PREPROCESS_FACE.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "PREPROCESS_FACE.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jMn_xw7DUykV",
        "outputId": "a6302da1-5850-4012-f6bf-7fedc3429425"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PeRSdcI6U7f3"
      },
      "source": [
        "import os\n",
        "import glob\n",
        "import cv2\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from pathlib import Path\n",
        "from PIL import Image"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CLbipDAzVXmt"
      },
      "source": [
        "DATA_PATH = Path('/content/drive/My Drive/WORKSPACE/HK192/AI/AI_COLAB/FG-NET/FGNET/images')\n",
        "ALIGN_FACE_PATH = DATA_PATH.resolve().parent / 'align_face_v2'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4hRCzVZhp4h3",
        "outputId": "8ab02c73-c271-4bf0-fbf3-9cd57e978d2b"
      },
      "source": [
        "%cd /content/drive/My\\ Drive/WORKSPACE/HK192/AI/AI_COLAB/FG-NET\n",
        "# !git clone https://github.com/TropComplique/mtcnn-pytorch.git"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/drive/My Drive/WORKSPACE/HK192/AI/AI_COLAB/FG-NET\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 269
        },
        "id": "PKqxLXrJr2KW",
        "outputId": "97d40ba4-01b0-427a-f4ff-8095430c0fec"
      },
      "source": [
        "exam_img = Image.open(os.path.join(str(DATA_PATH), '001A08.JPG')).convert('RGB')\n",
        "plt.imshow(exam_img)\n",
        "plt.show()"
      ],
      "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": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PXeNm9pqr9WW"
      },
      "source": [
        "import numpy as np\n",
        "import torch\n",
        "from PIL import Image\n",
        "from torch.autograd import Variable\n",
        "from mtcnn_pytorch.src.get_nets import PNet, RNet, ONet\n",
        "from mtcnn_pytorch.src.box_utils import nms, calibrate_box, get_image_boxes, convert_to_square\n",
        "from mtcnn_pytorch.src.first_stage import run_first_stage\n",
        "from mtcnn_pytorch.src.align_trans import get_reference_facial_points, warp_and_crop_face\n",
        "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
        "\n",
        "class MTCNN():\n",
        "    def __init__(self):\n",
        "        self.pnet = PNet().to(device)\n",
        "        self.rnet = RNet().to(device)\n",
        "        self.onet = ONet().to(device)\n",
        "        self.pnet.eval()\n",
        "        self.rnet.eval()\n",
        "        self.onet.eval()\n",
        "        self.refrence = get_reference_facial_points(default_square= True)\n",
        "        \n",
        "    def align(self, img):\n",
        "        _, landmarks = self.detect_faces(img)\n",
        "        facial5points = [[landmarks[0][j],landmarks[0][j+5]] for j in range(5)]\n",
        "        warped_face = warp_and_crop_face(np.array(img), facial5points, self.refrence, crop_size=(112,112))\n",
        "        return Image.fromarray(warped_face)\n",
        "    \n",
        "    def align_multi(self, img, limit=None, min_face_size=30.0):\n",
        "        boxes, landmarks = self.detect_faces(img, min_face_size)\n",
        "        if limit:\n",
        "            boxes = boxes[:limit]\n",
        "            landmarks = landmarks[:limit]\n",
        "        faces = []\n",
        "        for landmark in landmarks:\n",
        "            facial5points = [[landmark[j],landmark[j+5]] for j in range(5)]\n",
        "            warped_face = warp_and_crop_face(np.array(img), facial5points, self.refrence, crop_size=(112,112))\n",
        "            faces.append(Image.fromarray(warped_face))\n",
        "        return boxes, faces\n",
        "\n",
        "    def detect_faces(self, image, min_face_size=20.0,\n",
        "                     thresholds=[0.6, 0.7, 0.8],\n",
        "                     nms_thresholds=[0.7, 0.7, 0.7]):\n",
        "        \"\"\"\n",
        "        Arguments:\n",
        "            image: an instance of PIL.Image.\n",
        "            min_face_size: a float number.\n",
        "            thresholds: a list of length 3.\n",
        "            nms_thresholds: a list of length 3.\n",
        "\n",
        "        Returns:\n",
        "            two float numpy arrays of shapes [n_boxes, 4] and [n_boxes, 10],\n",
        "            bounding boxes and facial landmarks.\n",
        "        \"\"\"\n",
        "\n",
        "        # BUILD AN IMAGE PYRAMID\n",
        "        width, height = image.size\n",
        "        min_length = min(height, width)\n",
        "\n",
        "        min_detection_size = 12\n",
        "        factor = 0.707  # sqrt(0.5)\n",
        "\n",
        "        # scales for scaling the image\n",
        "        scales = []\n",
        "\n",
        "        # scales the image so that\n",
        "        # minimum size that we can detect equals to\n",
        "        # minimum face size that we want to detect\n",
        "        m = min_detection_size/min_face_size\n",
        "        min_length *= m\n",
        "\n",
        "        factor_count = 0\n",
        "        while min_length > min_detection_size:\n",
        "            scales.append(m*factor**factor_count)\n",
        "            min_length *= factor\n",
        "            factor_count += 1\n",
        "\n",
        "        # STAGE 1\n",
        "\n",
        "        # it will be returned\n",
        "        bounding_boxes = []\n",
        "\n",
        "        with torch.no_grad():\n",
        "            # run P-Net on different scales\n",
        "            for s in scales:\n",
        "                boxes = run_first_stage(image, self.pnet, scale=s, threshold=thresholds[0])\n",
        "                bounding_boxes.append(boxes)\n",
        "\n",
        "            # collect boxes (and offsets, and scores) from different scales\n",
        "            bounding_boxes = [i for i in bounding_boxes if i is not None]\n",
        "            bounding_boxes = np.vstack(bounding_boxes)\n",
