Quick_Draw_Dataset.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Quick_Draw_Dataset.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyOS5x0I19AGwz38dV0Znq2x",
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Kushagratandon12/Machine-Learning-Projects/blob/master/Quick_Draw_Dataset.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PTJC6xPWOmK9"
      },
      "source": [
        "# # !gsutil -m cp 'gs://quickdraw_dataset/full/simplified/*.ndjson' . - Complete Dataset For Drawing Detection\n",
        "# !gsutil -m cp 'gs://quickdraw_dataset/full/numpy_bitmap/The Eiffel Tower.npy' .\n",
        "# !gsutil -m cp 'gs://quickdraw_dataset/full/simplified/The Eiffel Tower.ndjson' .\n",
        "# # !gsutil -m cp 'gs://quickdraw_dataset/full/simplified/animal migration.ndjson' ."
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "23GyLTnJOwaX",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b6dfb9dd-3817-4ea9-cc0f-e77cf9e9db4c"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "import cv2\n",
        "import random\n",
        "import json\n",
        "import gc\n",
        "import matplotlib.pyplot as plt\n",
        "import warnings\n",
        "!pip install cairocffi\n",
        "import cairocffi as cairo"
      ],
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: cairocffi in /usr/local/lib/python3.7/dist-packages (1.2.0)\n",
            "Requirement already satisfied: cffi>=1.1.0 in /usr/local/lib/python3.7/dist-packages (from cairocffi) (1.14.5)\n",
            "Requirement already satisfied: pycparser in /usr/local/lib/python3.7/dist-packages (from cffi>=1.1.0->cairocffi) (2.20)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "6LYKqx_HRK9Y",
        "outputId": "4dec09c6-a4ac-4b10-963f-517a94efa35b"
      },
      "source": [
        "data = map(json.loads, open('/content/The Eiffel Tower.ndjson'))\n",
        "df_data = pd.DataFrame(data)\n",
        "df_data = df_data [df_data['recognized']==True]\n",
        "df_data.head()"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>word</th>\n",
              "      <th>countrycode</th>\n",
              "      <th>timestamp</th>\n",
              "      <th>recognized</th>\n",
              "      <th>key_id</th>\n",
              "      <th>drawing</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>The Eiffel Tower</td>\n",
              "      <td>GB</td>\n",
              "      <td>2017-03-11 14:47:44.05242 UTC</td>\n",
              "      <td>True</td>\n",
              "      <td>5027286841556992</td>\n",
              "      <td>[[[0, 22, 37, 64, 255], [218, 220, 227, 228, 2...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>The Eiffel Tower</td>\n",
              "      <td>FR</td>\n",
              "      <td>2017-03-12 22:51:18.52595 UTC</td>\n",
              "      <td>True</td>\n",
              "      <td>5716269791707136</td>\n",
              "      <td>[[[47, 47, 36, 26, 0, 10, 23, 46, 46, 63, 68, ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>The Eiffel Tower</td>\n",
              "      <td>GB</td>\n",
              "      <td>2017-03-29 01:12:00.36798 UTC</td>\n",
              "      <td>True</td>\n",
              "      <td>5942899998982144</td>\n",
              "      <td>[[[184, 115, 67, 57, 36, 18], [251, 103, 12, 1...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>The Eiffel Tower</td>\n",
              "      <td>US</td>\n",
              "      <td>2017-03-29 16:48:54.9129 UTC</td>\n",
              "      <td>True</td>\n",
              "      <td>6226163091374080</td>\n",
              "      <td>[[[0, 187, 177, 132, 105, 79, 38, 19, 11], [24...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>The Eiffel Tower</td>\n",
              "      <td>GB</td>\n",
              "      <td>2017-03-04 15:50:45.19801 UTC</td>\n",
              "      <td>True</td>\n",
              "      <td>4889008825958400</td>\n",
              "      <td>[[[0, 21, 43, 83, 97, 158, 169, 172], [162, 16...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "               word  ...                                            drawing\n",
