Face_Detection_in_Images.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "8bfbac13",
   "metadata": {},
   "source": [
    "# Face Detection and Calculation in Images\n",
    "\n",
    "by Nan BAI\n",
    "\n",
    "adapted from the original code of facenet-pytorch library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ce5b6f9c",
   "metadata": {},
   "outputs": [],
   "source": [
    "from facenet_pytorch import MTCNN, InceptionResnetV1\n",
    "import torch\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision import datasets\n",
    "from torch.autograd import Variable as V\n",
    "from torchvision import transforms as trn\n",
    "from torch.nn import functional as F\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os\n",
    "from PIL import Image, ImageDraw\n",
    "from PIL.ImageOps import colorize\n",
    "from IPython import display\n",
    "\n",
    "workers = 0 if os.name == 'nt' else 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "325e73cf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running on device: cuda:0\n"
     ]
    }
   ],
   "source": [
    "device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')\n",
    "print('Running on device: {}'.format(device))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "aff10f80",
   "metadata": {},
   "outputs": [],
   "source": [
    "mtcnn = MTCNN(\n",
    "    image_size=160, margin=0, min_face_size=20,\n",
    "    thresholds=[0.6, 0.7, 0.7], factor=0.709, post_process=True,\n",
    "    device=device\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "ae74b839",
   "metadata": {},
   "outputs": [],
   "source": [
    "resnet = InceptionResnetV1(pretrained='vggface2').eval().to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "5794dd25",
   "metadata": {},
   "outputs": [],
   "source": [
    "def returnTF():\n",
    "# load the image transformer\n",
    "    tf = trn.Compose([\n",
    "        trn.Resize((150,150)),\n",
    "        trn.ToTensor(),\n",
    "        #trn.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ])\n",
    "    return tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "25dad6c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_images(images_folder):\n",
    "    all_images = []\n",
    "    for file in os.listdir(images_folder):\n",
    "        if os.path.splitext(file)[1].lower() in ('.jpg', '.jpeg', '.png'):\n",
    "            all_images.append(file)\n",
    "    return all_images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "30850d59",
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_images_from_category_folders(images_folder):\n",
    "    all_images = {}\n",
    "    all_imgs = []\n",
    "    categories = []\n",
    "    category_dict = {}\n",
    "    for file in os.listdir(images_folder):\n",
    "        if os.path.isdir(images_folder + '/' + file):\n",
    "            print(file)\n",
    "            categories.append(file)\n",
    "            category_dict[len(category_dict)] = file\n",
    "            all_images[file] = []\n",
    "            for f in os.listdir(images_folder + '/' + file):\n",
    "                if os.path.splitext(f)[1].lower() in ('.jpg', '.jpeg', '.png'):\n",
    "                    all_images[file].append(f)\n",
    "                    all_imgs.append((f, file, len(categories)))\n",
    "    return all_images, all_imgs, categories, category_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "27e2574f",
   "metadata": {},
   "outputs": [],
   "source": [
    "images_folder = 'Venezia/data_storage/images/150/'\n",
    "all_images = load_images(images_folder)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "b34bda85",
   "metadata": {},
   "outputs": [],
   "source": [
    "images_folder_Suzhou = 'Suzhou/data_storage/images/150/'\n",
    "all_images_Suzhou = load_images(images_folder_Suzhou)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "1c8cd603",
   "metadata": {},
   "outputs": [],
   "source": [
    "images_folder_Amsterdam = 'Amsterdam/data_storage/images/150/'\n",
    "all_images_Amsterdam = load_images(images_folder_Amsterdam)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "56e477f0",
   "metadata": {},
   "outputs": [],
   "source": [
    "images_folder_large = 'data_storage/images/grid-150/'\n",
    "all_images_large = load_images(images_folder_large)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "ffb1cd0e",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_one_image(ID = 0, all_images=all_images, images_folder=images_folder):\n",
    "    # load the test image\n",
    "    if type(all_images[0]) is tuple:\n",
    "        img_name = all_images[ID][0]\n",
    "        img = Image.open(images_folder + '/' + all_images[ID][1] +'/'+img_name)\n",
    "    else:\n",
    "        img_name = all_images[ID]\n",
    "        #os.system('wget %s -q -O test.jpg' % img_url)\n",
    "        img = Image.open(images_folder + img_name)\n",
    "    if img.mode == 'L':\n",
    "        img = colorize(img, black=(0,0,0), white=(255,255,255))\n",
    "    if img.mode == 'CMYK':\n",
    "        img = img.convert('RGB')\n",
    "    if not img.mode == 'RGB':\n",
    "        img = img.convert('RGB')\n",
    "    tf = returnTF()\n",
    "    input_img = V(tf(img).permute(1,2,0).unsqueeze(0))\n",
    "    img = img.resize((150,150))\n",
    "    return img_name, img, input_img*255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "d4a8fdef",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_one_image_name(name, all_images=all_images, images_folder=images_folder):\n",
    "    if type(all_images[0]) is tuple:\n",
    "        img_list = [a[0] for a in all_images]\n",
    "        cat_list = [a[1] for a in all_images]\n",
    "        assert name+'.jpg' in [a[0] for a in all_images]\n",
    "        ID = img_list.index(name+'.jpg')\n",
    "        img = Image.open(images_folder + '/' + cat_list[ID] +'/'+name+ '.jpg')\n",
    "    else:\n",
    "        assert name+'.jpg' in all_images\n",