        "\n",
        "            keep = nms(bounding_boxes[:, 0:5], nms_thresholds[0])\n",
        "            bounding_boxes = bounding_boxes[keep]\n",
        "\n",
        "            # use offsets predicted by pnet to transform bounding boxes\n",
        "            bounding_boxes = calibrate_box(bounding_boxes[:, 0:5], bounding_boxes[:, 5:])\n",
        "            # shape [n_boxes, 5]\n",
        "\n",
        "            bounding_boxes = convert_to_square(bounding_boxes)\n",
        "            bounding_boxes[:, 0:4] = np.round(bounding_boxes[:, 0:4])\n",
        "\n",
        "            # STAGE 2\n",
        "\n",
        "            img_boxes = get_image_boxes(bounding_boxes, image, size=24)\n",
        "            img_boxes = torch.FloatTensor(img_boxes).to(device)\n",
        "\n",
        "            output = self.rnet(img_boxes)\n",
        "            offsets = output[0].cpu().data.numpy()  # shape [n_boxes, 4]\n",
        "            probs = output[1].cpu().data.numpy()  # shape [n_boxes, 2]\n",
        "\n",
        "            keep = np.where(probs[:, 1] > thresholds[1])[0]\n",
        "            bounding_boxes = bounding_boxes[keep]\n",
        "            bounding_boxes[:, 4] = probs[keep, 1].reshape((-1,))\n",
        "            offsets = offsets[keep]\n",
        "\n",
        "            keep = nms(bounding_boxes, nms_thresholds[1])\n",
        "            bounding_boxes = bounding_boxes[keep]\n",
        "            bounding_boxes = calibrate_box(bounding_boxes, offsets[keep])\n",
        "            bounding_boxes = convert_to_square(bounding_boxes)\n",
        "            bounding_boxes[:, 0:4] = np.round(bounding_boxes[:, 0:4])\n",
        "\n",
        "            # STAGE 3\n",
        "\n",
        "            img_boxes = get_image_boxes(bounding_boxes, image, size=48)\n",
        "            if len(img_boxes) == 0: \n",
        "                return [], []\n",
        "            img_boxes = torch.FloatTensor(img_boxes).to(device)\n",
        "            output = self.onet(img_boxes)\n",
        "            landmarks = output[0].cpu().data.numpy()  # shape [n_boxes, 10]\n",
        "            offsets = output[1].cpu().data.numpy()  # shape [n_boxes, 4]\n",
        "            probs = output[2].cpu().data.numpy()  # shape [n_boxes, 2]\n",
        "\n",
        "            keep = np.where(probs[:, 1] > thresholds[2])[0]\n",
        "            bounding_boxes = bounding_boxes[keep]\n",
        "            bounding_boxes[:, 4] = probs[keep, 1].reshape((-1,))\n",
        "            offsets = offsets[keep]\n",
        "            landmarks = landmarks[keep]\n",
        "\n",
        "            # compute landmark points\n",
        "            width = bounding_boxes[:, 2] - bounding_boxes[:, 0] + 1.0\n",
        "            height = bounding_boxes[:, 3] - bounding_boxes[:, 1] + 1.0\n",
        "            xmin, ymin = bounding_boxes[:, 0], bounding_boxes[:, 1]\n",
        "            landmarks[:, 0:5] = np.expand_dims(xmin, 1) + np.expand_dims(width, 1)*landmarks[:, 0:5]\n",
        "            landmarks[:, 5:10] = np.expand_dims(ymin, 1) + np.expand_dims(height, 1)*landmarks[:, 5:10]\n",
        "\n",
        "            bounding_boxes = calibrate_box(bounding_boxes, offsets)\n",
        "            keep = nms(bounding_boxes, nms_thresholds[2], mode='min')\n",
        "            bounding_boxes = bounding_boxes[keep]\n",
        "            landmarks = landmarks[keep]\n",
        "\n",
        "        return bounding_boxes, landmarks"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qxdQz1GBtoAm"
      },
      "source": [
        "mtcnn = MTCNN()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "djJffMBr54Kw"
      },
      "source": [
        "def preprocess_images(img_dir, save_dir):\n",
        "  if not os.path.exists(save_dir):\n",
        "    os.mkdir(save_dir)\n",
        "\n",
        "  for idx, img in enumerate(sorted(os.listdir(img_dir))):\n",
        "    path = os.path.join(img_dir, img)\n",
        "    name = img.split(os.sep)[-1]\n",
        "    face = Image.open(path).convert('RGB')\n",
        "    boxces, align_faces = mtcnn.align_multi(face, limit=1)\n",
        "\n",
        "    if len(align_faces) == 0:\n",
        "      print('Cannot extract face from {}'.format(name))\n",
        "      resize_face = align_faces[0].resize((256, 256))\n",
        "      resize_face.save(os.path.join(save_dir, name))\n",
        "      print('{} finish'.format(name))\n",
        "      continue\n",
        "\n",
        "    resize_face = align_faces[0].resize((256, 256))\n",
        "    resize_face.save(os.path.join(save_dir, name))\n",
        "    print('{} finish'.format(name))\n",
        "  print('Done')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LJRKQeJa8k8Z",
        "outputId": "96371e44-b4c3-48ef-b387-ec49feff70d0"
      },
      "source": [
        "preprocess_images(DATA_PATH, ALIGN_FACE_PATH)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/drive/MyDrive/WORKSPACE/HK192/AI/AI_COLAB/FG-NET/mtcnn_pytorch/src/first_stage.py:32: UserWarning: volatile was removed and now has no effect. Use `with torch.no_grad():` instead.\n",
            "  img = Variable(torch.FloatTensor(_preprocess(img)), volatile=True)\n",
            "/content/drive/MyDrive/WORKSPACE/HK192/AI/AI_COLAB/FG-NET/mtcnn_pytorch/src/get_nets.py:70: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
            "  a = F.softmax(a)\n",
            "/content/drive/MyDrive/WORKSPACE/HK192/AI/AI_COLAB/FG-NET/mtcnn_pytorch/src/get_nets.py:115: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
            "  a = F.softmax(a)\n",
            "/content/drive/MyDrive/WORKSPACE/HK192/AI/AI_COLAB/FG-NET/mtcnn_pytorch/src/get_nets.py:168: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
            "  a = F.softmax(a)\n",
            "mtcnn_pytorch/src/matlab_cp2tform.py:90: FutureWarning: `rcond` parameter will change to the default of machine precision times ``max(M, N)`` where M and N are the input matrix dimensions.\n",
