{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Facial Expression Embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-04-19 12:14:54.399884: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX512F AVX512_VNNI FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "import random\n",
    "from typing import *\n",
    "import tensorflow as tf\n",
    "from pathlib import Path\n",
    "from tensorflow.keras import applications\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras import losses\n",
    "from tensorflow.keras import optimizers\n",
    "from tensorflow.keras import metrics\n",
    "from tensorflow.keras import Model\n",
    "from tensorflow.keras.applications import resnet\n",
    "import pandas as pd\n",
    "import mediapipe as mp\n",
    "import plotly.express as px\n",
    "from plotly.subplots import make_subplots\n",
    "import plotly.graph_objects as go\n",
    "import requests\n",
    "from tqdm import tqdm\n",
    "import base64\n",
    "from concurrent.futures import ThreadPoolExecutor, as_completed\n",
    "from retrying import retry\n",
    "import swifter\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset Loading"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "IMAGE_DIR = \"images\"\n",
    "TRAINING_DATASET = \"training_dataset\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "@retry(stop_max_attempt_number=3)\n",
    "def image_downloader(url: str):\n",
    "    get_name = lambda url: base64.urlsafe_b64encode(url.encode()).decode()\n",
    "    Path(IMAGE_DIR).mkdir(exist_ok=True)\n",
    "    filename = get_name(url)\n",
    "    if os.path.exists(os.path.join(IMAGE_DIR, filename)):\n",
    "        return filename\n",
    "    res = requests.get(url, timeout=10)\n",
    "    if not res.ok:\n",
    "        return None\n",
    "    with open(os.path.join(IMAGE_DIR, filename), \"wb\") as f:\n",
    "        f.write(res.content)\n",
    "    return filename\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_column_names():\n",
    "    names = []\n",
    "    for i in range(1, 4):\n",
    "        names += [\n",
    "            f\"img{i}_url\",\n",
    "            f\"img{i}_tl_col\",\n",
    "            f\"img{i}_br_col\",\n",
    "            f\"img{i}_tl_row\",\n",
    "            f\"img{i}_br_row\",\n",
    "        ]\n",
    "    names += [\"triplet_type\"]\n",
    "    for i in range(6):\n",
    "        names += [f\"annotator{i+1}_id\", f\"annotation{i+1}\"]\n",
    "    return names\n",
    "\n",
    "\n",
    "def get_local_storage_column_names():\n",
    "    names = []\n",
    "    for i in range(1, 4):\n",
    "        names += [\n",
    "            f\"img{i}_id\",\n",
    "            f\"img{i}_tl_col\",\n",
    "            f\"img{i}_br_col\",\n",
    "            f\"img{i}_tl_row\",\n",
    "            f\"img{i}_br_row\",\n",
    "        ]\n",
    "    names += [\"triplet_type\"]\n",
    "    names += [\"annotator1_id\"]\n",
    "    names += [\"annotation\"]\n",
    "    return names\n",
    "\n",
    "\n",
    "def get_label(annotations: pd.Series):\n",
    "    def mode(x):\n",
    "        s = pd.Series(x)\n",
    "        if s.value_counts(normalize=True).max() < 0.5:\n",
    "            return np.nan\n",
    "        return s.mode().at[0]\n",
    "\n",
    "    return annotations.swifter.apply(mode)\n",
    "\n",
    "\n",
    "def fecnet_dataset_loader(dataset_csv: str):\n",
    "    if isinstance(dataset_csv, bytes):\n",
    "        dataset_csv = dataset_csv.decode()\n",
    "    df = pd.read_csv(\n",
    "        dataset_csv, header=None, names=get_column_names(), nrows=10000\n",
    "    )  # TODO: remove nrows\n",
    "\n",
    "    # download images\n",
    "    df[\"img1_url\"] = df[\"img1_url\"].swifter.apply(image_downloader)\n",
    "    df[\"img2_url\"] = df[\"img2_url\"].swifter.apply(image_downloader)\n",
    "    df[\"img3_url\"] = df[\"img3_url\"].swifter.apply(image_downloader)\n",
    "    df.dropna(subset=[\"img1_url\", \"img2_url\", \"img3_url\"], inplace=True)\n",
    "\n",
    "    # determine label\n",
    "    df[\"label\"] = get_label(\n",
    "        pd.Series(df[[f\"annotation{i}\" for i in range(1, 7)]].values.tolist())\n",
    "    )\n",
    "    df.dropna(subset=[\"label\"], inplace=True)\n",
    "\n",
    "    samples = {\n",
    "        \"img1\": [],\n",
    "        \"img1_box\": [],\n",
    "        \"img2\": [],\n",
    "        \"img2_box\": [],\n",
    "        \"img3\": [],\n",
    "        \"img3_box\": [],\n",
    "    }\n",
    "\n",
    "    for _, row in df.iterrows():\n",
    "        img1_idx, img2_idx, img3_idx = 1, 2, 3\n",
    "        if row.label == 1:\n",
    "            img1_idx, img3_idx = img3_idx, img1_idx\n",
    "        elif row.label == 2:\n",
    "            img2_idx, img3_idx = img3_idx, img2_idx\n",
    "        bounding_boxes = (\n",
    "            (\n",
    "                (row[f\"img{img1_idx}_tl_col\"], row[f\"img{img1_idx}_tl_row\"]),\n",
    "                (row[f\"img{img1_idx}_br_col\"], row[f\"img{img1_idx}_br_row\"]),\n",
    "            ),\n",
    "            (\n",
    "                (row[f\"img{img2_idx}_tl_col\"], row[f\"img{img2_idx}_tl_row\"]),\n",
    "                (row[f\"img{img2_idx}_br_col\"], row[f\"img{img2_idx}_br_row\"]),\n",
    "            ),\n",
    "            (\n",
    "                (row[f\"img{img3_idx}_tl_col\"], row[f\"img{img3_idx}_tl_row\"]),\n",
    "                (row[f\"img{img3_idx}_br_col\"], row[f\"img{img3_idx}_br_row\"]),\n",
    "            ),\n",
    "        )\n",
    "        samples[\"img1\"].append(row[f\"img{img1_idx}_url\"])\n",
    "        samples[\"img1_box\"].append(bounding_boxes[0])\n",
    "        samples[\"img2\"].append(row[f\"img{img2_idx}_url\"])\n",
    "        samples[\"img2_box\"].append(bounding_boxes[1])\n",
    "        samples[\"img3\"].append(row[f\"img{img3_idx}_url\"])\n",
    "        samples[\"img3_box\"].append(bounding_boxes[2])\n",
    "    return samples\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_landmarks(image):\n",
    "    with mp.solutions.face_mesh.FaceMesh(\n",
    "        static_image_mode=True,\n",
    "        max_num_faces=1,\n",
    "        refine_landmarks=True,\n",
    "        min_detection_confidence=0.5,\n",
    "    ) as face_mesh:\n",
    "        results = face_mesh.process(image.numpy())\n",
    "        if results.multi_face_landmarks:\n",
    "            landmarks = results.multi_face_landmarks[0]\n",
    "            landmarks = np.array(\n",
    "                [[lm.x, lm.y, lm.z] for lm in landmarks.landmark], dtype=np.float32\n",
    "            )\n",
    "            landmarks = landmarks.flatten()\n",
    "        else:\n",
    "            landmarks = np.zeros(478 * 3, dtype=np.float32)\n",
    "    return landmarks\n",
    "\n",
    "\n",
    "def preprocess_image(filename: str, tl: Tuple[float, float], br: Tuple[float, float]):\n",
    "    image_string = tf.io.read_file(tf.strings.join([IMAGE_DIR, \"/\", filename]))\n",
    "    image = tf.image.decode_jpeg(image_string, channels=3)\n",
    "    image = tf.image.convert_image_dtype(image, tf.uint8)\n",
    "\n",
    "    # crop image\n",
    "    tl = tf.cast(tf.multiply(tl, tf.cast(tf.shape(image)[:2][::-1], tf.float32)), tf.int32)\n",
    "    br = tf.cast(tf.multiply(br, tf.cast(tf.shape(image)[:2][::-1], tf.float32)), tf.int32)\n",
    "    image = tf.image.crop_to_bounding_box(\n",
    "        image, tl[1], tl[0], br[1] - tl[1], br[0] - tl[0]\n",
    "    )\n",
    "\n",
    "    # extract landmarks using facemesh\n",
    "    return tf.py_function(extract_landmarks, [image], tf.float32)\n",
    "    # return image\n",
    "\n",
    "\n",
    "ImgType = Tuple[str, Tuple[float, float], Tuple[float, float]]\n",
    "\n",
    "\n",
    "def preprocess_triplets(triplet: dict):\n",
    "    anchor: ImgType = (triplet[\"img1\"], triplet[\"img1_box\"][0], triplet[\"img1_box\"][1])\n",
    "    positive: ImgType = (\n",
    "        triplet[\"img2\"],\n",
    "        triplet[\"img2_box\"][0],\n",
    "        triplet[\"img2_box\"][1],\n",
    "    )\n",
    "    negative: ImgType = (\n",
    "        triplet[\"img3\"],\n",
    "        triplet[\"img3_box\"][0],\n",
    "        triplet[\"img3_box\"][1],\n",
    "    )\n",
    "    return (\n",
    "        preprocess_image(*anchor),\n",
    "        preprocess_image(*positive),\n",
    "        preprocess_image(*negative),\n",
