1.21990270e-05, 9.99987841e-01],\n",
       "       [6.83730781e-01, 3.16269249e-01],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.00000000e+00, 0.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.85382692e-38, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [1.24220794e-26, 1.00000000e+00],\n",
       "       [0.00000000e+00, 1.00000000e+00],\n",
       "       [2.21362046e-20, 1.00000000e+00],\n",
       "       [1.00000000e+00, 8.23503264e-32],\n",
       "       [0.00000000e+00, 1.00000000e+00]], dtype=float32)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Model式的模型只能用 XX.predict 跑出每個類別的機率\n",
    "# 反觀Sequential是的模型還可以用XX.predict_classes可以直接跑出機率最高的類別(就是判斷的答案)\n",
    "pre = cnn.predict(x_test_np) \n",
    "pre #他有沒有亂猜 像是全部猜一個答案"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(200, 2)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_test.shape.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 檢討考卷"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 最好都先轉成numpy array\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "# 前行提要:Model式的模型只能用 XX.predict，但不能用XX.predict_classes 才這麼麻煩\n",
    "\n",
    "# XX.argmax會找最大值的索引，axis=0 這找每行的最大值，axis=-1是找每列的最大值\n",
    "pre_ori = pre.argmax(axis=-1) #可以分別印出pre和pre_ori比對看看\n",
    "# 也就是你寫的答案\n",
    "\n",
    "\n",
    "# 這邊開始都是複製mnist\n",
    "# test_img_list_ori是一個大list，收集了200張圖(矩陣還是原本的樣子)，再轉numpy。 就是原本題目的樣子\n",
    "test_img_list_ori = np.array(test_img_list_ori)\n",
    "y_test_ori = np.array(y_test_ori) #y_test_ori是真正的答案(沒有做onehot-encoding)\n",
    "\n",
    "# pre_ori是你寫的答案拿去和真正的答案比對，把錯的題號存成變數false_index\n",
    "false_index = np.nonzero(pre_ori != y_test_ori)[0]\n",
    "print(\"到底有幾個預測錯誤:\", len(false_index))\n",
    "false_img = test_img_list_ori[false_index] # 用預測錯誤的索引 抓出你寫錯的題目\n",
    "ori_label = y_test_ori[false_index]        # 用預測錯誤的索引 抓出你寫錯題目的答案\n",
    "pre_label = pre_ori[false_index] #回憶:pre_ori是預測的答案(你寫的答案) \n",
    "#用預測錯誤的索引 抓出錯的那幾題你寫了什麼答案\n",
    "\n",
    "width = 10\n",
    "height = int(len(false_index) / 10) + 1\n",
    "plt.figure(figsize=(14, 7))\n",
    "for (index, img) in enumerate(false_img):\n",
    "    plt.subplot(height, width, index + 1)\n",
    "    msg = \"[O]:\" + str(ori_label[index]) + \"[P]:\" + str(pre_label[index])\n",
    "    plt.title(msg)\n",
    "    plt.axis(\"off\")\n",
    "    plt.imshow(img)"
   ]
  }
 ],
 "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": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}