            "To use the future default and silence this warning we advise to pass `rcond=None`, to keep using the old, explicitly pass `rcond=-1`.\n",
            "  r, _, _, _ = lstsq(X, U)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "001A02.JPG finish\n",
            "001A05.JPG finish\n",
            "001A08.JPG finish\n",
            "001A10.JPG finish\n",
            "001A14.JPG finish\n",
            "001A16.JPG finish\n",
            "001A18.JPG finish\n",
            "001A19.JPG finish\n",
            "001A22.JPG finish\n",
            "001A28.JPG finish\n",
            "001A29.JPG finish\n",
            "001A33.JPG finish\n",
            "001A40.JPG finish\n",
            "001A43a.JPG finish\n",
            "001A43b.JPG finish\n",
            "002A03.JPG finish\n",
            "002A04.JPG finish\n",
            "002A05.JPG finish\n",
            "002A07.JPG finish\n",
            "002A12.JPG finish\n",
            "002A15.JPG finish\n",
            "002A16.JPG finish\n",
            "002A18.JPG finish\n",
            "002A20.JPG finish\n",
            "002A21.JPG finish\n",
            "002A23.JPG finish\n",
            "002A26.JPG finish\n",
            "002A29.JPG finish\n",
            "002A31.JPG finish\n",
            "002A36.JPG finish\n",
            "002A38.JPG finish\n",
            "003A18.JPG finish\n",
            "003A20.JPG finish\n",
            "003A23.JPG finish\n",
            "003A25.JPG finish\n",
            "003A35.JPG finish\n",
            "003A38.JPG finish\n",
            "003A47.JPG finish\n",
            "003A49.JPG finish\n",
            "003A51.JPG finish\n",
            "003A58.JPG finish\n",
            "003A60.JPG finish\n",
            "003A61.JPG finish\n",
            "004A19.JPG finish\n",
            "004A21.JPG finish\n",
            "004A26.JPG finish\n",
            "004A28.JPG finish\n",
            "004A30.JPG finish\n",
            "004A37.JPG finish\n",
            "004A40.JPG finish\n",
            "004A48.JPG finish\n",
            "004A51.JPG finish\n",
            "004A53.JPG finish\n",
            "004A62.JPG finish\n",
            "004A63.JPG finish\n",
            "005A18.JPG finish\n",
            "005A24.JPG finish\n",
            "005A30.JPG finish\n",
            "005A31.JPG finish\n",
            "005A35.JPG finish\n",
            "005A40.JPG finish\n",
            "005A45.JPG finish\n",
            "005A48.JPG finish\n",
            "005A49.JPG finish\n",
            "005A52.JPG finish\n",
            "005A61.JPG finish\n",
            "006A24.JPG finish\n",
            "006A28.JPG finish\n",
            "006A31.JPG finish\n",
            "006A36.JPG finish\n",
            "006A40.JPG finish\n",
            "006A42.JPG finish\n",
            "006A46.JPG finish\n",
            "006A51.JPG finish\n",
            "006A55.JPG finish\n",
            "006A61.JPG finish\n",
            "006A67.JPG finish\n",
            "006A69.JPG finish\n",
            "007A01.JPG finish\n",
            "007A18.JPG finish\n",
            "007A22.JPG finish\n",
            "007A23.JPG finish\n",
            "007A26.JPG finish\n",
            "007A37.JPG finish\n",
            "007A38.JPG finish\n",
            "007A39.JPG finish\n",
            "007A45.JPG finish\n",
            "008A03.JPG finish\n",
            "008A06.JPG finish\n",
            "008A08.JPG finish\n",
            "008A12.JPG finish\n",
            "008A13.JPG finish\n",
            "008A16.JPG finish\n",
            "008A17.JPG finish\n",
            "008A18.JPG finish\n",
            "008A21.JPG finish\n",
            "008A29.JPG finish\n",
            "008A30.JPG finish\n",
            "008A31.JPG finish\n",
            "008A33.JPG finish\n",
            "008A35.JPG finish\n",
            "008A41.JPG finish\n",
            "009A00.JPG finish\n",
            "009A01.JPG finish\n",
            "009A03.JPG finish\n",
            "009A05.JPG finish\n",
            "009A09.JPG finish\n",
            "009A11.JPG finish\n",
            "009A13.JPG finish\n",
            "009A14.JPG finish\n",
            "009A16a.JPG finish\n",
            "009A16b.JPG finish\n",
            "009A18.JPG finish\n",
            "009A22a.JPG finish\n",
            "009A22b.JPG finish\n",
            "010A01.JPG finish\n",
            "010A04.JPG finish\n",
            "010A05.JPG finish\n",
            "010A06.JPG finish\n",
            "010A07a.JPG finish\n",
            "010A07b.JPG finish\n",
            "010A09.JPG finish\n",
            "010A10.JPG finish\n",
            "010A12.JPG finish\n",
            "010A15.JPG finish\n",
            "010A16.JPG finish\n",
            "010A18.JPG finish\n",
            "011A02.JPG finish\n",
            "011A05.JPG finish\n",
            "011A07.JPG finish\n",
            "011A11.JPG finish\n",
            "011A13.JPG finish\n",
            "011A14.JPG finish\n",
            "011A15.JPG finish\n",
            "011A17.JPG finish\n",
            "011A20.JPG finish\n",
            "011A27.JPG finish\n",
            "011A30.JPG finish\n",
            "011A34.JPG finish\n",
            "011A40.JPG finish\n",
            "011A42.JPG finish\n",
            "012A04.JPG finish\n",
            "012A07.JPG finish\n",
            "012A12.JPG finish\n",
            "012A14.JPG finish\n",
            "012A18.JPG finish\n",
            "012A21.JPG finish\n",
            "012A23.JPG finish\n",
            "012A24.JPG finish\n",
            "012A25.JPG finish\n",
            "012A26.JPG finish\n",
            "012A27.JPG finish\n",
            "012A30.JPG finish\n",
            "012A32.JPG finish\n",
            "012A34.JPG finish\n",
            "012A36.JPG finish\n",
            "013A00.JPG finish\n",
            "013A08.JPG finish\n",
            "013A16.JPG finish\n",
            "013A18.JPG finish\n",
            "013A19.JPG finish\n",
            "013A21.JPG finish\n",
            "013A23.JPG finish\n",
            "013A25.JPG finish\n",
            "013A30.JPG finish\n",
            "013A34.JPG finish\n",