    "    )\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dc545e1717044afd9ddc0b32fecd72c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Pandas Apply:   0%|          | 0/10000 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0e653babd8104a72a0e4425167b44005",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Pandas Apply:   0%|          | 0/10000 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5fe7ddb8d18c41c599a69b9ca87cd8cd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Pandas Apply:   0%|          | 0/10000 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "38e2dc2d2517465f896d4fe4ca05da75",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Pandas Apply:   0%|          | 0/8143 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-04-19 13:04:21.919175: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX512F AVX512_VNNI FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n"
     ]
    }
   ],
   "source": [
    "if not TRAINING_DATASET:\n",
    "    df = fecnet_dataset_loader(\"data/faceexp-comparison-data-train-public.csv\")\n",
    "    image_count = len(df)\n",
    "\n",
    "    dataset = tf.data.Dataset.from_tensor_slices(\n",
    "        df\n",
    "    )\n",
    "\n",
    "    dataset = dataset.shuffle(buffer_size=1024)\n",
    "    dataset = dataset.map(preprocess_triplets)\n",
    "else:\n",
    "    dataset = tf.data.Dataset.load(TRAINING_DATASET)\n",
    "\n",
    "train_dataset = dataset.take(round(image_count * 0.8))\n",
    "val_dataset = dataset.skip(round(image_count * 0.8))\n",
    "\n",
    "train_dataset = train_dataset.batch(32, drop_remainder=False)\n",
    "train_dataset = train_dataset.prefetch(8)\n",
    "\n",
    "val_dataset = val_dataset.batch(32, drop_remainder=False)\n",
    "val_dataset = val_dataset.prefetch(8)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Visualisation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def visualise_face_mesh(landmarks):\n",
    "    landmarks = landmarks.reshape(-1, 3)\n",
    "    fig = go.Figure(\n",
    "        go.Mesh3d(\n",
    "            x=landmarks[:, 0],\n",
    "            y=landmarks[:, 1],\n",
    "            z=landmarks[:, 2],\n",
    "            color=\"lightpink\",\n",
    "            opacity=0.50,\n",
    "            contour=dict(  \n",
    "                color=\"grey\",\n",
    "                width=1,\n",
    "            ),\n",
    "        )\n",
    "    )\n",
    "    fig.show()\n",
    "\n",
    "def visualise_face_mesh_triplets(anchor, positive, negative):\n",
    "    fig = make_subplots(\n",
    "        rows=anchor.shape[0],\n",
    "        cols=3,\n",
    "        specs=[[{\"type\": \"surface\"}, {\"type\": \"surface\"}, {\"type\": \"surface\"}]],\n",
    "    )\n",
    "    for i in range(anchor.shape[0]):\n",
    "        fig.add_trace(\n",
    "            go.Mesh3d(\n",
    "                x=anchor[i, :, 0],\n",
    "                y=anchor[i, :, 1],\n",
    "                z=anchor[i, :, 2],\n",
    "                color=\"lightpink\",\n",
    "                opacity=0.50,\n",
    "            ),\n",
    "            row=i + 1,\n",
    "            col=1,\n",
    "        )\n",
    "        fig.add_trace(\n",
    "            go.Mesh3d(\n",
    "                x=positive[i, :, 0],\n",
    "                y=positive[i, :, 1],\n",
    "                z=positive[i, :, 2],\n",
    "                color=\"lightpink\",\n",
    "                opacity=0.50,\n",
    "            ),\n",
    "            row=i + 1,\n",
    "            col=2,\n",
    "        )\n",
    "        fig.add_trace(\n",
    "            go.Mesh3d(\n",
    "                x=negative[i, :, 0],\n",
    "                y=negative[i, :, 1],\n",
    "                z=negative[i, :, 2],\n",
    "                color=\"lightpink\",\n",
    "                opacity=0.50,\n",
    "            ),\n",
    "            row=i + 1,\n",
    "            col=3,\n",
    "        )\n",
    "    fig.show()\n",
    "\n",
    "def visualise_image(image):\n",
    "    fig = px.imshow(image)\n",
    "    fig.show()\n",
    "\n",
    "\n",
    "def visualise_triplet(anchor, positive, negative):\n",
    "    pass\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# ds_samples = dataset.as_numpy_iterator()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "plotlyServerURL": "https://plot.ly"
       },
       "data": [
        {
         "color": "lightpink",
         "contour": {
          "color": "grey",
          "width": 1
         },
         "opacity": 0.5,
         "type": "mesh3d",
         "x": [
          0.35013529658317566,
          0.3202477991580963,
          0.34950482845306396,
          0.2947392463684082,
          0.3107353150844574,
          0.30867958068847656,
          0.31206899881362915,
          0.19437134265899658,
          0.3033527135848999,
          0.29292023181915283,
          0.2628871500492096,
          0.3529404401779175,
          0.3594517111778259,
          0.3670436143875122,
          0.3770252466201782,
          0.3787930905818939,
          0.38357824087142944,
          0.3916940689086914,
          0.41902995109558105,
          0.33183610439300537,
          0.32221633195877075,
          0.1292702555656433,
          0.2556365728378296,
          0.23406024277210236,
          0.21422530710697174,
          0.19232195615768433,
          0.2719995081424713,
          0.2027609646320343,
          0.2285306602716446,
          0.18247291445732117,
          0.17337092757225037,
          0.18227389454841614,
          0.3651861548423767,
          0.18877878785133362,
          0.15935450792312622,
          0.16724026203155518,
          0.2536448836326599,
          0.3214626908302307,
          0.33914950489997864,
          0.3032814562320709,
          0.2989189028739929,
          0.32689595222473145,
          0.3215159773826599,
          0.3173191547393799,
          0.3041236400604248,
          0.29213953018188477,
          0.14093244075775146,
          0.27950426936149597,
          0.27511918544769287,
          0.2730633616447449,
          0.20276901125907898,
          0.29134082794189453,
          0.16574198007583618,
          0.14575889706611633,
          0.12337398529052734,
          0.2573159337043762,
          0.25341469049453735,
          0.29400718212127686,
          0.2752702832221985,
          0.2978423535823822,
          0.3163992166519165,
          0.31312257051467896,
          0.32013198733329773,
          0.13626331090927124,
          0.2830720543861389,
          0.2005755603313446,
          0.18996721506118774,
          0.160194993019104,
          0.13002368807792664,
          0.17717117071151733,
          0.1345946490764618,
          0.13120043277740479,
          0.3281206488609314,
          0.3140168786048889,
          0.3089359998703003,
          0.30569905042648315,
          0.3165404796600342,
          0.31785061955451965,
          0.3235434293746948,
          0.29580214619636536,
          0.33239829540252686,
          0.33826249837875366,
          0.3498963415622711,
          0.3905647397041321,
          0.36034095287323,
          0.35492366552352905,
          0.3543073832988739,
          0.3569784462451935,
          0.335256427526474,
          0.32934874296188354,
          0.32290810346603394,
          0.31925898790359497,
          0.2890661358833313,
          0.22059209644794464,
          0.3411254584789276,
          0.33027714490890503,
          0.3239477276802063,
          0.32445842027664185,
          0.2993137240409851,
          0.3205776810646057,
          0.26374733448028564,
          0.23641261458396912,
          0.2775478661060333,
          0.13223609328269958,
          0.14423814415931702,
          0.1550685167312622,
          0.33759602904319763,
          0.23853854835033417,
          0.21993188560009003,
          0.20177412033081055,
          0.2002149224281311,
          0.17113888263702393,
          0.28158074617385864,
          0.1633250117301941,
          0.2873168885707855,
          0.27693700790405273,
          0.1632269024848938,
          0.17775171995162964,
          0.1971782147884369,
          0.23263566195964813,
          0.25902259349823,
          0.2767479717731476,
          0.2962176203727722,
          0.17507284879684448,
          0.1533837616443634,
          0.32181215286254883,
          0.281211256980896,
          0.1761782169342041,
          0.28724032640457153,
          0.28236931562423706,
          0.18191057443618774,
          0.2759503126144409,
          0.24542783200740814,
          0.28246355056762695,
          0.28175079822540283,
