Pixel_Brightness_And_Algorithmic_Bias.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "The Difference In Average Pixel Brightness For Black Faces May Account For Algorithmic Bias in Computer Vision Models",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "l4yEn-8D61ZC",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b3a10d41-d36a-4ed4-9f37-5387e15224d5"
      },
      "source": [
        "!pip install --upgrade pillow"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: pillow in /usr/local/lib/python3.7/dist-packages (8.3.2)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "a062ofLVnFHj"
      },
      "source": [
        "#Restart after running the above cell."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uJukDg8hCIKY",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "189b6d1c-f29b-4a39-a5f0-9bf71512828b"
      },
      "source": [
        "!gdown https://drive.google.com/uc?id=0BxYys69jI14kYVM3aVhKS1VhRUk"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading...\n",
            "From: https://drive.google.com/uc?id=0BxYys69jI14kYVM3aVhKS1VhRUk\n",
            "To: /content/UTKFace.tar.gz\n",
            "107MB [00:01, 100MB/s]  \n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "X9-L3LfqCIpU"
      },
      "source": [
        "!tar -xzvf UTKFace.tar.gz"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j7Yr1GEf6EPF"
      },
      "source": [
        "#There are too many images of Caucasian people in this dataset.  We need to reduce the number so the training will not be biased.\n",
        "import os\n",
        "import random\n",
        "directory = \"/content/UTKFace/\"\n",
        "\n",
        "white_images = []\n",
        "removed = 0\n",
        "\n",
        "for file in os.listdir(directory):\n",
        "\n",
        "    if str(file)[-32] == '0':\n",
        "      white_images.append(file)\n",
        "\n",
        "print(white_images)\n",
        "\n",
        "for i in range(10076):\n",
        "  try:\n",
        "    remove_idx = random.randint(0,len(white_images))\n",
        "    print(remove_idx)\n",
        "    img_to_remove = white_images[remove_idx]\n",
        "    print(img_to_remove)\n",
        "    os.remove(directory+img_to_remove)\n",
        "    removed+=1\n",
        "    print(\"removed: \",removed)\n",
        "  except:\n",
        "    pass"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hgfaB9RR4fzC"
      },
      "source": [
        "#This code converts all images to greyscale for averaging.\n",
        "import os\n",
        "import cv2\n",
        "from google.colab.patches import cv2_imshow\n",
        "\n",
        "directory = '/content/UTKFace/'\n",
        "\n",
        "for file in os.listdir(directory):\n",
        "  print(file)\n",
        "\n",
        "  image = cv2.imread(\"/content/UTKFace/\" + file)\n",
        "  image_grey = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)\n",
        "  cv2.imwrite(\"/content/UTKFace/\" + file, image_grey)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h41iiC7Xkrna"
      },
      "source": [
        "#@title Do you want to increase brightness of photos of persons of African descent by 26% ?\n",
        "\n",
        "\n",
        "increase_brightness = \"No\"  #@param ['Yes', 'No']\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "q7tcZAjUyIgi"
      },
      "source": [
        "# This will increase the brightness of all images of people of African descent, by 26%\n",
        "# Don't run this cell if you want to see how the original dataset looks.\n",
        "from PIL import Image, ImageEnhance\n",
        "import os\n",
        "import cv2\n",
        "from google.colab.patches import cv2_imshow\n",
        "\n",
        "directory = '/content/UTKFace/'\n",
        "\n",
        "for file in os.listdir(directory):\n",
        "  \n",
        "  if str(file)[-32] == '1' and increase_brightness=='Yes':\n",
        "\n",
        "    print(file)\n",
        "  \n",
        "    im = Image.open(r\"/content/UTKFace/\" + file)\n",
        "\n",
        "    im3 = ImageEnhance.Brightness(im)\n",
        "    im3.enhance(1.26).save(\"/content/UTKFace/\" + file)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zO2klXbdxjVZ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "204f3f72-8897-42ba-b1b2-a4001591a85a"
      },
      "source": [
        "import os\n",
        "from skimage import io\n",
        "import numpy as np\n",
        "\n",
        "directory = \"/content/UTKFace/\"\n",
        "\n",
        "white_mean_tot = 0\n",
        "black_mean_tot = 0\n",
        "asian_mean_tot = 0\n",