            "013A41.JPG finish\n",
            "013A44.JPG finish\n",
            "Cannot extract face from 014A03.JPG\n",
            "014A12.JPG finish\n",
            "014A14.JPG finish\n",
            "014A16.JPG finish\n",
            "014A18.JPG finish\n",
            "014A22.JPG finish\n",
            "014A30.JPG finish\n",
            "014A36.JPG finish\n",
            "014A40.JPG finish\n",
            "014A42.JPG finish\n",
            "015A00.JPG finish\n",
            "015A01.JPG finish\n",
            "015A03.JPG finish\n",
            "015A04.JPG finish\n",
            "015A05.JPG finish\n",
            "015A07.JPG finish\n",
            "015A09.JPG finish\n",
            "015A10.JPG finish\n",
            "015A12.JPG finish\n",
            "015A14.JPG finish\n",
            "015A15.JPG finish\n",
            "015A17.JPG finish\n",
            "015A19.JPG finish\n",
            "016A01.JPG finish\n",
            "016A03.JPG finish\n",
            "016A04.JPG finish\n",
            "016A05.JPG finish\n",
            "016A07.JPG finish\n",
            "016A08.JPG finish\n",
            "016A10.JPG finish\n",
            "016A12.JPG finish\n",
            "016A14.JPG finish\n",
            "016A16.JPG finish\n",
            "016A17.JPG finish\n",
            "016A18.JPG finish\n",
            "016A19.JPG finish\n",
            "017A05.JPG finish\n",
            "017A07.JPG finish\n",
            "017A12.JPG finish\n",
            "017A15.JPG finish\n",
            "017A17.JPG finish\n",
            "017A19.JPG finish\n",
            "017A22.JPG finish\n",
            "017A26.JPG finish\n",
            "017A29.JPG finish\n",
            "017A32.JPG finish\n",
            "017A36.JPG finish\n",
            "017A38.JPG finish\n",
            "017A42.JPG finish\n",
            "018A06.JPG finish\n",
            "018A11.JPG finish\n",
            "018A16.JPG finish\n",
            "018A18.JPG finish\n",
            "018A22.JPG finish\n",
            "018A25.JPG finish\n",
            "018A28.JPG finish\n",
            "018A29.JPG finish\n",
            "018A33.JPG finish\n",
            "018A34.JPG finish\n",
            "018A36.JPG finish\n",
            "019A00.JPG finish\n",
            "019A03.JPG finish\n",
            "019A05.JPG finish\n",
            "019A07.JPG finish\n",
            "019A12.JPG finish\n",
            "019A15.JPG finish\n",
            "019A18.JPG finish\n",
            "019A23.JPG finish\n",
            "019A27.JPG finish\n",
            "019A30.JPG finish\n",
            "020A02.JPG finish\n",
            "020A04.JPG finish\n",
            "020A08.JPG finish\n",
            "020A11.JPG finish\n",
            "020A13.JPG finish\n",
            "020A16.JPG finish\n",
            "020A18.JPG finish\n",
            "020A20.JPG finish\n",
            "020A22.JPG finish\n",
            "020A24.JPG finish\n",
            "020A26.JPG finish\n",
            "020A30.JPG finish\n",
            "020A36.JPG finish\n",
            "021A03.JPG finish\n",
            "021A05.JPG finish\n",
            "021A06.JPG finish\n",
            "021A11.JPG finish\n",
            "021A16.JPG finish\n",
            "021A18.JPG finish\n",
            "021A20.JPG finish\n",
            "021A23.JPG finish\n",
            "021A27.JPG finish\n",
            "021A30.JPG finish\n",
            "021A35.JPG finish\n",
            "021A39.JPG finish\n",
            "022A00.JPG finish\n",
            "022A07.JPG finish\n",
            "022A11.JPG finish\n",
            "022A14.JPG finish\n",
            "022A16.JPG finish\n",
            "022A18.JPG finish\n",
            "022A20.JPG finish\n",
            "022A23.JPG finish\n",
            "022A25.JPG finish\n",
            "022A26.JPG finish\n",
            "022A27.JPG finish\n",
            "022A28.JPG finish\n",
            "022A30.JPG finish\n",
            "023A00.JPG finish\n",
            "023A03.JPG finish\n",
            "023A05.JPG finish\n",
            "023A07.JPG finish\n",
            "023A09.JPG finish\n",
            "023A14.JPG finish\n",
            "023A16.JPG finish\n",
            "023A18.JPG finish\n",
            "023A21.JPG finish\n",
            "023A22.JPG finish\n",
            "023A25.JPG finish\n",
            "023A29.JPG finish\n",
            "024A00.JPG finish\n",
            "024A05.JPG finish\n",
            "024A07.JPG finish\n",
            "024A10.JPG finish\n",
            "024A17.JPG finish\n",
            "024A18.JPG finish\n",
            "024A19.JPG finish\n",
            "024A23.JPG finish\n",
            "024A25.JPG finish\n",
            "024A28.JPG finish\n",
            "024A30.JPG finish\n",
            "025A00.JPG finish\n",
            "025A03.JPG finish\n",
            "025A07.JPG finish\n",
            "025A12.JPG finish\n",
            "025A15.JPG finish\n",
            "025A18.JPG finish\n",
            "025A19.JPG finish\n",
            "025A22.JPG finish\n",
            "025A23.JPG finish\n",
            "025A28.JPG finish\n",
            "025A34.JPG finish\n",
            "025A39.JPG finish\n",
            "026A00.JPG finish\n",
            "026A02.JPG finish\n",
            "026A04.JPG finish\n",
            "026A06.JPG finish\n",
            "026A07.JPG finish\n",
            "026A11.JPG finish\n",
            "026A13.JPG finish\n",
            "026A15.JPG finish\n",
            "026A17.JPG finish\n",
            "026A18.JPG finish\n",
            "026A19.JPG finish\n",
            "027A02.JPG finish\n",
            "027A11.JPG finish\n",
            "027A13.JPG finish\n",
            "027A16.JPG finish\n",
            "027A18.JPG finish\n",
            "027A20.JPG finish\n",
            "027A22.JPG finish\n",
            "027A25.JPG finish\n",
            "027A30.JPG finish\n",
            "027A36.JPG finish\n",
            "027A41.JPG finish\n",
            "028A14.JPG finish\n",
            "028A16.JPG finish\n",
            "028A18.JPG finish\n",
            "028A20.JPG finish\n",
            "028A26.JPG finish\n",
            "028A28.JPG finish\n",
