clean code-1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import youtube_dl\n",
    "import moviepy\n",
    "import librosa\n",
    "import matplotlib.pyplot as plt\n",
    "import soundfile\n",
    "import sklearn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "pandas version = 0.25.1\n",
      "numpy version = 1.16.5\n",
      "tensorflow version = 2.0.0\n",
      "moviepy version = 1.0.1\n",
      "librosa version = 0.7.2\n"
     ]
    }
   ],
   "source": [
    "print('pandas version = {}'.format(pd.__version__))\n",
    "print('numpy version = {}'.format(np.__version__))\n",
    "print('tensorflow version = {}'.format(tf.__version__))\n",
    "print('moviepy version = {}'.format(moviepy.__version__))\n",
    "print('librosa version = {}'.format(librosa.__version__))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**This csv file contains youtube link for different types of videos containing different sounds.\n",
    "Useful fields are link-part, Duration of the audio, Corresponding label (Police/Ambulance/Fire Truck)**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Swaraj Badhei\\Anaconda3\\lib\\site-packages\\IPython\\core\\interactiveshell.py:3058: DtypeWarning: Columns (12,13,14,15,16,17) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  interactivity=interactivity, compiler=compiler, result=result)\n"
     ]
    }
   ],
   "source": [
    "data_file = pd.read_csv('unbalanced_train_segments.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No of instances = 1048572\n",
      "No of Columns = 18\n",
      "Columns are as follows-:\n",
      "Index(['---1_cCGK4M', '0', '10', ' \"/m/01g50p', '/m/0284vy3', '/m/06d_3',\n",
      "       '/m/07jdr', '/m/07rwm0c\"', 'Unnamed: 8', 'Unnamed: 9', 'Unnamed: 10',\n",
      "       'Unnamed: 11', 'Unnamed: 12', 'Unnamed: 13', 'Unnamed: 14',\n",
      "       'Unnamed: 15', 'Unnamed: 16', 'Unnamed: 17'],\n",
      "      dtype='object')\n"
     ]
    }
   ],
   "source": [
    "print('No of instances = {}\\nNo of Columns = {}'.format(data_file.shape[0],data_file.shape[1]))\n",
    "print('Columns are as follows-:\\n{}'.format(data_file.columns))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "targets=data_file.values[:,3:]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Next up, we need to separate the links for different labels(Emergency Vehicles in our case).#\n",
    "Useful data are as follows:\n",
    "#Link\n",
    "#Starting time of audio part\n",
    "#Ending time of audio part"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "#tags got from the documentaion for the audio dataset\n",
    "#Takes a bit of time..so not recommended to run each time.\n",
    "police='/m/04qvtq'\n",
    "ambu='/m/012n7d'\n",
    "fire='/m/012ndj'\n",
    "emer='/03j1ly'\n",
    "emer_lst=[]\n",
    "police_lst=[]\n",
    "ambu_lst=[]\n",
    "fire_lst=[]\n",
    "for i in range(targets.shape[0]):\n",
    "    if police in targets[i,:]:\n",
    "        police_lst.append([data_file.iloc[i,0],data_file.iloc[i,1],data_file.iloc[i,2]])\n",
    "    elif fire in targets[i,:]:\n",
    "        fire_lst.append([data_file.iloc[i,0],data_file.iloc[i,1],data_file.iloc[i,2]])\n",
    "    elif ambu in targets[i,:]:\n",
    "        ambu_lst.append([data_file.iloc[i,0],data_file.iloc[i,1],data_file.iloc[i,2]])\n",
    "    else:\n",
    "        pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No of Police Car instances = 242\n",
      "No of Fire Truck instances = 213\n",
      "No of Ambulanece instances = 30\n"
     ]
    }
   ],
   "source": [
    "print('No of Police Car instances = {}'.format(len(police_lst)))\n",
    "print('No of Fire Truck instances = {}'.format(len(fire_lst)))\n",
    "print('No of Ambulanece instances = {}'.format(len(ambu_lst)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Here comes up the downloading part\n",
    "#1.Download the audio file from youtube\n",
    "#2.Clip the audio file as per got starting and ending time\n",
    "#3.Delete the previous audio file downloaded\n",
    "#4.Save the clipped audio"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#for the record...dont fuck with this now...it just works\n",
    "'''for i in range(np.array(ambu_lst).shape[0]):\n",
    "    try:\n",
    "        ts=ambu_lst[i][1]\n",
    "        te=ambu_lst[i][2]\n",
    "        #print(te,ts)\n",
    "        y_id=ambu_lst[i][0]\n",
    "        video_url='https://www.youtube.com/watch?v={}'.format(y_id)\n",
    "        video=pafy.new(video_url)\n",
    "        best_audio=video.getbestaudio()\n",
    "        audio_name=best_audio.filename\n",
    "        best_audio.download()\n",
    "        if os.path.exists(audio_name):\n",
    "            output_file='Ambulance_'+str(i)+'.webm'\n",
    "            ffmpeg_extract_subclip(audio_name,ts,te,targetname=output_file)\n",
    "            os.remove(audio_name)\n",
    "        print(\"DONE FOR {}\".format(i+1))\n",
    "    except:\n",
    "        print('Error in file : '.format(i+1))'''"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### For Audio visualization part Go to \"audio_visualization_file.ipynb\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Preparation to store in .csv file please refer to \"final_code_for_feature_extraction\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### Deep Learning part\n",
    "#### 1. Reading data from previously prepared .csv file\n",
    "#### 2. Preparing one simple Dense net\n",
    "#### 3. Feeding and training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.models import Model\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from keras.optimizers import Adam,SGD\n",