    "        img = Image.open(images_folder + name + '.jpg')\n",
    "          \n",
    "    if img.mode == 'L':\n",
    "        img = colorize(img, black=(0,0,0), white=(255,255,255))\n",
    "    tf = returnTF()\n",
    "    input_img = V(tf(img).permute(1,2,0).unsqueeze(0))\n",
    "    img = img.resize((150,150))\n",
    "    return img, input_img*255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "b2f4c39b",
   "metadata": {},
   "outputs": [],
   "source": [
    "def process_one_image(img, input_img, model=mtcnn):\n",
    "    boxes, prob = mtcnn.detect(input_img)\n",
    "    for i in range(len(boxes)):\n",
    "        if len(prob[i])<=1:\n",
    "            if not prob[i][0]:\n",
    "                boxes[i] = []\n",
    "                prob[i] = [0]\n",
    "    return boxes, prob"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "da70b0d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "def draw_one_image(img, input_img, model=mtcnn):\n",
    "    boxes, prob = process_one_image(img, input_img, model=model)\n",
    "    frame = img.copy()\n",
    "    frame_draw = frame.copy()\n",
    "    draw = ImageDraw.Draw(frame_draw)\n",
    "    for box in boxes[0]:\n",
    "        draw.rectangle(box.tolist(), outline=(255, 0, 0), width=3)\n",
    "        \n",
    "    # Add to frame list\n",
    "    frames_draw = frame_draw.resize((150, 150), Image.BILINEAR)\n",
    "    d = display.display(frames_draw, display_id=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "d7b03913",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_areas(boxes):\n",
    "    areas = []\n",
    "    for box in boxes:\n",
    "        area = (box[2]-box[0])*(box[3]-box[1])\n",
    "        areas.append(area)\n",
    "    return np.array(areas).sum(),areas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "4fe4ceab",
   "metadata": {},
   "outputs": [],
   "source": [
    "def report_one_image(img, input_img, model=mtcnn):\n",
    "    boxes, prob = process_one_image(img, input_img, model=model)\n",
    "    report = {}\n",
    "    report['prob'] = prob[0]\n",
    "    report['boxes'] = boxes\n",
    "    report['num_faces'] = len(boxes)\n",
    "    area, areas = get_areas(boxes)\n",
    "    report['area'] = area\n",
    "    report['areas'] = areas\n",
    "    report['ratio'] = area/(img.size()[0]*img.size()[1])\n",
    "    return report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "9393f82e",
   "metadata": {},
   "outputs": [],
   "source": [
    "def report_many_images(img, input_img, image_names, model=mtcnn):\n",
    "    if len(image_names)==1:\n",
    "        return report_one_image(img, input_img, model=model)\n",
    "    \n",
    "    boxes, prob= process_one_image(img, input_img, model)\n",
    "    all_report = {}\n",
    "    \n",
    "    for n in range(len(image_names)):\n",
    "        report = {}\n",
    "        report['Face_prob'] = prob[n][0]\n",
    "        report['Face_boxes'] = boxes[n]\n",
    "        report['Face_num_faces'] = len(boxes[n])\n",
    "        area, areas = get_areas(boxes[n])\n",
    "        report['Face_area'] = area\n",
    "        report['Face_areas'] = areas\n",
    "        report['Face_ratio'] = area/(img[n].size()[0]*img[n].size()[1])\n",
    "        \n",
    "        all_report[image_names[n][:-4]] = report\n",
    "\n",
    "    return all_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "9f5be05f",
   "metadata": {},
   "outputs": [],
   "source": [
    "def report_images(IDs=[0,1], images_folder=images_folder, all_images=all_images, model=mtcnn):\n",
    "    image_names = [get_one_image(ID, all_images, images_folder)[0] for ID in IDs]\n",
    "    #for ID in IDs:\n",
    "    #    print(get_one_image(ID, all_images, images_folder)[0], get_one_image(ID, all_images, images_folder)[1])\n",
    "    imgs = torch.as_tensor(np.stack([get_one_image(ID, all_images, images_folder)[1] for ID in IDs]))\n",
    "    input_img = torch.cat(tuple([get_one_image(ID, all_images, images_folder)[-1] for ID in IDs]))\n",
    "    report = report_many_images(imgs, input_img, image_names, model=model)\n",
    "    return report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "a9402778",
   "metadata": {},
   "outputs": [],
   "source": [
    "def images_to_df(IDs=[0,1], images_folder=images_folder, all_images=all_images, model=mtcnn):\n",
    "    report = report_images(IDs, images_folder=images_folder, all_images=all_images, model=model)\n",
    "    image_names = report.keys()\n",
    "    df_all = pd.DataFrame(report)\n",
    "    return df_all"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "67369375",
   "metadata": {},
   "outputs": [],
   "source": [
    "def batch_df(images_folder=images_folder, all_images=all_images, batch_size = 8, model=mtcnn, ls_df = []):\n",
    "    size = len(all_images)\n",
    "    batch_num = int(size/batch_size)\n",
    "    #ls_df = []\n",
    "    for i in range(batch_num):\n",
    "        IDs = list(range(batch_size*i, batch_size*(i+1)))\n",
    "        temp_df = images_to_df(IDs, images_folder=images_folder, all_images=all_images, model=model)\n",
    "        ls_df.append(temp_df.T)\n",
    "        print('{}/{} batches finished'.format(i+1, batch_num+1))\n",
    "    IDs = list(range(batch_size*batch_num, size))\n",
    "    temp_df = images_to_df(IDs, images_folder=images_folder, all_images=all_images, model=model)\n",
    "    ls_df.append(temp_df.T)\n",
    "    df_all = pd.concat(ls_df, axis=0)\n",
    "    return df_all"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "2a67760e",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\utils\\detect_face.py:183: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  batch_boxes, batch_points = np.array(batch_boxes), np.array(batch_points)\n",
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\mtcnn.py:339: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  boxes = np.array(boxes)\n",