            "028A34.JPG finish\n",
            "028A35.JPG finish\n",
            "028A37.JPG finish\n",
            "028A41.JPG finish\n",
            "028A46.JPG finish\n",
            "029A01.JPG finish\n",
            "029A04.JPG finish\n",
            "029A05.JPG finish\n",
            "029A10.JPG finish\n",
            "029A11.JPG finish\n",
            "029A13.JPG finish\n",
            "029A16.JPG finish\n",
            "029A20.JPG finish\n",
            "029A22.JPG finish\n",
            "029A23.JPG finish\n",
            "029A29.JPG finish\n",
            "029A31.JPG finish\n",
            "029A33.JPG finish\n",
            "030A00.JPG finish\n",
            "030A04.JPG finish\n",
            "030A05.JPG finish\n",
            "030A07.JPG finish\n",
            "030A10.JPG finish\n",
            "030A16.JPG finish\n",
            "030A17.JPG finish\n",
            "030A18.JPG finish\n",
            "030A20.JPG finish\n",
            "030A23.JPG finish\n",
            "030A26.JPG finish\n",
            "031A00.JPG finish\n",
            "031A02.JPG finish\n",
            "031A03.JPG finish\n",
            "031A04.JPG finish\n",
            "031A06.JPG finish\n",
            "031A07.JPG finish\n",
            "031A09.JPG finish\n",
            "031A11.JPG finish\n",
            "031A12.JPG finish\n",
            "031A14.JPG finish\n",
            "031A15.JPG finish\n",
            "031A16.JPG finish\n",
            "031A19.JPG finish\n",
            "032A06.JPG finish\n",
            "032A08.JPG finish\n",
            "032A10.JPG finish\n",
            "032A13.JPG finish\n",
            "032A15.JPG finish\n",
            "032A18.JPG finish\n",
            "032A20.JPG finish\n",
            "032A24.JPG finish\n",
            "032A28.JPG finish\n",
            "032A33.JPG finish\n",
            "032A35.JPG finish\n",
            "032A40.JPG finish\n",
            "Cannot extract face from 033A02.JPG\n",
            "033A05.JPG finish\n",
            "033A10.JPG finish\n",
            "033A14.JPG finish\n",
            "033A18.JPG finish\n",
            "033A21.JPG finish\n",
            "033A25.JPG finish\n",
            "033A30.JPG finish\n",
            "033A35.JPG finish\n",
            "033A39.JPG finish\n",
            "033A44.JPG finish\n",
            "034A02.JPG finish\n",
            "034A04.JPG finish\n",
            "034A11.JPG finish\n",
            "034A13.JPG finish\n",
            "034A15.JPG finish\n",
            "034A16.JPG finish\n",
            "034A18.JPG finish\n",
            "034A21.JPG finish\n",
            "034A25.JPG finish\n",
            "034A30.JPG finish\n",
            "034A35.JPG finish\n",
            "034A39.JPG finish\n",
            "034A44.JPG finish\n",
            "035A00.JPG finish\n",
            "035A01.JPG finish\n",
            "035A03.JPG finish\n",
            "035A05.JPG finish\n",
            "035A07.JPG finish\n",
            "035A08.JPG finish\n",
            "035A09.JPG finish\n",
            "035A12.JPG finish\n",
            "035A14.JPG finish\n",
            "035A16.JPG finish\n",
            "035A18.JPG finish\n",
            "035A21.JPG finish\n",
            "035A23a.JPG finish\n",
            "035A23b.JPG finish\n",
            "036A00.JPG finish\n",
            "036A01.JPG finish\n",
            "036A03.JPG finish\n",
            "036A05.JPG finish\n",
            "036A06.JPG finish\n",
            "036A08.JPG finish\n",
            "036A11.JPG finish\n",
            "036A13.JPG finish\n",
            "036A15.JPG finish\n",
            "036A17.JPG finish\n",
            "036A18.JPG finish\n",
            "036A19.JPG finish\n",
            "036A20.JPG finish\n",
            "037A00.JPG finish\n",
            "037A01.JPG finish\n",
            "037A02.JPG finish\n",
            "037A04.JPG finish\n",
            "037A06.JPG finish\n",
            "037A08.JPG finish\n",
            "037A09.JPG finish\n",
            "037A11.JPG finish\n",
            "037A13.JPG finish\n",
            "037A17.JPG finish\n",
            "037A18.JPG finish\n",
            "037A19.JPG finish\n",
            "038A00.JPG finish\n",
            "038A01.JPG finish\n",
            "038A03.JPG finish\n",
            "038A04.JPG finish\n",
            "038A05.JPG finish\n",
            "038A08.JPG finish\n",
            "038A10.JPG finish\n",
            "038A11.JPG finish\n",
            "038A13.JPG finish\n",
            "038A14.JPG finish\n",
            "038A17.JPG finish\n",
            "038A21.JPG finish\n",
            "038A23.JPG finish\n",
            "038A34.JPG finish\n",
            "039A02.JPG finish\n",
            "039A05.JPG finish\n",
            "039A07.JPG finish\n",
            "039A10.JPG finish\n",
            "039A11.JPG finish\n",
            "039A13.JPG finish\n",
            "039A15.JPG finish\n",
            "039A16.JPG finish\n",
            "039A17.JPG finish\n",
            "039A19.JPG finish\n",
            "039A35.JPG finish\n",
            "039A45.JPG finish\n",
            "039A50.JPG finish\n",
            "039A52.JPG finish\n",
            "040A00.JPG finish\n",
            "040A02.JPG finish\n",
            "040A03a.JPG finish\n",
            "040A03b.JPG finish\n",
            "040A05.JPG finish\n",
            "040A06.JPG finish\n",
            "040A07.JPG finish\n",
            "040A09.JPG finish\n",
            "040A10.JPG finish\n",
            "040A12.JPG finish\n",
            "040A14.JPG finish\n",
            "040A16.JPG finish\n",
            "040A19.JPG finish\n",
            "040A21.JPG finish\n",
            "041A00.JPG finish\n",
            "041A02.JPG finish\n",
            "041A04.JPG finish\n",
            "041A06.JPG finish\n",
            "041A09.JPG finish\n",
            "041A13.JPG finish\n",
            "041A19.JPG finish\n",
            "041A20.JPG finish\n",
            "041A21.JPG finish\n",
            "041A24.JPG finish\n",
            "042A00.JPG finish\n",
            "042A01.JPG finish\n",
            "042A04.JPG finish\n",
            "042A05.JPG finish\n",