          0.2925601601600647,
          0.322528213262558,
          0.18736642599105835,
          0.26578789949417114,
          0.14396250247955322,
          0.37482091784477234,
          0.33261752128601074,
          0.271721750497818,
          0.16378885507583618,
          0.21123595535755157,
          0.22955989837646484,
          0.31326746940612793,
          0.19698596000671387,
          0.4261090159416199,
          0.3720523715019226,
          0.35031408071517944,
          0.2751613259315491,
          0.4718414545059204,
          0.2487933337688446,
          0.2659434378147125,
          0.27677154541015625,
          0.1496465802192688,
          0.2618972361087799,
          0.2421388477087021,
          0.22194741666316986,
          0.2048594206571579,
          0.19574737548828125,
          0.14686748385429382,
          0.2006738781929016,
          0.3573332130908966,
          0.29936978220939636,
          0.29349109530448914,
          0.3287666440010071,
          0.31181642413139343,
          0.3223838806152344,
          0.3485766649246216,
          0.4094846844673157,
          0.30081480741500854,
          0.2751452326774597,
          0.2894112765789032,
          0.45546016097068787,
          0.39512237906455994,
          0.21169638633728027,
          0.3435114324092865,
          0.3384382128715515,
          0.33481577038764954,
          0.33553093671798706,
          0.36127132177352905,
          0.3200308084487915,
          0.3103123605251312,
          0.3017609119415283,
          0.2863670587539673,
          0.21239569783210754,
          0.29185211658477783,
          0.2841160297393799,
          0.2777450680732727,
          0.3287966847419739,
          0.24352115392684937,
          0.29005372524261475,
          0.3634802997112274,
          0.3102151155471802,
          0.2951728105545044,
          0.3112820088863373,
          0.2836800515651703,
          0.4415362775325775,
          0.42968249320983887,
          0.39485836029052734,
          0.30872637033462524,
          0.2746482193470001,
          0.336324006319046,
          0.23370853066444397,
          0.26550430059432983,
          0.23641857504844666,
          0.39913901686668396,
          0.2807650566101074,
          0.3073270320892334,
          0.3376671075820923,
          0.28107959032058716,
          0.22107289731502533,
          0.2685129642486572,
          0.2380293309688568,
          0.26531392335891724,
          0.2859867811203003,
          0.286306768655777,
          0.28346776962280273,
          0.28228336572647095,
          0.2581683397293091,
          0.21725869178771973,
          0.1858476996421814,
          0.16497018933296204,
          0.15712428092956543,
          0.17520159482955933,
          0.17026746273040771,
          0.18985232710838318,
          0.20713093876838684,
          0.23303402960300446,
          0.2584986686706543,
          0.2767183482646942,
          0.28761813044548035,
          0.2011682689189911,
          0.2900546193122864,
          0.286752313375473,
          0.293531596660614,
          0.31491926312446594,
          0.3009755611419678,
          0.3011189103126526,
          0.3166291415691376,
          0.32679590582847595,
          0.2874455451965332,
          0.29399558901786804,
          0.29295897483825684,
          0.19137704372406006,
          0.17200034856796265,
          0.33345794677734375,
          0.5165848731994629,
          0.36468279361724854,
          0.6297245025634766,
          0.4502667486667633,
          0.47633248567581177,
          0.5015144348144531,
          0.5343148708343506,
          0.426510751247406,
          0.460232675075531,
          0.4335858225822449,
          0.4887502193450928,
          0.5113205909729004,
          0.5653001666069031,
          0.541712760925293,
          0.5257027745246887,
          0.6740575432777405,
          0.5946303606033325,
          0.48840880393981934,
          0.38389983773231506,
          0.39036402106285095,
          0.4307761490345001,
          0.4634186625480652,
          0.42382970452308655,
          0.451711505651474,
          0.5367835164070129,
          0.3413296937942505,
          0.3359428644180298,
          0.5427399277687073,
          0.41957995295524597,
          0.4249035716056824,
          0.4272385537624359,
          0.5784129500389099,
          0.3331250250339508,
          0.4599546790122986,
          0.504022479057312,
          0.5697509050369263,
          0.3514682352542877,
          0.41314804553985596,
          0.5489563345909119,
          0.7560851573944092,
          0.41146716475486755,
          0.3917941451072693,
          0.5043962597846985,
          0.49487701058387756,
          0.5158083438873291,
          0.4358724057674408,
          0.40945297479629517,
          0.4059695899486542,
          0.4149421453475952,
          0.5406082272529602,
          0.40419265627861023,
          0.5600824356079102,
          0.5958532094955444,
          0.38704413175582886,
          0.4267544448375702,
          0.4575740098953247,
          0.40850165486335754,
          0.49946314096450806,
          0.4883684515953064,
          0.4903327226638794,
          0.3852199912071228,
          0.4472481906414032,
          0.42134445905685425,
          0.3933267295360565,
          0.45462092757225037,
          0.42959362268447876,
          0.4193694293498993,
          0.41059499979019165,
          0.4052921235561371,
          0.4550853967666626,
          0.4628187417984009,
          0.4740072786808014,
          0.4849053621292114,
          0.4906916916370392,
          0.7479798793792725,
          0.4724788963794708,
          0.4793197810649872,
          0.38570523262023926,
          0.4360443353652954,
          0.38855594396591187,
          0.4531913697719574,
          0.5007776021957397,
          0.4399547576904297,
          0.4991367757320404,
          0.47803518176078796,
          0.46327677369117737,
          0.5160042643547058,
          0.3462086021900177,
          0.34027785062789917,
          0.33941394090652466,
          0.5231478214263916,
          0.6046774387359619,
          0.4097871482372284,
          0.5468049049377441,
          0.3934779465198517,
          0.3971065282821655,
          0.6478445529937744,
          0.5864382982254028,
          0.5495205521583557,
          0.4948408305644989,
          0.4540863037109375,
          0.4216562509536743,
          0.33921191096305847,
          0.6606267094612122,
          0.5706796646118164,
          0.3441013991832733,
          0.4196983575820923,
          0.7090785503387451,
          0.39462149143218994,
          0.4492813050746918,
          0.5398746132850647,
          0.399263471364975,
          0.7585204839706421,
          0.4087865650653839,
          0.36433231830596924,
          0.6719770431518555,
          0.6992827653884888,
          0.7148271799087524,
          0.7057991027832031,
          0.6363992691040039,
          0.561163067817688,
          0.3505931496620178,
          0.45125308632850647,
          0.6255184412002563,
          0.49153614044189453,
          0.4703032672405243,
          0.49767056107521057,
          0.6768105030059814,
          0.5258975625038147,
          0.6157470941543579,
          0.6549895405769348,
          0.44951823353767395,
          0.4293901324272156,
          0.4162120819091797,
          0.5935925841331482,
          0.4273167550563812,
          0.44744372367858887,
          0.46681931614875793,
          0.4885428547859192,
          0.5046107769012451,
          0.6733250617980957,
          0.5059229731559753,
          0.4558345079421997,
          0.40657898783683777,
          0.3944751024246216,
          0.6374261975288391,
          0.6023849248886108,
          0.5109530091285706,
          0.7339633703231812,
          0.41390347480773926,
          0.3638405501842499,
          0.5724732875823975,
          0.7273882627487183,
          0.4317135810852051,
          0.4396117627620697,
          0.45061978697776794,
          0.4618709981441498,
          0.48905760049819946,
          0.47551092505455017,
          0.4825282096862793,
          0.4889715611934662,
          0.5229483246803284,
          0.628064751625061,
          0.36532044410705566,
          0.3764932453632355,
          0.3972977101802826,
          0.4693567156791687,
          0.6636986136436462,
          0.34619778394699097,
          0.517635703086853,