    "import sklearn\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensorflow version = 2.0.0\n",
      "numpy version = 1.16.5\n",
      "pandas version = 0.25.1\n",
      "scikit-learn version = 0.21.3\n"
     ]
    }
   ],
   "source": [
    "print(\"tensorflow version = {}\".format(tf.__version__))\n",
    "print(\"numpy version = {}\".format(np.__version__))\n",
    "print(\"pandas version = {}\".format(pd.__version__))\n",
    "print(\"scikit-learn version = {}\".format(sklearn.__version__))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_file=pd.read_csv('feature/final_feature.csv').values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "target_file=pd.read_csv('feature/final_targets.csv').values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total no of data points = 301\n",
      "Total no of Features = 1\n"
     ]
    }
   ],
   "source": [
    "print(\"Total no of data points = {}\\nTotal no of Features = {}\".format(target_file.shape[0],target_file.shape[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of datapoint set = (301, 861, 40)\n"
     ]
    }
   ],
   "source": [
    "feature_file=feature_file.reshape(301,861,40)\n",
    "print('Shape of datapoint set = {}'.format(feature_file.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of Target Vector = (301, 3)\n"
     ]
    }
   ],
   "source": [
    "ohe=OneHotEncoder(categories='auto')\n",
    "onehot_target=ohe.fit_transform(target_file.reshape(-1,1)).todense()\n",
    "print('Shape of Target Vector = {}'.format(onehot_target.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(feature_file, onehot_target, test_size=0.15, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "model=tf.keras.Sequential([\n",
    "    tf.keras.layers.Flatten(input_shape=(861,40)),\n",
    "    tf.keras.layers.Dense(1024,activation=tf.nn.relu),\n",
    "    tf.keras.layers.Dense(512,activation=tf.nn.relu),\n",
    "    tf.keras.layers.Dense(512,activation=tf.nn.relu),\n",
    "    tf.keras.layers.Dense(512,activation=tf.nn.relu),\n",
    "    tf.keras.layers.Dense(256,activation=tf.nn.relu),\n",
    "    tf.keras.layers.Dense(256,activation=tf.nn.relu),\n",
    "    tf.keras.layers.Dense(128,activation=tf.nn.relu),\n",
    "    tf.keras.layers.Dense(64,activation=tf.nn.relu),\n",
    "    tf.keras.layers.Dense(3,activation=tf.nn.softmax)\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "flatten (Flatten)            (None, 34440)             0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 1024)              35267584  \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 512)               524800    \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 512)               262656    \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 512)               262656    \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 256)               65792     \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 3)                 195       \n",
      "=================================================================\n",
      "Total params: 36,556,163\n",
      "Trainable params: 36,556,163\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "adam=Adam(lr=0.001,decay=0.0001)\n",
    "sgd=SGD(lr=0.001,decay=0.0001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(optimizer='adam',loss='categorical_crossentropy',\n",
    "             metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 255 samples, validate on 46 samples\n",
      "Epoch 1/10\n",
      "255/255 [==============================] - 4s 15ms/sample - loss: 66.8572 - acc: 0.3490 - val_loss: 67.6783 - val_acc: 0.5870\n",
      "Epoch 2/10\n",
      "255/255 [==============================] - 2s 7ms/sample - loss: 34.2662 - acc: 0.5333 - val_loss: 22.1390 - val_acc: 0.5870\n",
      "Epoch 3/10\n",
      "255/255 [==============================] - 2s 7ms/sample - loss: 29.4684 - acc: 0.4275 - val_loss: 17.8237 - val_acc: 0.3261\n",
      "Epoch 4/10\n",
      "255/255 [==============================] - 2s 7ms/sample - loss: 17.0272 - acc: 0.4275 - val_loss: 8.9734 - val_acc: 0.5870\n",
      "Epoch 5/10\n",
      "255/255 [==============================] - 2s 7ms/sample - loss: 8.0507 - acc: 0.4588 - val_loss: 4.0002 - val_acc: 0.5870\n",
      "Epoch 6/10\n",
      "255/255 [==============================] - 2s 7ms/sample - loss: 2.0671 - acc: 0.5569 - val_loss: 1.4875 - val_acc: 0.4565\n",
      "Epoch 7/10\n",
      "255/255 [==============================] - 2s 7ms/sample - loss: 1.0554 - acc: 0.5804 - val_loss: 1.3858 - val_acc: 0.4348\n",
      "Epoch 8/10\n",
      "255/255 [==============================] - 2s 8ms/sample - loss: 0.7441 - acc: 0.6667 - val_loss: 1.3409 - val_acc: 0.4565\n",
      "Epoch 9/10\n",
      "255/255 [==============================] - 2s 8ms/sample - loss: 0.5120 - acc: 0.7686 - val_loss: 1.6760 - val_acc: 0.4130\n",
      "Epoch 10/10\n",
      "255/255 [==============================] - 2s 7ms/sample - loss: 0.5274 - acc: 0.7843 - val_loss: 1.3177 - val_acc: 0.5652\n"
     ]
    }
   ],
   "source": [
    "history=model.fit(X_train,y_train,epochs=10,validation_data=(X_test,y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['loss', 'acc', 'val_loss', 'val_acc'])"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history.history.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x1d780629d88>]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['acc'],history.history['loss'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}