            "042A06.JPG finish\n",
            "042A08.JPG finish\n",
            "042A10.JPG finish\n",
            "042A12.JPG finish\n",
            "042A14.JPG finish\n",
            "042A18.JPG finish\n",
            "042A20.JPG finish\n",
            "042A24.JPG finish\n",
            "042A28.JPG finish\n",
            "043A00.JPG finish\n",
            "043A03.JPG finish\n",
            "043A04.JPG finish\n",
            "043A05.JPG finish\n",
            "043A10.JPG finish\n",
            "043A13.JPG finish\n",
            "043A15.JPG finish\n",
            "043A17.JPG finish\n",
            "043A19.JPG finish\n",
            "043A23.JPG finish\n",
            "043A27.JPG finish\n",
            "044A01.JPG finish\n",
            "044A03.JPG finish\n",
            "044A05.JPG finish\n",
            "044A07.JPG finish\n",
            "044A13.JPG finish\n",
            "044A15.JPG finish\n",
            "044A17.JPG finish\n",
            "044A18.JPG finish\n",
            "044A21.JPG finish\n",
            "044A25.JPG finish\n",
            "045A03.JPG finish\n",
            "045A06.JPG finish\n",
            "045A08.JPG finish\n",
            "045A12.JPG finish\n",
            "045A17.JPG finish\n",
            "045A18.JPG finish\n",
            "045A19.JPG finish\n",
            "045A25.JPG finish\n",
            "045A28.JPG finish\n",
            "045A32.JPG finish\n",
            "045A40.JPG finish\n",
            "045A43.JPG finish\n",
            "045A48.JPG finish\n",
            "046A00.JPG finish\n",
            "046A01.JPG finish\n",
            "046A02.JPG finish\n",
            "046A04.JPG finish\n",
            "046A06.JPG finish\n",
            "046A08.JPG finish\n",
            "046A10.JPG finish\n",
            "046A12.JPG finish\n",
            "046A14.JPG finish\n",
            "046A16.JPG finish\n",
            "046A17.JPG finish\n",
            "046A18.JPG finish\n",
            "046A20.JPG finish\n",
            "047A05.JPG finish\n",
            "047A06.JPG finish\n",
            "047A09.JPG finish\n",
            "047A11.JPG finish\n",
            "047A12.JPG finish\n",
            "047A13.JPG finish\n",
            "047A15.JPG finish\n",
            "047A18.JPG finish\n",
            "047A20.JPG finish\n",
            "047A23.JPG finish\n",
            "047A27.JPG finish\n",
            "047A33.JPG finish\n",
            "047A45.JPG finish\n",
            "047A55.JPG finish\n",
            "048A00.JPG finish\n",
            "048A09.JPG finish\n",
            "048A10.JPG finish\n",
            "048A11.JPG finish\n",
            "048A15.JPG finish\n",
            "048A16.JPG finish\n",
            "048A17.JPG finish\n",
            "048A18.JPG finish\n",
            "048A27.JPG finish\n",
            "048A30.JPG finish\n",
            "048A33.JPG finish\n",
            "048A46.JPG finish\n",
            "048A49.JPG finish\n",
            "048A50.JPG finish\n",
            "048A52.JPG finish\n",
            "048A54.JPG finish\n",
            "049A04.JPG finish\n",
            "049A05.JPG finish\n",
            "049A06.JPG finish\n",
            "049A07.JPG finish\n",
            "049A08.JPG finish\n",
            "049A10.JPG finish\n",
            "049A12.JPG finish\n",
            "049A15.JPG finish\n",
            "049A16.JPG finish\n",
            "049A17.JPG finish\n",
            "Cannot extract face from 050A00.JPG\n",
            "050A03.JPG finish\n",
            "050A06.JPG finish\n",
            "050A10.JPG finish\n",
            "050A14.JPG finish\n",
            "050A16.JPG finish\n",
            "050A18.JPG finish\n",
            "050A22.JPG finish\n",
            "051A00.JPG finish\n",
            "051A02.JPG finish\n",
            "051A05.JPG finish\n",
            "051A08.JPG finish\n",
            "051A12.JPG finish\n",
            "051A14.JPG finish\n",
            "051A15.JPG finish\n",
            "051A16.JPG finish\n",
            "051A17.JPG finish\n",
            "051A18.JPG finish\n",
            "051A24.JPG finish\n",
            "052A02.JPG finish\n",
            "052A04.JPG finish\n",
            "052A07.JPG finish\n",
            "052A10.JPG finish\n",
            "052A13.JPG finish\n",
            "052A14.JPG finish\n",
            "052A16.JPG finish\n",
            "052A17.JPG finish\n",
            "052A18.JPG finish\n",
            "052A19.JPG finish\n",
            "052A20.JPG finish\n",
            "053A00.JPG finish\n",
            "053A01.JPG finish\n",
            "053A02.JPG finish\n",
            "053A03.JPG finish\n",
            "053A04.JPG finish\n",
            "053A06.JPG finish\n",
            "053A08.JPG finish\n",
            "053A10.JPG finish\n",
            "053A11.JPG finish\n",
            "053A12.JPG finish\n",
            "053A13.JPG finish\n",
            "053A15.JPG finish\n",
            "053A18.JPG finish\n",
            "054A00.JPG finish\n",
            "054A02.JPG finish\n",
            "054A04.JPG finish\n",
            "054A06.JPG finish\n",
            "054A09.JPG finish\n",
            "054A10.JPG finish\n",
            "054A11.JPG finish\n",
            "054A12.JPG finish\n",
            "054A13.JPG finish\n",
            "054A16.JPG finish\n",
            "054A17.JPG finish\n",
            "054A18.JPG finish\n",
            "054A19.JPG finish\n",
            "055A04.JPG finish\n",
            "055A05.JPG finish\n",
            "055A06.JPG finish\n",
            "055A08.JPG finish\n",
            "055A12.JPG finish\n",
            "055A14.JPG finish\n",
            "055A22.JPG finish\n",
            "055A23.JPG finish\n",
            "056A01.JPG finish\n",
            "056A02.JPG finish\n",
            "056A03.JPG finish\n",
            "056A04.JPG finish\n",
            "056A06.JPG finish\n",
            "056A10.JPG finish\n",
            "056A12.JPG finish\n",
            "056A15.JPG finish\n",
            "057A00.JPG finish\n",
            "057A05.JPG finish\n",
            "057A06.JPG finish\n",
            "057A07.JPG finish\n",
            "057A12.JPG finish\n",
            "057A16.JPG finish\n",