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\mtcnn.py:340: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  probs = np.array(probs)\n",
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\mtcnn.py:341: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  points = np.array(points)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/12 batches finished\n",
      "2/12 batches finished\n",
      "3/12 batches finished\n",
      "4/12 batches finished\n",
      "5/12 batches finished\n",
      "6/12 batches finished\n",
      "7/12 batches finished\n",
      "8/12 batches finished\n",
      "9/12 batches finished\n",
      "10/12 batches finished\n",
      "11/12 batches finished\n"
     ]
    }
   ],
   "source": [
    "Face_pred = batch_df(images_folder, all_images,256)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "2c93b5bf",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\utils\\detect_face.py:183: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  batch_boxes, batch_points = np.array(batch_boxes), np.array(batch_points)\n",
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\mtcnn.py:339: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  boxes = np.array(boxes)\n",
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\mtcnn.py:340: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  probs = np.array(probs)\n",
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\mtcnn.py:341: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  points = np.array(points)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/13 batches finished\n",
      "2/13 batches finished\n",
      "3/13 batches finished\n",
      "4/13 batches finished\n",
      "5/13 batches finished\n",
      "6/13 batches finished\n",
      "7/13 batches finished\n",
      "8/13 batches finished\n",
      "9/13 batches finished\n",
      "10/13 batches finished\n",
      "11/13 batches finished\n",
      "12/13 batches finished\n"
     ]
    }
   ],
   "source": [
    "Face_pred_Suzhou = batch_df(images_folder_Suzhou, all_images_Suzhou,256)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "33433ad9",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\utils\\detect_face.py:183: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  batch_boxes, batch_points = np.array(batch_boxes), np.array(batch_points)\n",
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\mtcnn.py:339: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  boxes = np.array(boxes)\n",
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\mtcnn.py:340: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  probs = np.array(probs)\n",
      "D:\\surfdrive\\TUD\\env\\lib\\site-packages\\facenet_pytorch\\models\\mtcnn.py:341: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  points = np.array(points)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/15 batches finished\n",
      "2/15 batches finished\n",
      "3/15 batches finished\n",
      "4/15 batches finished\n",
      "5/15 batches finished\n",
      "6/15 batches finished\n",
      "7/15 batches finished\n",
      "8/15 batches finished\n",
      "9/15 batches finished\n",
      "10/15 batches finished\n",
      "11/15 batches finished\n",
      "12/15 batches finished\n",
      "13/15 batches finished\n",
      "14/15 batches finished\n"
     ]
    }
   ],
   "source": [
    "Face_pred_Amsterdam = batch_df(images_folder_Amsterdam, all_images_Amsterdam,256)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "id": "de02b6f1",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/317 batches finished\n",
      "2/317 batches finished\n",
      "3/317 batches finished\n",
      "4/317 batches finished\n",
      "5/317 batches finished\n",
      "6/317 batches finished\n",
      "7/317 batches finished\n",
      "8/317 batches finished\n",
      "9/317 batches finished\n",
      "10/317 batches finished\n",
      "11/317 batches finished\n",
      "12/317 batches finished\n",
      "13/317 batches finished\n",
      "14/317 batches finished\n",
      "15/317 batches finished\n",
      "16/317 batches finished\n",
      "17/317 batches finished\n",
      "18/317 batches finished\n",
      "19/317 batches finished\n",
      "20/317 batches finished\n",
      "21/317 batches finished\n",
      "22/317 batches finished\n",
      "23/317 batches finished\n",
      "24/317 batches finished\n",
      "25/317 batches finished\n",
      "26/317 batches finished\n",
      "27/317 batches finished\n",
      "28/317 batches finished\n",
      "29/317 batches finished\n",
      "30/317 batches finished\n",
      "31/317 batches finished\n",
      "32/317 batches finished\n",
      "33/317 batches finished\n",
      "34/317 batches finished\n",
      "35/317 batches finished\n",
      "36/317 batches finished\n",
      "37/317 batches finished\n",
      "38/317 batches finished\n",
      "39/317 batches finished\n",
      "40/317 batches finished\n",
      "41/317 batches finished\n",
      "42/317 batches finished\n",
      "43/317 batches finished\n",
      "44/317 batches finished\n",
      "45/317 batches finished\n",
      "46/317 batches finished\n",
      "47/317 batches finished\n",
      "48/317 batches finished\n",
      "49/317 batches finished\n",
      "50/317 batches finished\n",
      "51/317 batches finished\n",
      "52/317 batches finished\n",
      "53/317 batches finished\n",
      "54/317 batches finished\n",
      "55/317 batches finished\n",
      "56/317 batches finished\n",
      "57/317 batches finished\n",
      "58/317 batches finished\n",
      "59/317 batches finished\n",
      "60/317 batches finished\n",
      "61/317 batches finished\n",
      "62/317 batches finished\n",
      "63/317 batches finished\n",
      "64/317 batches finished\n",
      "65/317 batches finished\n",
      "66/317 batches finished\n",
      "67/317 batches finished\n",
      "68/317 batches finished\n",
      "69/317 batches finished\n",
      "70/317 batches finished\n",