              "0  The Eiffel Tower  ...  [[[0, 22, 37, 64, 255], [218, 220, 227, 228, 2...\n",
              "1  The Eiffel Tower  ...  [[[47, 47, 36, 26, 0, 10, 23, 46, 46, 63, 68, ...\n",
              "2  The Eiffel Tower  ...  [[[184, 115, 67, 57, 36, 18], [251, 103, 12, 1...\n",
              "3  The Eiffel Tower  ...  [[[0, 187, 177, 132, 105, 79, 38, 19, 11], [24...\n",
              "4  The Eiffel Tower  ...  [[[0, 21, 43, 83, 97, 158, 169, 172], [162, 16...\n",
              "\n",
              "[5 rows x 6 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 35
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "y8spoWyaX7Zc",
        "outputId": "db269b1a-1920-4c53-fe65-401dbf307076"
      },
      "source": [
        "df_data.size"
      ],
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "782598"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 36
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        },
        "id": "HStsZRkMoXiJ",
        "outputId": "b8b8a277-b1ff-4139-e423-1dd1d5e60a23"
      },
      "source": [
        "vector_drawings = df_data['drawing']\n",
        "vector_label = df_data['word'].unique()\n",
        "img_inspect = vector_drawings[random.randint(1,len(vector_drawings))]\n",
        "# plt.grid(True)\n",
        "for x , y in img_inspect:\n",
        "  plt.plot(x, y, marker='.')\n",
        "  plt.title(vector_label)"
      ],
      "execution_count": 98,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uLzE2j6HJCDU",
        "outputId": "4772ec71-20d5-4ba4-fc4a-c917887644ed"
      },
      "source": [
        "gc.collect()\n",
        "df_data['drawing'] = df_data['drawing'].apply(json.dumps)\n",
        "df_data['drawing'] = df_data['drawing'].apply(json.loads)\n",
        "drawing=df_data['drawing'].values\n",
        "drawing"
      ],
      "execution_count": 99,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([list([[[0, 22, 37, 64, 255], [218, 220, 227, 228, 211]], [[76, 95, 135, 141, 150, 159, 166, 180, 186, 201], [220, 138, 31, 0, 63, 79, 117, 150, 191, 224]], [[94, 104, 111, 119, 127, 141, 143, 142, 180, 191], [212, 167, 149, 80, 59, 41, 30, 134, 202, 232]], [[109, 127, 137, 147, 150, 162, 172, 185], [122, 120, 104, 97, 99, 124, 128, 128]], [[75, 130, 158], [162, 159, 150]]]),\n",
              "       list([[[47, 47, 36, 26, 0, 10, 23, 46, 46, 63, 68, 105, 107, 112, 134, 152, 170, 151, 125, 101, 82, 74, 70, 60], [18, 98, 167, 201, 255, 248, 245, 248, 235, 183, 181, 187, 202, 216, 244, 252, 239, 180, 140, 93, 42, 1, 1, 21]], [[76, 68, 78, 101, 102, 93], [71, 131, 135, 136, 126, 101]]]),\n",
              "       list([[[184, 115, 67, 57, 36, 18], [251, 103, 12, 109, 193, 247]], [[145, 154, 150, 48, 24], [180, 176, 175, 178, 173]], [[55, 50, 38, 0, 28, 31, 60, 66, 71], [128, 152, 179, 253, 254, 247, 114, 26, 0]], [[72, 76, 81, 83, 86, 99, 93, 86], [26, 35, 71, 250, 255, 254, 189, 5]], [[67, 74, 97, 172], [12, 16, 49, 194]]]),\n",
              "       ...,\n",
              "       list([[[0, 13, 32, 44, 59, 104, 120, 124, 135, 151, 172, 195, 205, 209], [255, 249, 229, 212, 184, 84, 29, 0, 33, 119, 183, 237, 252, 253]], [[126, 112, 115, 137, 139, 132], [0, 17, 20, 21, 12, 5]], [[123, 119, 100, 134, 142, 132], [74, 97, 147, 158, 158, 84]], [[177, 169, 156, 130, 92, 89, 70, 22], [247, 212, 182, 175, 172, 173, 210, 246]]]),\n",
              "       list([[[104, 96, 77, 50, 0], [1, 66, 119, 169, 244]], [[109, 106, 107, 130, 151, 169], [0, 12, 41, 148, 216, 247]], [[1, 18, 44, 71], [247, 234, 227, 254]], [[74, 89, 106, 113, 130], [255, 240, 234, 237, 255]]]),\n",
              "       list([[[48, 39, 32, 19, 11, 0], [131, 144, 161, 199, 234, 255]], [[53, 80, 107], [135, 186, 253]], [[48, 48, 56, 51, 51, 44, 58, 49], [126, 92, 46, 22, 0, 13, 8, 3]], [[52, 25, 50, 87, 89, 62, 56], [43, 126, 119, 120, 117, 69, 43]]])],\n",
              "      dtype=object)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 99
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6BZW3VPHQqv6",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bbacc987-2d93-48e8-d731-d86b3f0d70d0"