            "057A17.JPG finish\n",
            "057A18.JPG finish\n",
            "057A23.JPG finish\n",
            "057A25.JPG finish\n",
            "058A00.JPG finish\n",
            "058A01.JPG finish\n",
            "058A02.JPG finish\n",
            "058A03.JPG finish\n",
            "058A04.JPG finish\n",
            "058A05.JPG finish\n",
            "058A06.JPG finish\n",
            "058A10.JPG finish\n",
            "058A13.JPG finish\n",
            "058A15.JPG finish\n",
            "058A17.JPG finish\n",
            "059A02a.JPG finish\n",
            "059A02b.JPG finish\n",
            "059A03.JPG finish\n",
            "059A04.JPG finish\n",
            "059A06.JPG finish\n",
            "059A07.JPG finish\n",
            "059A12.JPG finish\n",
            "059A14.JPG finish\n",
            "059A16.JPG finish\n",
            "060A02.JPG finish\n",
            "060A04.JPG finish\n",
            "060A05.JPG finish\n",
            "060A06.JPG finish\n",
            "060A07.JPG finish\n",
            "060A08.JPG finish\n",
            "060A10.JPG finish\n",
            "060A11.JPG finish\n",
            "060A12.JPG finish\n",
            "060A17.JPG finish\n",
            "060A19.JPG finish\n",
            "060A20.JPG finish\n",
            "061A04.JPG finish\n",
            "061A06.JPG finish\n",
            "061A07.JPG finish\n",
            "061A10.JPG finish\n",
            "061A15.JPG finish\n",
            "061A17.JPG finish\n",
            "061A18.JPG finish\n",
            "061A19.JPG finish\n",
            "061A25.JPG finish\n",
            "061A26.JPG finish\n",
            "061A29.JPG finish\n",
            "061A30.JPG finish\n",
            "061A32.JPG finish\n",
            "062A03.JPG finish\n",
            "062A15.JPG finish\n",
            "062A16.JPG finish\n",
            "062A18.JPG finish\n",
            "062A30.JPG finish\n",
            "062A35.JPG finish\n",
            "062A38.JPG finish\n",
            "062A41.JPG finish\n",
            "062A45.JPG finish\n",
            "062A47.JPG finish\n",
            "062A53.JPG finish\n",
            "062A54.JPG finish\n",
            "063A00.JPG finish\n",
            "063A02.JPG finish\n",
            "063A04.JPG finish\n",
            "063A08.JPG finish\n",
            "063A09.JPG finish\n",
            "063A18.JPG finish\n",
            "063A20.JPG finish\n",
            "063A22.JPG finish\n",
            "063A25.JPG finish\n",
            "063A33.JPG finish\n",
            "064A01.JPG finish\n",
            "064A03.JPG finish\n",
            "064A04.JPG finish\n",
            "064A10.JPG finish\n",
            "064A15.JPG finish\n",
            "064A19.JPG finish\n",
            "065A02a.JPG finish\n",
            "065A02b.JPG finish\n",
            "065A03a.JPG finish\n",
            "065A03b.JPG finish\n",
            "065A04.JPG finish\n",
            "065A06a.JPG finish\n",
            "065A06b.JPG finish\n",
            "065A07a.JPG finish\n",
            "065A07b.JPG finish\n",
            "065A08.JPG finish\n",
            "065A09.JPG finish\n",
            "065A10.JPG finish\n",
            "065A11.JPG finish\n",
            "065A12.JPG finish\n",
            "065A13.JPG finish\n",
            "066A00.JPG finish\n",
            "066A02a.JPG finish\n",
            "066A02b.JPG finish\n",
            "066A03.JPG finish\n",
            "066A04.JPG finish\n",
            "066A06a.JPG finish\n",
            "066A06b.JPG finish\n",
            "066A07.JPG finish\n",
            "066A08.JPG finish\n",
            "066A09.JPG finish\n",
            "066A10.JPG finish\n",
            "066A11.JPG finish\n",
            "067A04.JPG finish\n",
            "067A14.JPG finish\n",
            "067A17.JPG finish\n",
            "067A21.JPG finish\n",
            "067A23.JPG finish\n",
            "067A26.JPG finish\n",
            "067A28.JPG finish\n",
            "067A30.JPG finish\n",
            "067A33.JPG finish\n",
            "067A39.JPG finish\n",
            "068A00.JPG finish\n",
            "068A03.JPG finish\n",
            "068A04.JPG finish\n",
            "068A09.JPG finish\n",
            "068A10a.JPG finish\n",
            "068A10b.JPG finish\n",
            "068A11.JPG finish\n",
            "068A12.JPG finish\n",
            "068A13.JPG finish\n",
            "068A14.JPG finish\n",
            "069A00.JPG finish\n",
            "069A02.JPG finish\n",
            "069A03.JPG finish\n",
            "069A04.JPG finish\n",
            "069A05.JPG finish\n",
            "069A06.JPG finish\n",
            "069A07.JPG finish\n",
            "069A09.JPG finish\n",
            "069A10.JPG finish\n",
            "069A11.JPG finish\n",
            "069A13.JPG finish\n",
            "070A00.JPG finish\n",
            "070A01.JPG finish\n",
            "070A02.JPG finish\n",
            "070A03.JPG finish\n",
            "070A05.JPG finish\n",
            "070A06.JPG finish\n",
            "070A07.JPG finish\n",
            "070A08.JPG finish\n",
            "070A09.JPG finish\n",
            "070A12.JPG finish\n",
            "070A14.JPG finish\n",
            "071A00.JPG finish\n",
            "071A10.JPG finish\n",
            "071A18.JPG finish\n",
            "071A22.JPG finish\n",
            "071A23.JPG finish\n",
            "071A24.JPG finish\n",
            "071A27.JPG finish\n",
            "071A35.JPG finish\n",
            "071A40.JPG finish\n",
            "071A42.JPG finish\n",
            "071A43.JPG finish\n",
            "071A44.JPG finish\n",
            "071A45.JPG finish\n",
            "072A02.JPG finish\n",
            "072A07.JPG finish\n",
            "072A11.JPG finish\n",
            "072A13.JPG finish\n",
            "072A14.JPG finish\n",
            "072A17.JPG finish\n",
            "072A19.JPG finish\n",
            "072A21.JPG finish\n",
            "072A25.JPG finish\n",
            "072A27.JPG finish\n",
            "072A29.JPG finish\n",
            "072A34.JPG finish\n",
            "072A41.JPG finish\n",
            "072A45.JPG finish\n",