      "71/317 batches finished\n",
      "72/317 batches finished\n",
      "73/317 batches finished\n",
      "74/317 batches finished\n",
      "75/317 batches finished\n",
      "76/317 batches finished\n",
      "77/317 batches finished\n",
      "78/317 batches finished\n",
      "79/317 batches finished\n",
      "80/317 batches finished\n",
      "81/317 batches finished\n",
      "82/317 batches finished\n",
      "83/317 batches finished\n",
      "84/317 batches finished\n",
      "85/317 batches finished\n",
      "86/317 batches finished\n",
      "87/317 batches finished\n",
      "88/317 batches finished\n",
      "89/317 batches finished\n",
      "90/317 batches finished\n",
      "91/317 batches finished\n",
      "92/317 batches finished\n",
      "93/317 batches finished\n",
      "94/317 batches finished\n",
      "95/317 batches finished\n",
      "96/317 batches finished\n",
      "97/317 batches finished\n",
      "98/317 batches finished\n",
      "99/317 batches finished\n",
      "100/317 batches finished\n",
      "101/317 batches finished\n",
      "102/317 batches finished\n",
      "103/317 batches finished\n",
      "104/317 batches finished\n",
      "105/317 batches finished\n",
      "106/317 batches finished\n",
      "107/317 batches finished\n",
      "108/317 batches finished\n",
      "109/317 batches finished\n",
      "110/317 batches finished\n",
      "111/317 batches finished\n",
      "112/317 batches finished\n",
      "113/317 batches finished\n",
      "114/317 batches finished\n",
      "115/317 batches finished\n",
      "116/317 batches finished\n",
      "117/317 batches finished\n",
      "118/317 batches finished\n",
      "119/317 batches finished\n",
      "120/317 batches finished\n",
      "121/317 batches finished\n",
      "122/317 batches finished\n",
      "123/317 batches finished\n",
      "124/317 batches finished\n",
      "125/317 batches finished\n",
      "126/317 batches finished\n",
      "127/317 batches finished\n",
      "128/317 batches finished\n",
      "129/317 batches finished\n",
      "130/317 batches finished\n",
      "131/317 batches finished\n",
      "132/317 batches finished\n",
      "133/317 batches finished\n",
      "134/317 batches finished\n",
      "135/317 batches finished\n",
      "136/317 batches finished\n",
      "137/317 batches finished\n",
      "138/317 batches finished\n",
      "139/317 batches finished\n",
      "140/317 batches finished\n",
      "141/317 batches finished\n",
      "142/317 batches finished\n",
      "143/317 batches finished\n",
      "144/317 batches finished\n",
      "145/317 batches finished\n",
      "146/317 batches finished\n",
      "147/317 batches finished\n",
      "148/317 batches finished\n",
      "149/317 batches finished\n",
      "150/317 batches finished\n",
      "151/317 batches finished\n",
      "152/317 batches finished\n",
      "153/317 batches finished\n",
      "154/317 batches finished\n",
      "155/317 batches finished\n",
      "156/317 batches finished\n",
      "157/317 batches finished\n",
      "158/317 batches finished\n",
      "159/317 batches finished\n",
      "160/317 batches finished\n",
      "161/317 batches finished\n",
      "162/317 batches finished\n",
      "163/317 batches finished\n",
      "164/317 batches finished\n",
      "165/317 batches finished\n",
      "166/317 batches finished\n",
      "167/317 batches finished\n",
      "168/317 batches finished\n",
      "169/317 batches finished\n",
      "170/317 batches finished\n",
      "171/317 batches finished\n",
      "172/317 batches finished\n",
      "173/317 batches finished\n",
      "174/317 batches finished\n",
      "175/317 batches finished\n",
      "176/317 batches finished\n",
      "177/317 batches finished\n",
      "178/317 batches finished\n",
      "179/317 batches finished\n",
      "180/317 batches finished\n",
      "181/317 batches finished\n",
      "182/317 batches finished\n",
      "183/317 batches finished\n",
      "184/317 batches finished\n",
      "185/317 batches finished\n",
      "186/317 batches finished\n",
      "187/317 batches finished\n",
      "188/317 batches finished\n",
      "189/317 batches finished\n",
      "190/317 batches finished\n",
      "191/317 batches finished\n",
      "192/317 batches finished\n",
      "193/317 batches finished\n",
      "194/317 batches finished\n",
      "195/317 batches finished\n",
      "196/317 batches finished\n",
      "197/317 batches finished\n",
      "198/317 batches finished\n",
      "199/317 batches finished\n",
      "200/317 batches finished\n",
      "201/317 batches finished\n",
      "202/317 batches finished\n",
      "203/317 batches finished\n",
      "204/317 batches finished\n",
      "205/317 batches finished\n",
      "206/317 batches finished\n",
      "207/317 batches finished\n",
      "208/317 batches finished\n",
      "209/317 batches finished\n",
      "210/317 batches finished\n",
      "211/317 batches finished\n",
      "212/317 batches finished\n",
      "213/317 batches finished\n",
      "214/317 batches finished\n",
      "215/317 batches finished\n",
      "216/317 batches finished\n",
      "217/317 batches finished\n",
      "218/317 batches finished\n",
      "219/317 batches finished\n",
      "220/317 batches finished\n",
      "221/317 batches finished\n",
      "222/317 batches finished\n",
      "223/317 batches finished\n",
      "224/317 batches finished\n",
      "225/317 batches finished\n",
      "226/317 batches finished\n",
      "227/317 batches finished\n",
      "228/317 batches finished\n",
      "229/317 batches finished\n",
      "230/317 batches finished\n",
      "231/317 batches finished\n",
      "232/317 batches finished\n",
      "233/317 batches finished\n",
      "234/317 batches finished\n",
      "235/317 batches finished\n",
      "236/317 batches finished\n",
      "237/317 batches finished\n",
