01-pipe_1-car_check.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pipe1 : \n",
    "## To check if user input image is a car or not. Tested against multiple cnn frameworks to find one with maximum accuracy.\n",
    "\n",
    "### Data0 - dataset of all images of cars(damaged and undamaged ones)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import h5py\n",
    "import urllib.request\n",
    "import json\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import pickle as pk\n",
    "from IPython.display import Image, display, clear_output\n",
    "from collections import Counter, defaultdict\n",
    "from sklearn.metrics import classification_report, confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras import optimizers\n",
    "from keras.applications.vgg16 import VGG16\n",
    "from keras.applications.vgg19 import VGG19\n",
    "from keras.applications.resnet50 import ResNet50\n",
    "from keras.applications.inception_v3 import InceptionV3\n",
    "from keras.applications.xception import Xception\n",
    "from keras.applications.imagenet_utils import preprocess_input, decode_predictions\n",
    "from keras.utils.data_utils import get_file\n",
    "from keras.utils.np_utils import to_categorical\n",
    "from keras.preprocessing.image import ImageDataGenerator, array_to_img, img_to_array, load_img\n",
    "from keras.models import Sequential, load_model\n",
    "from keras.layers import Convolution2D, MaxPooling2D, ZeroPadding2D, Activation, Dropout, Flatten, Dense\n",
    "from keras.callbacks import ModelCheckpoint, History"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "CLASS_INDEX = None\n",
    "CLASS_INDEX_PATH = 'https://s3.amazonaws.com/deep-learning-models/image-models/imagenet_class_index.json'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Anaconda3\\envs\\envdlcv\\lib\\site-packages\\tensorflow\\python\\framework\\op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n"
     ]
    }
   ],
   "source": [
    "vgg16 = VGG16(weights='imagenet')\n",
    "vgg19 = VGG19(weights='imagenet')\n",
    "resnet = ResNet50(weights='imagenet')\n",
    "inception = InceptionV3(weights='imagenet')\n",
    "xception = Xception(weights='imagenet')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def prepare_image_224(img_path):\n",
    "    img = load_img(img_path, target_size=(224,224))\n",
    "    x = img_to_array(img)\n",
    "    x = np.expand_dims(x, axis=0)\n",
    "    x = preprocess_input(x)\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def prepare_image_299(img_path):\n",
    "    img = load_img(img_path, target_size=(299,299))\n",
    "    x = img_to_array(img)\n",
    "    x = np.expand_dims(x, axis=0)\n",
    "    x = preprocess_input(x)\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_predictions(preds, top=5):\n",
    "    global CLASS_INDEX\n",
    "    if len(preds.shape) != 2 or preds.shape[1] != 1000:\n",
    "        raise ValueError('`decode_predictions` expects a batch of predictions (i.e. a 2D array of shape (samples, 1000)). Found array with shape: ' + str(preds.shape))\n",
    "    if CLASS_INDEX is None:\n",
    "        fpath = get_file('imagenet_class_index.json',CLASS_INDEX_PATH,cache_subdir='models')\n",
    "        CLASS_INDEX = json.load(open(fpath))\n",
    "    results = []\n",
    "    for pred in preds:\n",
    "        top_indices = pred.argsort()[-top:][::-1]\n",
    "        result = [tuple(CLASS_INDEX[str(i)]) + (pred[i],) for i in top_indices]\n",
    "        result.sort(key=lambda x: x[2], reverse=True)\n",
    "        results.append(result)\n",
    "    return results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image('testimage.jpg')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Testing with different models "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[('n03100240', 'convertible', 0.28325948), ('n03930630', 'pickup', 0.17927827), ('n03670208', 'limousine', 0.16034348), ('n04285008', 'sports_car', 0.102715686), ('n03770679', 'minivan', 0.07280146)]]\n"
     ]
    }
   ],
   "source": [
    "y = prepare_image_224('testimage.jpg')\n",
    "preds = vgg16.predict(y)\n",
    "print(get_predictions(preds, top=5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[('n03100240', 'convertible', 0.26360297), ('n03930630', 'pickup', 0.14805934), ('n03770679', 'minivan', 0.14723605), ('n04285008', 'sports_car', 0.12050969), ('n02814533', 'beach_wagon', 0.11459463)]]\n"
     ]
    }
   ],
   "source": [
    "z = prepare_image_224('testimage.jpg')\n",
    "preds = vgg19.predict(z)\n",
    "print(get_predictions(preds, top=5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[('n03770679', 'minivan', 0.23549458), ('n02814533', 'beach_wagon', 0.13592634), ('n04285008', 'sports_car', 0.11415414), ('n03930630', 'pickup', 0.10745103), ('n02974003', 'car_wheel', 0.081083775)]]\n"
     ]
    }
   ],
   "source": [
    "a = prepare_image_224('testimage.jpg')\n",
    "preds = resnet.predict(a)\n",
    "print(get_predictions(preds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[('n01924916', 'flatworm', 1.0), ('n03047690', 'clog', 8.942e-12), ('n03065424', 'coil', 1.4942996e-14), ('n04325704', 'stole', 1.5525325e-16), ('n02089973', 'English_foxhound', 6.774755e-17)]]\n"
     ]
    }
   ],
   "source": [
    "b = prepare_image_299('testimage.jpg')\n",
    "preds = inception.predict(b)\n",