          0.33655333518981934,
          0.39386242628097534,
          0.47414958477020264,
          0.5699188709259033,
          0.47869187593460083,
          0.549126148223877,
          0.5422247648239136,
          0.5116172432899475,
          0.5860729217529297,
          0.49369877576828003,
          0.4207926392555237,
          0.6077792644500732,
          0.5770455002784729,
          0.5803326964378357,
          0.6877860426902771,
          0.6267993450164795,
          0.7282576560974121,
          0.5479268431663513,
          0.39167848229408264,
          0.39047372341156006,
          0.41766995191574097,
          0.3659999966621399,
          0.3878336548805237,
          0.4232195317745209,
          0.45865774154663086,
          0.4935988485813141,
          0.5233235359191895,
          0.5636805295944214,
          0.697788655757904,
          0.5514645576477051,
          0.5239049792289734,
          0.4872211813926697,
          0.45203667879104614,
          0.42333337664604187,
          0.40192869305610657,
          0.7306671142578125,
          0.42416754364967346,
          0.36220744252204895,
          0.3638482093811035,
          0.3601621687412262,
          0.370847225189209,
          0.4228134751319885,
          0.3530462980270386,
          0.35901665687561035,
          0.39753565192222595,
          0.3821548819541931,
          0.3707219064235687,
          0.5157893300056458,
          0.530518651008606,
          0.23357976973056793,
          0.2589195966720581,
          0.22795309126377106,
          0.20859573781490326,
          0.23917651176452637,
          0.47791236639022827,
          0.5038635730743408,
          0.47284260392189026,
          0.45236286520957947,
          0.4831041693687439
         ],
         "y": [
          0.6485642790794373,
          0.582582414150238,
          0.6065047979354858,
          0.5122002959251404,
          0.5590110421180725,
          0.5286349058151245,
          0.4544658958911896,
          0.48525744676589966,
          0.39900508522987366,
          0.3683531582355499,
          0.2479459047317505,
          0.6580212116241455,
          0.6688521504402161,
          0.6765074729919434,
          0.7104316353797913,
          0.7244861125946045,
          0.7416667342185974,
          0.7575659155845642,
          0.76170814037323,
          0.59522545337677,
          0.6013530492782593,
          0.425597608089447,
          0.48798638582229614,
          0.4966822564601898,
          0.5013554692268372,
          0.4969438910484314,
          0.4752363860607147,
          0.4347844421863556,
          0.42922669649124146,
          0.4446660876274109,
          0.4572390615940094,
          0.5172831416130066,
          0.8128834962844849,
          0.48265188932418823,
          0.5241737365722656,
          0.5060642957687378,
          0.5938929915428162,
          0.6523133516311646,
          0.6760266423225403,
          0.6739839911460876,
          0.6959530711174011,
          0.6877308487892151,
          0.7010895013809204,
          0.7423572540283203,
          0.5873830318450928,
          0.5658705830574036,
          0.4446147680282593,
          0.5235385298728943,
          0.5952703356742859,
          0.5830937027931213,
          0.6089307069778442,
          0.5369102358818054,
          0.4067484736442566,
          0.42263954877853394,
          0.37835127115249634,
          0.39777642488479614,
          0.4301164448261261,
          0.723823070526123,
          0.7307397723197937,
          0.604042112827301,
          0.6063739657402039,
          0.7302286624908447,
          0.7262782454490662,
          0.4103720188140869,
          0.6053493022918701,
          0.39707809686660767,
          0.3795672357082367,
          0.2993575930595398,
          0.3977091908454895,
          0.3414604067802429,
          0.4383431673049927,
          0.4369620978832245,
          0.6651723384857178,
          0.6821042895317078,
          0.6993644833564758,
          0.6067893505096436,
          0.7285173535346985,
          0.7331249117851257,
          0.7239314317703247,
          0.5944957137107849,
          0.7004430294036865,
          0.6899611949920654,
          0.6820166707038879,
          0.7664796710014343,
          0.762989342212677,
          0.7464680075645447,
          0.728485107421875,
          0.7148585319519043,
          0.7195716500282288,
          0.728119969367981,
          0.7391125559806824,
          0.7500336170196533,
          0.6750990152359009,
          0.6231774687767029,
          0.6009430885314941,
          0.720997154712677,
          0.7269667983055115,
          0.6133565902709961,
          0.6164888739585876,
          0.6099555492401123,
          0.5432831645011902,
          0.5683454275131226,
          0.5940686464309692,
          0.3359910547733307,
          0.3655206561088562,
          0.3903098702430725,
          0.7530261278152466,
          0.3707354962825775,
          0.323819637298584,
          0.2718733549118042,
          0.501727819442749,
          0.5398740768432617,
          0.46388745307922363,
          0.47178661823272705,
          0.5043274760246277,
          0.5835996866226196,
          0.5648958683013916,
          0.5535296201705933,
          0.555377721786499,
          0.5395916104316711,
          0.5199916362762451,
          0.5012259483337402,
          0.4662066102027893,
          0.612809419631958,
          0.47340840101242065,
          0.5977627635002136,
          0.5457257628440857,
          0.5228996276855469,
          0.48252996802330017,
          0.5956723093986511,
          0.4859740734100342,
          0.5682836174964905,
          0.6763687133789062,
          0.45437848567962646,
          0.5503214597702026,
          0.7772854566574097,
          0.8009825348854065,
          0.6223053932189941,
          0.7527986764907837,
          0.4779493510723114,
          0.8370722532272339,
          0.602717936038971,
          0.567912757396698,
          0.514924943447113,
          0.48645490407943726,
          0.4834015369415283,
          0.7394698262214661,
          0.6581702828407288,
          0.863107442855835,
          0.8410794734954834,
          0.8246340155601501,
          0.30597537755966187,
          0.8602319955825806,
          0.47638922929763794,
          0.4663528800010681,
          0.45911905169487,
          0.47331973910331726,
          0.45037704706192017,
          0.4539639353752136,
          0.4596148729324341,
          0.4674178659915924,
          0.47436952590942383,
          0.4708400368690491,
          0.48637503385543823,
          0.6260759830474854,
          0.6549198627471924,
          0.6005723476409912,
          0.6358640789985657,
          0.4261813163757324,
          0.801131546497345,
          0.8187321424484253,
          0.8473207950592041,
          0.7725343704223633,
          0.45150500535964966,
          0.5066806077957153,
          0.843221127986908,
          0.8550012707710266,
          0.6711223125457764,
          0.7175735831260681,
          0.7291356325149536,
          0.7442406415939331,
          0.7600929141044617,
          0.7623715996742249,
          0.7139681577682495,
          0.7152912020683289,
          0.7144668102264404,
          0.6989864110946655,
          0.6636712551116943,
          0.4834255874156952,
          0.4330372214317322,
          0.43950027227401733,
          0.7111238837242126,
          0.7192444801330566,
          0.42774632573127747,
          0.786672830581665,
          0.5033930540084839,
          0.4905381202697754,
          0.47989535331726074,
          0.5469294786453247,
          0.8185184001922607,
          0.7884063720703125,
          0.7933444976806641,
          0.7549517154693604,
          0.6199367046356201,
          0.7739682793617249,
          0.6310218572616577,
          0.6503364443778992,
          0.6738605499267578,
          0.8234170079231262,
          0.5632441639900208,
          0.775279700756073,
          0.7948195934295654,
          0.7304477691650391,
          0.6960389614105225,
          0.7385753393173218,
          0.7151064276695251,
          0.6862692832946777,
          0.5257720947265625,
          0.59209144115448,
          0.6011517643928528,
          0.5739011168479919,
          0.41765475273132324,
          0.41604849696159363,
          0.42351603507995605,
          0.4355558753013611,
          0.45135337114334106,