      "238/317 batches finished\n",
      "239/317 batches finished\n",
      "240/317 batches finished\n",
      "241/317 batches finished\n",
      "242/317 batches finished\n",
      "243/317 batches finished\n",
      "244/317 batches finished\n",
      "245/317 batches finished\n",
      "246/317 batches finished\n",
      "247/317 batches finished\n",
      "248/317 batches finished\n",
      "249/317 batches finished\n",
      "250/317 batches finished\n",
      "251/317 batches finished\n",
      "252/317 batches finished\n",
      "253/317 batches finished\n",
      "254/317 batches finished\n",
      "255/317 batches finished\n",
      "256/317 batches finished\n",
      "257/317 batches finished\n",
      "258/317 batches finished\n",
      "259/317 batches finished\n",
      "260/317 batches finished\n",
      "261/317 batches finished\n",
      "262/317 batches finished\n",
      "263/317 batches finished\n",
      "264/317 batches finished\n",
      "265/317 batches finished\n",
      "266/317 batches finished\n",
      "267/317 batches finished\n",
      "268/317 batches finished\n",
      "269/317 batches finished\n",
      "270/317 batches finished\n",
      "271/317 batches finished\n",
      "272/317 batches finished\n",
      "273/317 batches finished\n",
      "274/317 batches finished\n",
      "275/317 batches finished\n",
      "276/317 batches finished\n",
      "277/317 batches finished\n",
      "278/317 batches finished\n",
      "279/317 batches finished\n",
      "280/317 batches finished\n",
      "281/317 batches finished\n",
      "282/317 batches finished\n",
      "283/317 batches finished\n",
      "284/317 batches finished\n",
      "285/317 batches finished\n",
      "286/317 batches finished\n",
      "287/317 batches finished\n",
      "288/317 batches finished\n",
      "289/317 batches finished\n",
      "290/317 batches finished\n",
      "291/317 batches finished\n",
      "292/317 batches finished\n",
      "293/317 batches finished\n",
      "294/317 batches finished\n",
      "295/317 batches finished\n",
      "296/317 batches finished\n",
      "297/317 batches finished\n",
      "298/317 batches finished\n",
      "299/317 batches finished\n",
      "300/317 batches finished\n",
      "301/317 batches finished\n",
      "302/317 batches finished\n",
      "303/317 batches finished\n",
      "304/317 batches finished\n",
      "305/317 batches finished\n",
      "306/317 batches finished\n",
      "307/317 batches finished\n",
      "308/317 batches finished\n",
      "309/317 batches finished\n",
      "310/317 batches finished\n",
      "311/317 batches finished\n",
      "312/317 batches finished\n",
      "313/317 batches finished\n",
      "314/317 batches finished\n",
      "315/317 batches finished\n",
      "316/317 batches finished\n"
     ]
    }
   ],
   "source": [
    "Face_pred_large = batch_df(images_folder_large, all_images_large,256)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "id": "250eeaae",
   "metadata": {},
   "outputs": [],
   "source": [
    "Face_pred_large = Face_pred_large.reset_index().drop_duplicates(subset = 'index').set_index('index')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "8e31f06e",
   "metadata": {},
   "outputs": [],
   "source": [
    "Face_pred.to_csv('Venezia/data_storage/Face_preds.csv',sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "08ae807b",
   "metadata": {},
   "outputs": [],
   "source": [
    "Face_pred_Suzhou.to_csv('Suzhou/data_storage/Face_preds.csv',sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "d17efbc0",
   "metadata": {},
   "outputs": [],
   "source": [
    "Face_pred_Amsterdam.to_csv('Amsterdam/data_storage/Face_preds.csv',sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "id": "78a6cebc",
   "metadata": {},
   "outputs": [],
   "source": [
    "Face_pred_large.to_csv('data_storage/images/Face_preds.csv',sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "f994f2ea",
   "metadata": {},
   "outputs": [],
   "source": [
    "Face_pred= pd.read_csv('Venezia/data_storage/Face_preds.csv',sep='\\t',index_col='Unnamed: 0')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "c480f646",
   "metadata": {},
   "outputs": [
    {
     "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>Face_prob</th>\n",
       "      <th>Face_boxes</th>\n",
       "      <th>Face_num_faces</th>\n",
       "      <th>Face_area</th>\n",
       "      <th>Face_areas</th>\n",
       "      <th>Face_ratio</th>\n",
       "      <th>index</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>51870743082</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>51871789478</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>51870742942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.929874</td>\n",
       "      <td>[[ 65.960014  45.39857  112.72479  105.72964 ]]</td>\n",
       "      <td>1</td>\n",
       "      <td>2821.36900</td>\n",
       "      <td>[2821.369]</td>\n",
       "      <td>0.125394</td>\n",
       "      <td>51870380541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>51863124066</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4981</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>51233430558</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4982</th>\n",
       "      <td>0.813669</td>\n",
       "      <td>[[67.875496  9.444743 83.15715  27.281273]]</td>\n",
       "      <td>1</td>\n",
       "      <td>272.57166</td>\n",
       "      <td>[272.57166]</td>\n",
       "      <td>0.012114</td>\n",
       "      <td>51234286630</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4983</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>51234294135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4984</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>51234291635</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4985</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>51233434888</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2951 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Face_prob                                       Face_boxes  \\\n",