    "print(get_predictions(preds, top=5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[('n03763968', 'military_uniform', 0.36981878), ('n03942813', 'ping-pong_ball', 0.36295855), ('n03825788', 'nipple', 0.09450442), ('n02099712', 'Labrador_retriever', 0.046820126), ('n02769748', 'backpack', 0.046718605)]]\n"
     ]
    }
   ],
   "source": [
    "c = prepare_image_299('testimage.jpg')\n",
    "preds = xception.predict(c)\n",
    "print(get_predictions(preds, top=5))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Result: VGG16 shows best accuracy."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Using VGG16 model for Pipe1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_car_categories():\n",
    "    d = defaultdict(float)\n",
    "    img_list = os.listdir('data1')\n",
    "    for i, img_path in enumerate(img_list):\n",
    "        img = prepare_image_224('data1/'+img_path)\n",
    "        out = vgg16.predict(img)\n",
    "        preds = get_predictions(out,top=5)\n",
    "        for pred in preds[0]:\n",
    "            d[pred[0:2]]+=pred[2]\n",
    "        if(i%50==0):\n",
    "            print(i,'/',len(img_list),'complete')\n",
    "    return Counter(d)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 / 918 complete\n",
      "50 / 918 complete\n",
      "100 / 918 complete\n",
      "150 / 918 complete\n",
      "200 / 918 complete\n",
      "250 / 918 complete\n",
      "300 / 918 complete\n",
      "350 / 918 complete\n",
      "400 / 918 complete\n",
      "450 / 918 complete\n",
      "500 / 918 complete\n",
      "550 / 918 complete\n",
      "600 / 918 complete\n",
      "650 / 918 complete\n",
      "700 / 918 complete\n",
      "750 / 918 complete\n",
      "800 / 918 complete\n",
      "850 / 918 complete\n",
      "900 / 918 complete\n"
     ]
    }
   ],
   "source": [
    "categ_count = get_car_categories()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('vgg16_cat_list.pk', 'wb') as f:\n",
    "    pk.dump(categ_count, f, -1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('vgg16_cat_list.pk', 'rb') as f:\n",
    "    categ_count = pk.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "categ_list = [k for k, v in categ_count.most_common()[:50]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imagenet Car Categories"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_car_categories(categ_list):\n",
    "    img_list = os.listdir('data1')\n",
    "    n = 0\n",
    "    bad_list = []\n",
    "    for i, img_path in enumerate(img_list):\n",
    "        img = prepare_image_224('data1/'+img_path)\n",
    "        out = vgg16.predict(img)\n",
    "        preds = get_predictions(out, top=5)\n",
    "        for pred in preds[0]:\n",
    "            if pred[0:2] in categ_list:\n",
    "                n+=1\n",
    "                break\n",
    "            else:\n",
    "                pass\n",
    "            bad_list.append(img_path)\n",
    "        if i%100 == 0:\n",
    "            print(i,'/',len(img_list), 'complete')\n",
    "    bad_list = [k for k, v in Counter(bad_list).items() if v == 5]\n",
    "    return n, bad_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 / 918 complete\n",
      "100 / 918 complete\n",
      "200 / 918 complete\n",
      "300 / 918 complete\n",
      "400 / 918 complete\n",
      "500 / 918 complete\n",
      "600 / 918 complete\n",
      "700 / 918 complete\n",
      "800 / 918 complete\n",
      "900 / 918 complete\n"
     ]
    }
   ],
   "source": [
    "num, bad_list = get_car_categories(categ_list)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pipe 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pipe1(img_path, categ_list):\n",
    "    urllib.request.urlretrieve(img_path, 'image.jpg')\n",
    "    img = prepare_image_224('image.jpg')\n",
    "    out = vgg16.predict(img)\n",
    "    preds = get_predictions(out, top=5)\n",
    "    print(\"Ensuring entered picture is a car...\")\n",
    "    for pred in preds[0]:\n",
    "        if pred[0:2] in categ_list:\n",
    "            print(pred[0:2])\n",
    "            return \"Successful. Proceeding to damage assessment...\"\n",
    "    return \"The entered image is a not a car. Please try again. Consider a different angle or lighting.\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ensuring entered picture is a car...\n",
      "('n03770679', 'minivan')\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'Successful. Proceeding to damage assessment...'"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pipe1('https://www.autoauctionmall.com/learning-center/contents/uploads/2015/10/damaged-car.jpg', categ_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ensuring entered picture is a car...\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'The entered image is a not a car. Please try again. Consider a different angle or lighting.'"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pipe1('https://cdn.nba.net/nba-drupal-prod/2017-05/lonzo-ball-sideline-iso.jpg',categ_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ensuring entered picture is a car...\n",
      "('n04037443', 'racer')\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'Successful. Proceeding to damage assessment...'"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pipe1('https://tse4.mm.bing.net/th?id=OIP.FaZela57De0uzfVxVY3JJQHaEo&pid=Api&P=0&w=289&h=181', categ_list)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:envdlcv] *",
   "language": "python",
   "name": "conda-env-envdlcv-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}