          0.4955683648586273,
          0.5737938284873962,
          0.5245038270950317,
          0.5251076221466064,
          0.5166606307029724,
          0.5025845170021057,
          0.4871779978275299,
          0.4740433394908905,
          0.5722470283508301,
          0.6069077849388123,
          0.5275518894195557,
          0.5889788269996643,
          0.5976216197013855,
          0.5933263301849365,
          0.6118854284286499,
          0.5983501076698303,
          0.6032703518867493,
          0.45284226536750793,
          0.4565606117248535,
          0.45945799350738525,
          0.4788980484008789,
          0.47041669487953186,
          0.5013483166694641,
          0.39130643010139465,
          0.5891731977462769,
          0.26805993914604187,
          0.4301952123641968,
          0.426037460565567,
          0.4186750650405884,
          0.3953601121902466,
          0.4293443560600281,
          0.36003631353378296,
          0.36995887756347656,
          0.35503384470939636,
          0.3571425974369049,
          0.4048507809638977,
          0.7780991792678833,
          0.38221651315689087,
          0.375266969203949,
          0.37950360774993896,
          0.5310734510421753,
          0.6355538368225098,
          0.6623284816741943,
          0.6427273750305176,
          0.6539143323898315,
          0.66261225938797,
          0.6672736406326294,
          0.6905454397201538,
          0.5768980383872986,
          0.5535698533058167,
          0.3293500542640686,
          0.48385298252105713,
          0.5550969839096069,
          0.5417131781578064,
          0.5078434348106384,
          0.5252175331115723,
          0.32459500432014465,
          0.3216426968574524,
          0.2340272068977356,
          0.37673598527908325,
          0.3847370147705078,
          0.6620195508003235,
          0.612559974193573,
          0.5734835267066956,
          0.5865715146064758,
          0.6774154901504517,
          0.6786665320396423,
          0.30282533168792725,
          0.5654998421669006,
          0.3407052159309387,
          0.3217375874519348,
          0.21293580532073975,
          0.2673113942146301,
          0.26658087968826294,
          0.31498438119888306,
          0.29271578788757324,
          0.6492006182670593,
          0.6531507968902588,
          0.6606115698814392,
          0.5798285007476807,
          0.6782470941543579,
          0.6863327622413635,
          0.6789563298225403,
          0.5687050223350525,
          0.6712284088134766,
          0.6686131358146667,
          0.6705335974693298,
          0.752515435218811,
          0.7450225949287415,
          0.729629635810852,
          0.7132340669631958,
          0.7015209794044495,
          0.6866973042488098,
          0.6911089420318604,
          0.6976304054260254,
          0.7049026489257812,
          0.6251015663146973,
          0.48012855648994446,
          0.682417094707489,
          0.6841904520988464,
          0.5980581641197205,
          0.5824921131134033,
          0.5925547480583191,
          0.49033063650131226,
          0.49561095237731934,
          0.5515713691711426,
          0.21434971690177917,
          0.2572726905345917,
          0.304477334022522,
          0.7126413583755493,
          0.3453207015991211,
          0.28395652770996094,
          0.22469562292099,
          0.4082232415676117,
          0.41267162561416626,
          0.4268799424171448,
          0.35888034105300903,
          0.473632276058197,
          0.5501964092254639,
          0.42894965410232544,
          0.4358370304107666,
          0.4552537500858307,
          0.4652335047721863,
          0.4643769860267639,
          0.4606035649776459,
          0.4534745514392853,
          0.48154738545417786,
          0.3511054515838623,
          0.591415524482727,
          0.5075382590293884,
          0.36743468046188354,
          0.4538077116012573,
          0.5523899793624878,
          0.3779571056365967,
          0.5345319509506226,
          0.5429238080978394,
          0.4188966155052185,
          0.5275136232376099,
          0.6931589245796204,
          0.7212470173835754,
          0.4792942404747009,
          0.6486352682113647,
          0.3298051953315735,
          0.8027075529098511,
          0.5974330902099609,
          0.5193739533424377,
          0.37666916847229004,
          0.4059479832649231,
          0.41304585337638855,
          0.6886794567108154,
          0.5327082872390747,
          0.8462651968002319,
          0.7969666719436646,
          0.765781819820404,
          0.4164455533027649,
          0.4175381064414978,
          0.4187663793563843,
          0.34054428339004517,
          0.4041135609149933,
          0.39580845832824707,
          0.38993334770202637,
          0.3854309320449829,
          0.3834839165210724,
          0.31048521399497986,
          0.3983473479747772,
          0.6159310936927795,
          0.5693028569221497,
          0.6192510724067688,
          0.735902726650238,
          0.7700165510177612,
          0.8287588357925415,
          0.672889232635498,
          0.41308435797691345,
          0.4854264557361603,
          0.8244576454162598,
          0.5365298986434937,
          0.6931396722793579,
          0.7011651992797852,
          0.7129989862442017,
          0.7261616587638855,
          0.7339270114898682,
          0.6727814674377441,
          0.6691363453865051,
          0.6648392677307129,
          0.6407595872879028,
          0.5561171174049377,
          0.46194371581077576,
          0.40936630964279175,
          0.40682780742645264,
          0.6750985980033875,
          0.6160704493522644,
          0.417363703250885,
          0.7550897598266602,
          0.47858819365501404,
          0.5167762041091919,
          0.7773205041885376,
          0.6955287456512451,
          0.56734299659729,
          0.7283923625946045,
          0.5500662326812744,
          0.5874913930892944,
          0.5848885178565979,
          0.8050228357315063,
          0.5252773761749268,
          0.7095112800598145,
          0.7459625601768494,
          0.6589949131011963,
          0.5777098536491394,
          0.6540405750274658,
          0.5923027992248535,
          0.6156118512153625,
          0.4961811304092407,
          0.562641441822052,
          0.5648633241653442,
          0.5505728125572205,
          0.3843691945075989,
          0.3591134548187256,
          0.3455658555030823,
          0.3407619595527649,
          0.34447282552719116,
          0.37946006655693054,
          0.4257481098175049,
          0.4193640351295471,
          0.4342425465583801,
          0.443255752325058,
          0.4459962248802185,
          0.4445638060569763,
          0.4416806399822235,
          0.4232228398323059,
          0.5716083645820618,
          0.5065553784370422,
          0.5693337321281433,
          0.5850460529327393,
          0.5736091136932373,
          0.5808728933334351,
          0.5880565643310547,
          0.5941392183303833,
          0.42205166816711426,
          0.43285754323005676,
          0.44047272205352783,
          0.3827030062675476,
          0.3635011315345764,
          0.46761175990104675,
          0.460554301738739,
          0.4486580789089203,
          0.47451990842819214,
          0.4864523410797119,
          0.39786893129348755,
          0.3909240961074829,
          0.3793075382709503,
          0.404601514339447,
          0.4163212180137634
         ],
         "z": [
          -0.09121585637331009,
          -0.13609382510185242,
          -0.08100872486829758,
          -0.07967185974121094,
          -0.14058925211429596,
          -0.12538911402225494,
          -0.04438198730349541,
          0.12718433141708374,
          -0.01621115952730179,
          -0.014007281512022018,
          0.03994101658463478,
          -0.08973724395036697,
          -0.08299174904823303,
          -0.07242238521575928,
          -0.07061827182769775,
          -0.07574466615915298,
          -0.08334501832723618,
          -0.08302035182714462,
          -0.06496825814247131,
          -0.12560433149337769,
          -0.08581506460905075,
          0.30247074365615845,
          0.07376182824373245,
          0.083957239985466,
          0.09916184097528458,
          0.13271541893482208,
          0.06719797104597092,
          0.08425556868314743,
          0.07325957715511322,
          0.100308358669281,
          0.11505148559808731,
          0.14803624153137207,