       "3      0.000000                                               []   \n",
       "4      0.000000                                               []   \n",
       "5      0.000000                                               []   \n",
       "6      0.929874  [[ 65.960014  45.39857  112.72479  105.72964 ]]   \n",
       "13     0.000000                                               []   \n",
       "...         ...                                              ...   \n",
       "4981   0.000000                                               []   \n",
       "4982   0.813669      [[67.875496  9.444743 83.15715  27.281273]]   \n",
       "4983   0.000000                                               []   \n",
       "4984   0.000000                                               []   \n",
       "4985   0.000000                                               []   \n",
       "\n",
       "      Face_num_faces   Face_area   Face_areas  Face_ratio        index  \n",
       "3                  0     0.00000           []    0.000000  51870743082  \n",
       "4                  0     0.00000           []    0.000000  51871789478  \n",
       "5                  0     0.00000           []    0.000000  51870742942  \n",
       "6                  1  2821.36900   [2821.369]    0.125394  51870380541  \n",
       "13                 0     0.00000           []    0.000000  51863124066  \n",
       "...              ...         ...          ...         ...          ...  \n",
       "4981               0     0.00000           []    0.000000  51233430558  \n",
       "4982               1   272.57166  [272.57166]    0.012114  51234286630  \n",
       "4983               0     0.00000           []    0.000000  51234294135  \n",
       "4984               0     0.00000           []    0.000000  51234291635  \n",
       "4985               0     0.00000           []    0.000000  51233434888  \n",
       "\n",
       "[2951 rows x 7 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "2c5356b4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    166.000000\n",
       "mean       0.930268\n",
       "std        0.098505\n",
       "min        0.700429\n",
       "25%        0.868059\n",
       "50%        0.991205\n",
       "75%        0.999570\n",
       "max        1.000000\n",
       "Name: Face_prob, dtype: float64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_prob'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "edf9e70a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    166.000000\n",
       "mean       0.076534\n",
       "std        0.184670\n",
       "min        0.003644\n",
       "25%        0.014804\n",
       "50%        0.025501\n",
       "75%        0.061156\n",
       "max        1.838786\n",
       "Name: Face_ratio, dtype: float64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_ratio'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "48511871",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    166.000000\n",
       "mean       1.349398\n",
       "std        0.785223\n",
       "min        1.000000\n",
       "25%        1.000000\n",
       "50%        1.000000\n",
       "75%        1.000000\n",
       "max        6.000000\n",
       "Name: Face_num_faces, dtype: float64"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_num_faces'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "3122b4cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "Face_pred= pd.read_csv('Suzhou/data_storage/Face_preds.csv',sep='\\t',index_col='Unnamed: 0')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "bcb5dce7",
   "metadata": {},
   "outputs": [
    {
     "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>Face_prob</th>\n",
       "      <th>Face_boxes</th>\n",
       "      <th>Face_num_faces</th>\n",
       "      <th>Face_area</th>\n",
       "      <th>Face_areas</th>\n",
       "      <th>Face_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>23903381607</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23903645987</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23919231927</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23940920237</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23952125997</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51859424958</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51859673444</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51859675789</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51859999785</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51860000170</th>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3137 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             Face_prob Face_boxes  Face_num_faces  Face_area Face_areas  \\\n",
       "23903381607        0.0         []               0        0.0         []   \n",
       "23903645987        0.0         []               0        0.0         []   \n",
       "23919231927        0.0         []               0        0.0         []   \n",
       "23940920237        0.0         []               0        0.0         []   \n",
       "23952125997        0.0         []               0        0.0         []   \n",
       "...                ...        ...             ...        ...        ...   \n",
       "51859424958        0.0         []               0        0.0         []   \n",
       "51859673444        0.0         []               0        0.0         []   \n",
       "51859675789        0.0         []               0        0.0         []   \n",
       "51859999785        0.0         []               0        0.0         []   \n",
       "51860000170        0.0         []               0        0.0         []   \n",
       "\n",
       "             Face_ratio  \n",
       "23903381607         0.0  \n",
       "23903645987         0.0  \n",