            "073A00.JPG finish\n",
            "073A01.JPG finish\n",
            "073A02.JPG finish\n",
            "073A03.JPG finish\n",
            "073A04.JPG finish\n",
            "073A05.JPG finish\n",
            "073A06.JPG finish\n",
            "073A07.JPG finish\n",
            "073A08.JPG finish\n",
            "073A09.JPG finish\n",
            "073A11.JPG finish\n",
            "073A12.JPG finish\n",
            "073A13.JPG finish\n",
            "073A14.JPG finish\n",
            "073A15.JPG finish\n",
            "073A16.JPG finish\n",
            "074A00.JPG finish\n",
            "074A01.JPG finish\n",
            "074A02.JPG finish\n",
            "074A03.JPG finish\n",
            "074A04.JPG finish\n",
            "074A05.JPG finish\n",
            "074A06.JPG finish\n",
            "074A07.JPG finish\n",
            "074A08.JPG finish\n",
            "074A09.JPG finish\n",
            "074A10.JPG finish\n",
            "074A11.JPG finish\n",
            "074A12.JPG finish\n",
            "074A13.JPG finish\n",
            "074A14.JPG finish\n",
            "074A15.JPG finish\n",
            "075A00.JPG finish\n",
            "075A01.JPG finish\n",
            "075A02.JPG finish\n",
            "075A03.JPG finish\n",
            "075A05.JPG finish\n",
            "075A06.JPG finish\n",
            "075A07.JPG finish\n",
            "075A08.JPG finish\n",
            "075A10.JPG finish\n",
            "075A11.JPG finish\n",
            "076A00.JPG finish\n",
            "076A01.JPG finish\n",
            "076A02.JPG finish\n",
            "076A03.JPG finish\n",
            "076A04.JPG finish\n",
            "076A05.JPG finish\n",
            "076A06.JPG finish\n",
            "076A07.JPG finish\n",
            "076A08.JPG finish\n",
            "076A09.JPG finish\n",
            "076A10.JPG finish\n",
            "076A11.JPG finish\n",
            "076A12.JPG finish\n",
            "076A13.JPG finish\n",
            "076A14.JPG finish\n",
            "076A15.JPG finish\n",
            "076A16.JPG finish\n",
            "076A17.JPG finish\n",
            "077A00.JPG finish\n",
            "077A01.JPG finish\n",
            "077A02.JPG finish\n",
            "077A03.JPG finish\n",
            "077A04.JPG finish\n",
            "077A06.JPG finish\n",
            "077A07.JPG finish\n",
            "077A08.JPG finish\n",
            "077A09.JPG finish\n",
            "077A10.JPG finish\n",
            "077A11.JPG finish\n",
            "077A12.JPG finish\n",
            "077A13.JPG finish\n",
            "077A14.JPG finish\n",
            "077A15.JPG finish\n",
            "077A16.JPG finish\n",
            "078A00.JPG finish\n",
            "078A01.JPG finish\n",
            "078A02.JPG finish\n",
            "078A03.JPG finish\n",
            "078A04.JPG finish\n",
            "078A05.JPG finish\n",
            "078A06.JPG finish\n",
            "078A07.JPG finish\n",
            "078A08.JPG finish\n",
            "078A09.JPG finish\n",
            "078A10.JPG finish\n",
            "078A11.JPG finish\n",
            "078A12.JPG finish\n",
            "078A13.JPG finish\n",
            "078A14.JPG finish\n",
            "078A15.JPG finish\n",
            "079A00.JPG finish\n",
            "079A01.JPG finish\n",
            "079A02.JPG finish\n",
            "079A03.JPG finish\n",
            "079A04.JPG finish\n",
            "079A05.JPG finish\n",
            "079A06.JPG finish\n",
            "079A07.JPG finish\n",
            "079A08.JPG finish\n",
            "079A09.JPG finish\n",
            "079A10.JPG finish\n",
            "079A11.JPG finish\n",
            "079A12.JPG finish\n",
            "079A13.JPG finish\n",
            "080A00.JPG finish\n",
            "080A01.JPG finish\n",
            "080A02.JPG finish\n",
            "080A03.JPG finish\n",
            "080A04.JPG finish\n",
            "080A05.JPG finish\n",
            "080A06.JPG finish\n",
            "080A07.JPG finish\n",
            "080A08.JPG finish\n",
            "080A09.JPG finish\n",
            "080A10.JPG finish\n",
            "080A12.JPG finish\n",
            "080A13.JPG finish\n",
            "080A14.JPG finish\n",
            "081A00.JPG finish\n",
            "081A01.JPG finish\n",
            "081A02.JPG finish\n",
            "081A03.JPG finish\n",
            "081A04.JPG finish\n",
            "081A05.JPG finish\n",
            "081A06.JPG finish\n",
            "081A07.JPG finish\n",
            "081A08.JPG finish\n",
            "081A10.JPG finish\n",
            "081A11.JPG finish\n",
            "081A12.JPG finish\n",
            "082A00.JPG finish\n",
            "082A03.JPG finish\n",
            "082A16.JPG finish\n",
            "082A18.JPG finish\n",
            "082A20.JPG finish\n",
            "082A21.JPG finish\n",
            "082A22.JPG finish\n",
            "082A23.JPG finish\n",
            "082A25.JPG finish\n",
            "082A27.JPG finish\n",
            "082A31.JPG finish\n",
            "Done\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zESLL8w9C4Gw"
      },
      "source": [
        "cant_extract_img = list(set(os.listdir(DATA_PATH)) - set(os.listdir(ALIGN_FACE_PATH)))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XWnU8t4sFWSE",
        "outputId": "2ca44481-1cc6-4942-f317-bf82d71af63f"
      },
      "source": [
        "cant_extract_img"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['050A00.JPG', '033A02.JPG', '014A03.JPG']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 35
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 190
        },
        "id": "gBeUSbK6FeOL",
        "outputId": "81f78a60-f3a9-4e0b-8231-6dc3b4da690f"
      },
      "source": [
        "for i, img in enumerate(sorted(cant_extract_img)):\n",
        "  im = Image.open(os.path.join(str(DATA_PATH), img)).convert('RGB')\n",
        "  im = np.asarray(im).astype('uint8')\n",
        "  plt.subplot(1, 3, i+1)\n",
        "  plt.imshow(im)\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 3 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}