          -0.01089327409863472,
          0.1395123451948166,
          0.2902640998363495,
          0.18638788163661957,
          0.02835126221179962,
          -0.07560916990041733,
          -0.06840875744819641,
          -0.04905732348561287,
          -0.021922873333096504,
          -0.047557372599840164,
          -0.023440662771463394,
          0.017584308981895447,
          -0.1275530606508255,
          -0.13006307184696198,
          0.14056991040706635,
          0.02077520824968815,
          -0.0442156121134758,
          -0.03376942500472069,
          0.09104868769645691,
          -0.10824217647314072,
          0.0746266320347786,
          0.10780426114797592,
          0.2511725425720215,
          0.011452984996140003,
          0.07140301167964935,
          0.03631165251135826,
          0.29466816782951355,
          -0.044521864503622055,
          -0.06079620495438576,
          0.02761082537472248,
          0.015762047842144966,
          0.1285913586616516,
          -0.03347942605614662,
          0.041757576167583466,
          0.04203008487820625,
          0.12463679164648056,
          0.18168705701828003,
          0.081479512155056,
          0.17099538445472717,
          0.2325398176908493,
          -0.07438021898269653,
          -0.049488600343465805,
          -0.025240955874323845,
          -0.04457782953977585,
          0.020716380327939987,
          -0.002523308154195547,
          0.01370294764637947,
          -0.07809291779994965,
          -0.02123580500483513,
          -0.039678845554590225,
          -0.05803755298256874,
          -0.053059957921504974,
          -0.06915974617004395,
          -0.06865312159061432,
          -0.06040692701935768,
          -0.055968090891838074,
          -0.01974307931959629,
          -0.024766674265265465,
          -0.029065119102597237,
          -0.023899393156170845,
          -0.004946161527186632,
          0.3694154620170593,
          -0.09597187489271164,
          0.000012797584531654138,
          -0.003148193471133709,
          -0.06472304463386536,
          -0.02459593303501606,
          -0.06348008662462234,
          0.03622623160481453,
          0.05461885407567024,
          -0.02102426253259182,
          0.1879878044128418,
          0.13223208487033844,
          0.0874381884932518,
          -0.0054074907675385475,
          0.009369843639433384,
          0.04150300472974777,
          0.07323696464300156,
          0.11736985296010971,
          0.16815035045146942,
          0.06773202121257782,
          0.14668476581573486,
          0.008282606489956379,
          -0.071965292096138,
          0.20318123698234558,
          0.13656383752822876,
          0.09944114834070206,
          0.07116102427244186,
          0.054457973688840866,
          0.039401888847351074,
          -0.0253420602530241,
          0.18562330305576324,
          0.16724233329296112,
          -0.11923962086439133,
          0.004662771243602037,
          0.3795381486415863,
          0.03174726292490959,
          0.0036454014480113983,
          0.14731495082378387,
          -0.059637364000082016,
          0.33986517786979675,
          0.07783091813325882,
          -0.08769088238477707,
          0.13169880211353302,
          0.16876262426376343,
          0.28076156973838806,
          0.18105612695217133,
          0.27027902007102966,
          -0.00010411344555905089,
          -0.09115592390298843,
          0.01699749566614628,
          0.21980278193950653,
          0.10290740430355072,
          0.0896163135766983,
          0.003572615794837475,
          0.17875154316425323,
          -0.019024638459086418,
          0.067782461643219,
          0.11336730420589447,
          0.011494852602481842,
          -0.046217139810323715,
          0.08075834065675735,
          0.0780411809682846,
          0.07957497984170914,
          0.20064634084701538,
          0.07890882343053818,
          0.08170289546251297,
          0.09049990773200989,
          0.10355793684720993,
          0.11570071429014206,
          0.3492725193500519,
          0.11641888320446014,
          -0.07878345996141434,
          -0.026020025834441185,
          -0.053032711148262024,
          -0.06521138548851013,
          -0.020475255325436592,
          0.08677523583173752,
          0.04394897073507309,
          -0.043961118906736374,
          0.23355354368686676,
          0.0785350650548935,
          -0.03400707244873047,
          -0.07007031887769699,
          0.02114160731434822,
          0.260515421628952,
          -0.037832293659448624,
          -0.042533136904239655,
          -0.0476418137550354,
          -0.04558650404214859,
          -0.029897741973400116,
          -0.0015982742188498378,
          0.0021507989149540663,
          0.005416004452854395,
          0.019851379096508026,
          0.11185712367296219,
          -0.006754759233444929,
          0.05032063648104668,
          0.06944146007299423,
          -0.0008659547311253846,
          0.13757598400115967,
          0.008817347697913647,
          -0.019660279154777527,
          -0.0973699688911438,
          -0.05554208904504776,
          -0.06937841325998306,
          -0.03533783555030823,
          -0.077920101583004,
          -0.07143185287714005,
          -0.052467841655015945,
          0.03586135804653168,
          0.013430037535727024,
          0.008818700909614563,
          0.045550551265478134,
          0.02259337343275547,
          0.06702230125665665,
          -0.052270594984292984,
          -0.015255318954586983,
          0.0618409663438797,
          0.026437796652317047,
          0.0546395517885685,
          0.17411206662654877,
          0.08781158924102783,
          0.23339404165744781,
          0.037738751620054245,
          -0.013058220967650414,
          -0.08045770972967148,
          -0.05290402099490166,
          -0.10097605735063553,
          0.05423634871840477,
          0.06433962285518646,
          0.07907649874687195,
          0.10031574219465256,
          0.12294679135084152,
          0.16158419847488403,
          0.2897143065929413,
          0.1288442462682724,
          0.10274598747491837,
          0.08050038665533066,
          0.06610122323036194,
          0.0558309406042099,
          0.04969083517789841,
          0.3845973014831543,
          -0.042286936193704605,
          -0.057174477726221085,
          -0.10785375535488129,
          -0.1062149852514267,
          -0.09619101881980896,
          -0.03945345804095268,
          -0.11460594087839127,
          -0.088462233543396,
          0.06775468587875366,
          0.04407636076211929,
          0.02071455866098404,
          0.12645506858825684,
          0.1323736608028412,
          -0.09889035671949387,
          -0.011927238665521145,
          -0.1006326973438263,
          0.09125715494155884,
          -0.015881065279245377,
          -0.023706989362835884,
          -0.02527143992483616,
          -0.014570984058082104,
          -0.0037606628611683846,
          -0.028006983920931816,
          -0.015431695617735386,
          -0.029498718678951263,
          -0.025449901819229126,
          -0.010369119234383106,
          -0.07560326159000397,
          -0.006348117254674435,
          0.0727023333311081,
          0.007910840213298798,
          -0.06127450242638588,
          -0.09910740703344345,
          -0.08981145173311234,
          -0.09785965085029602,
          -0.08875387907028198,
          -0.08978264778852463,
          -0.07960610836744308,
          -0.06756759434938431,
          -0.1399519443511963,
          -0.14495651423931122,
          -0.0265171080827713,
          -0.037832655012607574,
          -0.10015016794204712,
          -0.09146834909915924,
          -0.05790150165557861,
          -0.12721902132034302,
          -0.04072870314121246,
          -0.03675984591245651,
          0.06175284460186958,
          -0.026320427656173706,
          -0.0008182303281500936,
          -0.06646320968866348,
          0.09695709496736526,
          -0.08549613505601883,
          -0.08123213797807693,
          -0.061432793736457825,
          -0.06261935830116272,
          -0.025581125169992447,
          -0.08813691139221191,
          -0.037224844098091125,
          -0.0395049974322319,
          0.015527662821114063,
          0.01127002015709877,
          -0.012765360996127129,
          -0.008093909360468388,
          0.03835462033748627,
          -0.09772489964962006,
          -0.09530600160360336,