       "23919231927         0.0  \n",
       "23940920237         0.0  \n",
       "23952125997         0.0  \n",
       "...                 ...  \n",
       "51859424958         0.0  \n",
       "51859673444         0.0  \n",
       "51859675789         0.0  \n",
       "51859999785         0.0  \n",
       "51860000170         0.0  \n",
       "\n",
       "[3137 rows x 6 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "2f01ebda",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    303.000000\n",
       "mean       0.956071\n",
       "std        0.080793\n",
       "min        0.701789\n",
       "25%        0.964053\n",
       "50%        0.998158\n",
       "75%        0.999957\n",
       "max        1.000000\n",
       "Name: Face_prob, dtype: float64"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_prob'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "df5cd2c1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    303.000000\n",
       "mean       0.057062\n",
       "std        0.072769\n",
       "min        0.003033\n",
       "25%        0.012074\n",
       "50%        0.025756\n",
       "75%        0.083591\n",
       "max        0.483004\n",
       "Name: Face_ratio, dtype: float64"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_ratio'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "e2483f91",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    303.000000\n",
       "mean       1.402640\n",
       "std        0.706821\n",
       "min        1.000000\n",
       "25%        1.000000\n",
       "50%        1.000000\n",
       "75%        2.000000\n",
       "max        5.000000\n",
       "Name: Face_num_faces, dtype: float64"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_num_faces'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "c5ffd3e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "Face_pred= pd.read_csv('Amsterdam/data_storage/Face_preds.csv',sep='\\t',index_col='Unnamed: 0')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "7dd07123",
   "metadata": {},
   "outputs": [
    {
     "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>Face_prob</th>\n",
       "      <th>Face_boxes</th>\n",
       "      <th>Face_num_faces</th>\n",
       "      <th>Face_area</th>\n",
       "      <th>Face_areas</th>\n",
       "      <th>Face_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4074991040</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49286194036</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4995005091</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50996941904</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51042706771</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5335187703</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5335187791</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5683515939</th>\n",
       "      <td>0.999745</td>\n",
       "      <td>[[ 50.57038   40.098087  67.702614  61.921387]...</td>\n",
       "      <td>2</td>\n",
       "      <td>746.3048</td>\n",
       "      <td>[373.88184, 372.42297]</td>\n",
       "      <td>0.033169</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6526596205</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7296726574</th>\n",
       "      <td>0.900536</td>\n",
       "      <td>[[75.3812   46.39236  93.45019  69.09216 ]\\n [...</td>\n",
       "      <td>2</td>\n",
       "      <td>703.1636</td>\n",
       "      <td>[410.16238, 293.00122]</td>\n",
       "      <td>0.031252</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3727 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             Face_prob                                         Face_boxes  \\\n",
       "4074991040    0.000000                                                 []   \n",
       "49286194036   0.000000                                                 []   \n",
       "4995005091    0.000000                                                 []   \n",
       "50996941904   0.000000                                                 []   \n",
       "51042706771   0.000000                                                 []   \n",
       "...                ...                                                ...   \n",
       "5335187703    0.000000                                                 []   \n",
       "5335187791    0.000000                                                 []   \n",
       "5683515939    0.999745  [[ 50.57038   40.098087  67.702614  61.921387]...   \n",
       "6526596205    0.000000                                                 []   \n",
       "7296726574    0.900536  [[75.3812   46.39236  93.45019  69.09216 ]\\n [...   \n",
       "\n",
       "             Face_num_faces  Face_area              Face_areas  Face_ratio  \n",
       "4074991040                0     0.0000                      []    0.000000  \n",
       "49286194036               0     0.0000                      []    0.000000  \n",
       "4995005091                0     0.0000                      []    0.000000  \n",
       "50996941904               0     0.0000                      []    0.000000  \n",
       "51042706771               0     0.0000                      []    0.000000  \n",
       "...                     ...        ...                     ...         ...  \n",
       "5335187703                0     0.0000                      []    0.000000  \n",
       "5335187791                0     0.0000                      []    0.000000  \n",
       "5683515939                2   746.3048  [373.88184, 372.42297]    0.033169  \n",
       "6526596205                0     0.0000                      []    0.000000  \n",
       "7296726574                2   703.1636  [410.16238, 293.00122]    0.031252  \n",
       "\n",
       "[3727 rows x 6 columns]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "65ab674d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    667.000000\n",
       "mean       0.954624\n",
       "std        0.078805\n",