      },
      "source": [
        "def vector_to_raster(vector_images, side=28, line_diameter=16, padding=16, bg_color=(0,0,0), fg_color=(1,1,1)):\n",
        "    \"\"\"\n",
        "    padding and line_diameter are relative to the original 256x256 image.\n",
        "    \"\"\"\n",
        "    \n",
        "    original_side = 256.\n",
        "    \n",
        "    surface = cairo.ImageSurface(cairo.FORMAT_ARGB32, side, side)\n",
        "    ctx = cairo.Context(surface)\n",
        "    ctx.set_antialias(cairo.ANTIALIAS_BEST)\n",
        "    ctx.set_line_cap(cairo.LINE_CAP_ROUND)\n",
        "    ctx.set_line_join(cairo.LINE_JOIN_ROUND)\n",
        "    ctx.set_line_width(line_diameter)\n",
        "\n",
        "    # scale to match the new size\n",
        "    # add padding at the edges for the line_diameter\n",
        "    # and add additional padding to account for antialiasing\n",
        "    total_padding = padding * 2. + line_diameter\n",
        "    total_padding = padding * 2. + line_diameter\n",
        "    new_scale = float(side) / float(original_side + total_padding)\n",
        "    ctx.scale(new_scale, new_scale)\n",
        "    ctx.translate(total_padding / 2., total_padding / 2.)\n",
        "\n",
        "    raster_images = []\n",
        "    for vector_image in vector_images:\n",
        "        # clear background\n",
        "        ctx.set_source_rgb(*bg_color)\n",
        "        ctx.paint()\n",
        "        \n",
        "        bbox = np.hstack(vector_image).max(axis=1)\n",
        "        offset = ((original_side, original_side) - bbox) / 2.\n",
        "        offset = offset.reshape(-1,1)\n",
        "        centered = [stroke + offset for stroke in vector_image]\n",
        "\n",
        "        # draw strokes, this is the most cpu-intensive part\n",
        "        ctx.set_source_rgb(*fg_color)        \n",
        "        for xv, yv in centered:\n",
        "            ctx.move_to(xv[0], yv[0])\n",
        "            for x, y in zip(xv, yv):\n",
        "                ctx.line_to(x, y)\n",
        "            ctx.stroke()\n",
        "\n",
        "        data = surface.get_data()\n",
        "        raster_image = np.copy(np.asarray(data)[::4])\n",
        "        raster_images.append(raster_image)\n",
        "    \n",
        "    return raster_images\n",
        "\n",
        "images = vector_to_raster(drawing)\n",
        "gc.collect()"
      ],
      "execution_count": 100,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 100
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "M0Mwc4ZJSxdv",
        "outputId": "c1e80268-6e89-43c7-837e-f043d6adebfa"
      },
      "source": [
        "imgflat = images[random.randint(1,len(images))]\n",
        "img = np.reshape(imgflat,(28,28,1))\n",
        "print(img.shape)"
      ],
      "execution_count": 110,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(28, 28, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 291
        },
        "id": "lrhRXMatcHTh",
        "outputId": "03013330-e011-4708-fa98-a355174180cb"
      },
      "source": [
        "from google.colab.patches import cv2_imshow\n",
        "plt.figure(figsize=(10,10))\n",
        "img = cv2.resize(img,(256,256))\n",
        "\n",
        "kernel = np.ones((5,5),np.float32)/25\n",
        "dst = cv2.filter2D(img,-1,kernel)\n",
        "\n",
        "cv2_imshow(dst)"
      ],
      "execution_count": 118,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<PIL.Image.Image image mode=L size=256x256 at 0x7FE3B15CFCD0>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x720 with 0 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "id": "GzcSqSsyV5eP",
        "outputId": "c73eb922-61b9-4c1b-af05-85e401314a94"
      },
      "source": [
        "plt.imshow(images[random.randint(1,len(images))].reshape(28,28))"
      ],
      "execution_count": 108,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fe3b1d264d0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 108
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}