          -0.08628617972135544,
          -0.07817153632640839,
          -0.06166653335094452,
          -0.07227744907140732,
          -0.06240921840071678,
          -0.11619637161493301,
          -0.07443390786647797,
          -0.07921575754880905,
          -0.08004739135503769,
          -0.07414513826370239,
          -0.09233243763446808,
          -0.09264132380485535,
          -0.08427056670188904,
          -0.0784059464931488,
          -0.07313652336597443,
          -0.07830611616373062,
          -0.08461257070302963,
          -0.08369456976652145,
          -0.08220319449901581,
          0.14837966859340668,
          -0.0668543353676796,
          -0.07041570544242859,
          -0.08487234264612198,
          -0.06803347915410995,
          -0.08313804864883423,
          -0.04192331060767174,
          -0.05135186389088631,
          -0.08034216612577438,
          0.033207640051841736,
          -0.00489874416962266,
          -0.03497942164540291,
          -0.07535680383443832,
          -0.031857021152973175,
          -0.007430019322782755,
          0.018344657495617867,
          -0.02119283191859722,
          -0.008241543546319008,
          0.0046014003455638885,
          -0.014669954776763916,
          -0.03873741626739502,
          -0.11718544363975525,
          0.003271007677540183,
          -0.02897615358233452,
          -0.041027385741472244,
          -0.03592054173350334,
          -0.026965150609612465,
          -0.021603651344776154,
          -0.045224256813526154,
          -0.013225674629211426,
          -0.00931007880717516,
          -0.12911903858184814,
          -0.05192447081208229,
          0.1523587852716446,
          -0.015929657965898514,
          -0.055493175983428955,
          -0.005606541875749826,
          -0.10552632808685303,
          0.12867996096611023,
          0.0119915921241045,
          -0.12111679464578629,
          -0.016111431643366814,
          0.018796281889081,
          0.06348752975463867,
          0.006800343282520771,
          0.06270752847194672,
          -0.07059179246425629,
          -0.09720306098461151,
          -0.05274796113371849,
          0.02344159409403801,
          -0.01917886920273304,
          -0.018526457250118256,
          -0.07142385095357895,
          -0.012590430676937103,
          -0.057341400533914566,
          -0.0291037205606699,
          -0.008531619794666767,
          -0.01137626450508833,
          -0.001132537960074842,
          0.008641003631055355,
          0.010166378691792488,
          -0.001830465393140912,
          -0.012118567712605,
          -0.01873232051730156,
          -0.020996909588575363,
          -0.018036875873804092,
          0.12730459868907928,
          -0.016207730397582054,
          -0.08448704332113266,
          -0.09237361699342728,
          -0.08928672224283218,
          -0.032831307500600815,
          -0.05074578523635864,
          -0.08214619010686874,
          0.05725763738155365,
          0.006246543489396572,
          -0.06576965004205704,
          -0.0505618155002594,
          0.05277063325047493,
          -0.07773610204458237,
          -0.08399733901023865,
          -0.09140277653932571,
          -0.09134954959154129,
          -0.08046033978462219,
          -0.07109057903289795,
          -0.07410839200019836,
          -0.07608717679977417,
          -0.07576794922351837,
          -0.04881776124238968,
          -0.03963153809309006,
          0.008178438991308212,
          0.008539117872714996,
          -0.06765521317720413,
          -0.025535406544804573,
          -0.016103779897093773,
          -0.07576365768909454,
          -0.0726347342133522,
          -0.07719781994819641,
          -0.07935601472854614,
          -0.06254159659147263,
          -0.06031176447868347,
          -0.0691242665052414,
          -0.069569431245327,
          -0.06955363601446152,
          -0.06493905931711197,
          -0.0879756361246109,
          -0.06788983196020126,
          -0.05118369683623314,
          -0.06061539798974991,
          -0.06383516639471054,
          -0.0083423787727952,
          -0.0491381473839283,
          0.03599701076745987,
          -0.07112898677587509,
          -0.0580788291990757,
          -0.12162604928016663,
          -0.10017335414886475,
          -0.13350801169872284,
          -0.0034336443059146404,
          -0.019957521930336952,
          -0.03184174746274948,
          -0.03454732522368431,
          -0.0283445343375206,
          -0.0014535668306052685,
          0.06682955473661423,
          -0.020231906324625015,
          -0.026916366070508957,
          -0.026163825765252113,
          -0.01815035007894039,
          -0.007938317954540253,
          -0.002397038508206606,
          0.15823005139827728,
          -0.08857030421495438,
          -0.08997385948896408,
          -0.13407044112682343,
          -0.12137394398450851,
          -0.12296082824468613,
          -0.07815977185964584,
          -0.12686742842197418,
          -0.098884716629982,
          0.012813888490200043,
          0.003080498892813921,
          -0.01291661337018013,
          -0.013449185527861118,
          -0.016337916254997253,
          0.09795031696557999,
          0.09795031696557999,
          0.09795031696557999,
          0.09795031696557999,
          0.09795031696557999,
          -0.008936399593949318,
          -0.008936399593949318,
          -0.008936399593949318,
          -0.008936399593949318,
          -0.008936399593949318
         ]
        }
       ],
       "layout": {
        "template": {
         "data": {
          "bar": [
           {
            "error_x": {
             "color": "#2a3f5f"
            },
            "error_y": {
             "color": "#2a3f5f"
            },
            "marker": {
             "line": {
              "color": "#E5ECF6",
              "width": 0.5
             },
             "pattern": {
              "fillmode": "overlay",
              "size": 10,
              "solidity": 0.2
             }
            },
            "type": "bar"
           }
          ],
          "barpolar": [
           {
            "marker": {
             "line": {
              "color": "#E5ECF6",
              "width": 0.5
             },
             "pattern": {
              "fillmode": "overlay",
              "size": 10,
              "solidity": 0.2
             }
            },
            "type": "barpolar"
           }
          ],
          "carpet": [
           {
            "aaxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "white",
             "linecolor": "white",
             "minorgridcolor": "white",
             "startlinecolor": "#2a3f5f"
            },
            "baxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "white",
             "linecolor": "white",
             "minorgridcolor": "white",
             "startlinecolor": "#2a3f5f"
            },
            "type": "carpet"
           }
          ],
          "choropleth": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "choropleth"
           }
          ],
          "contour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "contour"
           }
          ],
          "contourcarpet": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "contourcarpet"
           }
          ],
          "heatmap": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmap"
           }
          ],
          "heatmapgl": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmapgl"
           }
          ],
          "histogram": [
           {
            "marker": {
             "pattern": {
              "fillmode": "overlay",
              "size": 10,
              "solidity": 0.2
             }
            },
            "type": "histogram"
           }
          ],
          "histogram2d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2d"
           }
          ],
          "histogram2dcontour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2dcontour"
           }
          ],
          "mesh3d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "mesh3d"
           }
          ],
          "parcoords": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "parcoords"
           }
          ],
          "pie": [
           {
            "automargin": true,
            "type": "pie"
           }
          ],
          "scatter": [
           {
            "fillpattern": {
             "fillmode": "overlay",
             "size": 10,
             "solidity": 0.2
            },