       "min        0.700180\n",
       "25%        0.953888\n",
       "50%        0.995914\n",
       "75%        0.999662\n",
       "max        1.000000\n",
       "Name: Face_prob, dtype: float64"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_prob'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "a7364f79",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    667.000000\n",
       "mean       0.049063\n",
       "std        0.111617\n",
       "min        0.002593\n",
       "25%        0.009706\n",
       "50%        0.019993\n",
       "75%        0.039661\n",
       "max        1.127565\n",
       "Name: Face_ratio, dtype: float64"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_ratio'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "4ea2976c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    667.000000\n",
       "mean       1.547226\n",
       "std        0.830411\n",
       "min        1.000000\n",
       "25%        1.000000\n",
       "50%        1.000000\n",
       "75%        2.000000\n",
       "max        6.000000\n",
       "Name: Face_num_faces, dtype: float64"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_num_faces'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "5efe54a2",
   "metadata": {},
   "outputs": [],
   "source": [
    "Face_pred= pd.read_csv('data_storage/images/Face_preds.csv',sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "b9964d56",
   "metadata": {},
   "outputs": [
    {
     "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>index</th>\n",
       "      <th>Face_prob</th>\n",
       "      <th>Face_boxes</th>\n",
       "      <th>Face_num_faces</th>\n",
       "      <th>Face_area</th>\n",
       "      <th>Face_areas</th>\n",
       "      <th>Face_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10003463424</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10003465315</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10003466745</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10003466824</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10003470025</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80959</th>\n",
       "      <td>9986056745</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80960</th>\n",
       "      <td>9986199265</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80961</th>\n",
       "      <td>9986220375</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80962</th>\n",
       "      <td>9986314863</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80963</th>\n",
       "      <td>999488335</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[]</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>80964 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             index  Face_prob Face_boxes  Face_num_faces  Face_area  \\\n",
       "0      10003463424        0.0         []               0        0.0   \n",
       "1      10003465315        0.0         []               0        0.0   \n",
       "2      10003466745        0.0         []               0        0.0   \n",
       "3      10003466824        0.0         []               0        0.0   \n",
       "4      10003470025        0.0         []               0        0.0   \n",
       "...            ...        ...        ...             ...        ...   \n",
       "80959   9986056745        0.0         []               0        0.0   \n",
       "80960   9986199265        0.0         []               0        0.0   \n",
       "80961   9986220375        0.0         []               0        0.0   \n",
       "80962   9986314863        0.0         []               0        0.0   \n",
       "80963    999488335        0.0         []               0        0.0   \n",
       "\n",
       "      Face_areas  Face_ratio  \n",
       "0             []         0.0  \n",
       "1             []         0.0  \n",
       "2             []         0.0  \n",
       "3             []         0.0  \n",
       "4             []         0.0  \n",
       "...          ...         ...  \n",
       "80959         []         0.0  \n",
       "80960         []         0.0  \n",
       "80961         []         0.0  \n",
       "80962         []         0.0  \n",
       "80963         []         0.0  \n",
       "\n",
       "[80964 rows x 7 columns]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "467bd2c2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    9288.000000\n",
       "mean        0.948147\n",
       "std         0.080700\n",
       "min         0.700076\n",
       "25%         0.930316\n",
       "50%         0.992544\n",
       "75%         0.999603\n",
       "max         1.000000\n",
       "Name: Face_prob, dtype: float64"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_prob'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "54b08ff0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    9288.000000\n",
       "mean        0.075740\n",
       "std         0.112140\n",
       "min         0.002026\n",
       "25%         0.017036\n",
       "50%         0.036783\n",
       "75%         0.084583\n",
       "max         1.440277\n",
       "Name: Face_ratio, dtype: float64"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_ratio'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "377bc37b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    9288.000000\n",
       "mean        1.298019\n",
       "std         0.650607\n",
       "min         1.000000\n",
       "25%         1.000000\n",
       "50%         1.000000\n",
       "75%         1.000000\n",
       "max        10.000000\n",
       "Name: Face_num_faces, dtype: float64"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Face_pred['Face_num_faces'][Face_pred['Face_prob']!=0].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e4ff0e13",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (env)",
   "language": "python",
   "name": "env"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}