        "indian_mean_tot = 0\n",
        "other_mean_tot = 0\n",
        "\n",
        "white_img_count = 0\n",
        "black_img_count = 0\n",
        "asian_img_count = 0\n",
        "indian_img_count = 0\n",
        "other_img_count = 0\n",
        "\n",
        "white_mean_list = []\n",
        "black_mean_list = []\n",
        "asian_mean_list = []\n",
        "indian_mean_list = []\n",
        "other_mean_list = []\n",
        "\n",
        "for file in os.listdir(directory):\n",
        "    image = io.imread(\"/content/UTKFace/\" + file)\n",
        "    image_mean = np.mean(image)\n",
        "\n",
        "    if str(file)[-32] == '0':\n",
        "      \n",
        "      white_mean_tot = white_mean_tot + image_mean\n",
        "      white_img_count = white_img_count + 1\n",
        "      white_mean_list.append(int(image_mean))\n",
        "    \n",
        "    if str(file)[-32] == '1':\n",
        "      black_mean_tot = black_mean_tot + image_mean\n",
        "      black_img_count = black_img_count + 1\n",
        "      black_mean_list.append(int(image_mean))\n",
        "\n",
        "    if str(file)[-32] == '2':\n",
        "      asian_mean_tot = asian_mean_tot + image_mean\n",
        "      asian_img_count = asian_img_count + 1\n",
        "      asian_mean_list.append(int(image_mean))\n",
        "\n",
        "    if str(file)[-32] == '3':\n",
        "      indian_mean_tot = indian_mean_tot + image_mean\n",
        "      indian_img_count = indian_img_count +1\n",
        "      indian_mean_list.append(int(image_mean))\n",
        "    \n",
        "    if str(file)[-32] == '4':\n",
        "      other_mean_tot = other_mean_tot + image_mean\n",
        "      other_img_count = other_img_count + 1\n",
        "      other_mean_list.append(int(image_mean))\n",
        "\n",
        "print(\"white_mean_tot\")\n",
        "print(white_mean_tot)\n",
        "print(\"black_mean_tot\")\n",
        "print(black_mean_tot)\n",
        "print(\"asian_mean_tot\")\n",
        "print(asian_mean_tot)\n",
        "print(\"indian_mean_tot\")\n",
        "print(indian_mean_tot)\n",
        "print(\"other_mean_tot\")\n",
        "print(other_mean_tot)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "white_mean_tot\n",
            "486114.2318500001\n",
            "black_mean_tot\n",
            "470636.58182500023\n",
            "asian_mean_tot\n",
            "464647.1501250005\n",
            "indian_mean_tot\n",
            "489218.6619499994\n",
            "other_mean_tot\n",
            "208556.58739999967\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BeQpdx337XIX",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d01c09fa-faac-4a2c-ad37-d41b8707d78a"
      },
      "source": [
        "import pandas as pd\n",
        "\n",
        "df_list=[]\n",
        "\n",
        "for i in white_mean_list:\n",
        "  df_list.append([i,'white'])\n",
        "\n",
        "for i in black_mean_list:\n",
        "  df_list.append([i,'black'])\n",
        "\n",
        "for i in asian_mean_list:\n",
        "  df_list.append([i,'asian'])\n",
        "\n",
        "for i in indian_mean_list:\n",
        "  df_list.append([i,'indian'])\n",
        "\n",
        "for i in other_mean_list:\n",
        "  df_list.append([i,'other'])\n",
        "\n",
        "df = pd.DataFrame(df_list, columns=['px_vals','Color'])\n",
        "\n",
        "print(df)\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "       px_vals  Color\n",
            "0          109  white\n",
            "1          111  white\n",
            "2          125  white\n",
            "3           93  white\n",
            "4          147  white\n",
            "...        ...    ...\n",
            "17291      103  other\n",
            "17292      148  other\n",
            "17293       97  other\n",
            "17294       87  other\n",
            "17295      144  other\n",
            "\n",
            "[17296 rows x 2 columns]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B1ERIFy16vvV"
      },
      "source": [
        "import seaborn as sns"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wFuxaySu6-bH",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 387
        },
        "outputId": "704fd547-24a4-4588-aca9-d049868f1a27"
      },
      "source": [
        "sns.displot(df, x=\"px_vals\", hue=\"Color\",kind=\"kde\", bw_adjust=2,fill=True)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<seaborn.axisgrid.FacetGrid at 0x7f5c1f29db10>"
            ]
          },
          "metadata": {},
          "execution_count": 11
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 427x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EUUYX3OUBxXr",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 387
        },
        "outputId": "1e1739b0-3cf6-4900-f27e-5752b437bd64"
      },
      "source": [
        "sns.displot(df, x=\"px_vals\", hue=\"Color\", stat=\"density\",element=\"step\",common_norm=False,)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<seaborn.axisgrid.FacetGrid at 0x7f5c095f12d0>"
            ]
          },
          "metadata": {},
          "execution_count": 12
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 427x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YwFD9SSeHCAb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 305
        },
        "outputId": "2c2f45ba-3855-4023-fcbe-cc2f0caf9b08"
      },
      "source": [
        "sns.set_theme(style=\"ticks\")\n",
        "\n",
        "# Draw a nested boxplot to show bills by day and time\n",
        "sns.boxplot(x=\"Color\", y=\"px_vals\",\n",
        "            data=df)\n",
        "sns.despine(offset=20, trim=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LGvT673d0YKN",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "af055301-d808-45a8-a69e-f870a0de84a7"
      },
      "source": [
        "all_pixel_averages = white_mean_tot + black_mean_tot + asian_mean_tot + indian_mean_tot + other_mean_tot\n",
        "\n",
        "print(\"white %\")\n",
        "print(white_mean_tot/all_pixel_averages)\n",
        "print(\"black %\")\n",
        "print(black_mean_tot/all_pixel_averages)\n",
        "print(\"asian %\")\n",
        "print(asian_mean_tot/all_pixel_averages)\n",
        "print(\"indian %\")\n",
        "print(indian_mean_tot/all_pixel_averages)\n",
        "print(\"other %\")\n",
        "print(other_mean_tot/all_pixel_averages)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "white %\n",
            "0.2293886261083047\n",
            "black %\n",
            "0.2220849994255224\n",
            "asian %\n",
            "0.21925869355168734\n",
            "indian %\n",
            "0.23085355124077406\n",
            "other %\n",
            "0.09841412967371135\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1ulFv9Ql1s37",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ddf667d9-7e69-4ed9-87e1-98515a23328f"
      },
      "source": [
        "print('white average pixel average')\n",
        "print(white_mean_tot/white_img_count)\n",
        "print('black average pixel average')\n",
        "print(black_mean_tot/black_img_count)\n",
        "print('asian average pixel average')\n",
        "print(asian_mean_tot/asian_img_count)\n",
        "print('indian average pixel average')\n",
        "print(indian_mean_tot/indian_img_count)\n",
        "print('other average pixel average')\n",
        "print(other_mean_tot/other_img_count)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "white average pixel average\n",
            "132.492295407468\n",
            "black average pixel average\n",
            "103.96213426662254\n",
            "asian average pixel average\n",
            "135.34726190649593\n",
            "indian average pixel average\n",
            "123.07387722012564\n",
            "other average pixel average\n",
            "123.26039444444424\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mYenR_cibPts"
      },
      "source": [
        "#Results:\n",
        "\n",
        "*   White images are on average 132 (avg pixel value)\n",
        "\n",
        "*   Asian images have an average pixel brightness of 135\n",
        "\n",
        "*   Indian images have an average pixel brightness of 123\n",
        "\n",
        "*   Other images have an average pixel brightness of 123\n",
        "\n",
        "*   Black images have an average pixel brightness of 103\n",
        "\n",
        "\n",
        "\n",
        "#Percentage Difference:\n",
        "Using White as a baseline:\n",
        "\n",
        "*   Asian is + 2%\n",
        "\n",
        "*   Indian is - 7%\n",
        "\n",
        "*   Other is - 7%\n",
        "\n",
        "*   Black is - 22%\n",
        "\n",
        "\n",
        "Faces in the 'Black' category are 22% darker than White faces, and 15% darker than their nearest neighbors, Indian and Other.  Training using this dataset is likely to lead to bias, as the model will be well-trained on the lighter images and poorly trained on the images of persons of African descent."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qCpr1lsVBmRh"
      },
      "source": [
        "UTK Face Dataset: http://aicip.eecs.utk.edu/wiki/UTKFace"
      ]
    }
  ]
}