            "type": "scatter"
           }
          ],
          "scatter3d": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter3d"
           }
          ],
          "scattercarpet": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattercarpet"
           }
          ],
          "scattergeo": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergeo"
           }
          ],
          "scattergl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergl"
           }
          ],
          "scattermapbox": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattermapbox"
           }
          ],
          "scatterpolar": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolar"
           }
          ],
          "scatterpolargl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolargl"
           }
          ],
          "scatterternary": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterternary"
           }
          ],
          "surface": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "surface"
           }
          ],
          "table": [
           {
            "cells": {
             "fill": {
              "color": "#EBF0F8"
             },
             "line": {
              "color": "white"
             }
            },
            "header": {
             "fill": {
              "color": "#C8D4E3"
             },
             "line": {
              "color": "white"
             }
            },
            "type": "table"
           }
          ]
         },
         "layout": {
          "annotationdefaults": {
           "arrowcolor": "#2a3f5f",
           "arrowhead": 0,
           "arrowwidth": 1
          },
          "autotypenumbers": "strict",
          "coloraxis": {
           "colorbar": {
            "outlinewidth": 0,
            "ticks": ""
           }
          },
          "colorscale": {
           "diverging": [
            [
             0,
             "#8e0152"
            ],
            [
             0.1,
             "#c51b7d"
            ],
            [
             0.2,
             "#de77ae"
            ],
            [
             0.3,
             "#f1b6da"
            ],
            [
             0.4,
             "#fde0ef"
            ],
            [
             0.5,
             "#f7f7f7"
            ],
            [
             0.6,
             "#e6f5d0"
            ],
            [
             0.7,
             "#b8e186"
            ],
            [
             0.8,
             "#7fbc41"
            ],
            [
             0.9,
             "#4d9221"
            ],
            [
             1,
             "#276419"
            ]
           ],
           "sequential": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ],
           "sequentialminus": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ]
          },
          "colorway": [
           "#636efa",
           "#EF553B",
           "#00cc96",
           "#ab63fa",
           "#FFA15A",
           "#19d3f3",
           "#FF6692",
           "#B6E880",
           "#FF97FF",
           "#FECB52"
          ],
          "font": {
           "color": "#2a3f5f"
          },
          "geo": {
           "bgcolor": "white",
           "lakecolor": "white",
           "landcolor": "#E5ECF6",
           "showlakes": true,
           "showland": true,
           "subunitcolor": "white"
          },
          "hoverlabel": {
           "align": "left"
          },
          "hovermode": "closest",
          "mapbox": {
           "style": "light"
          },
          "paper_bgcolor": "white",
          "plot_bgcolor": "#E5ECF6",
          "polar": {
           "angularaxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           },
           "bgcolor": "#E5ECF6",
           "radialaxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           }
          },
          "scene": {
           "xaxis": {
            "backgroundcolor": "#E5ECF6",
            "gridcolor": "white",
            "gridwidth": 2,
            "linecolor": "white",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "white"
           },
           "yaxis": {
            "backgroundcolor": "#E5ECF6",
            "gridcolor": "white",
            "gridwidth": 2,
            "linecolor": "white",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "white"
           },
           "zaxis": {
            "backgroundcolor": "#E5ECF6",
            "gridcolor": "white",
            "gridwidth": 2,
            "linecolor": "white",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "white"
           }
          },
          "shapedefaults": {
           "line": {
            "color": "#2a3f5f"
           }
          },
          "ternary": {
           "aaxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           },
           "baxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           },
           "bgcolor": "#E5ECF6",
           "caxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           }
          },
          "title": {
           "x": 0.05
          },
          "xaxis": {
           "automargin": true,
           "gridcolor": "white",
           "linecolor": "white",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "white",
           "zerolinewidth": 2
          },
          "yaxis": {
           "automargin": true,
           "gridcolor": "white",
           "linecolor": "white",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "white",
           "zerolinewidth": 2
          }
         }
        }
       }
      }
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# item = next(ds_samples)[0]\n",
    "# visualise_face_mesh(item)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Definition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_embedding_model():\n",
    "    input_layer = layers.Input(shape=(478, 3))\n",
    "    flatten = layers.Flatten()(input_layer)\n",
    "    dense1 = layers.Dense(512, activation=\"relu\")(flatten)\n",
    "    dense1 = layers.BatchNormalization()(dense1)\n",
    "    dense2 = layers.Dense(256, activation=\"relu\")(dense1)\n",
    "    dense2 = layers.BatchNormalization()(dense2)\n",
    "    output = layers.Dense(16)(dense2)\n",
    "\n",
    "    embedding = Model(input_layer, output, name=\"Embedding\")\n",
    "    return embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"Embedding\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_1 (InputLayer)        [(None, 478, 3)]          0         \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 1434)              0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 512)               734720    \n",
      "                                                                 \n",
      " batch_normalization (BatchN  (None, 512)              2048      \n",
      " ormalization)                                                   \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 256)               131328    \n",
      "                                                                 \n",
      " batch_normalization_1 (Batc  (None, 256)              1024      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 16)                4112      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 873,232\n",
      "Trainable params: 871,696\n",
      "Non-trainable params: 1,536\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "embedding = create_embedding_model()\n",
    "embedding.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DistanceLayer(layers.Layer):\n",
    "    def __init__(self, **kwargs):\n",
    "        super().__init__(**kwargs)\n",
    "\n",
    "    def call(self, anchor, positive, negative):\n",
    "        ap_distance = tf.reduce_sum(tf.square(anchor - positive), -1)\n",
    "        an_distance = tf.reduce_sum(tf.square(anchor - negative), -1)\n",
    "        return (ap_distance, an_distance)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "anchor_input = layers.Input(name=\"anchor\", shape=(478,) + (3,))\n",
    "positive_input = layers.Input(name=\"positive\", shape=(478,) + (3,))\n",
    "negative_input = layers.Input(name=\"negative\", shape=(478,) + (3,))\n",
    "\n",
    "distances = DistanceLayer()(\n",
    "    embedding(anchor_input),\n",
    "    embedding(positive_input),\n",
    "    embedding(negative_input),\n",
    ")\n",
    "\n",
    "siamese_network = Model(\n",
    "    inputs=[anchor_input, positive_input, negative_input], outputs=distances\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "class SiameseModel(Model):\n",
    "    \"\"\"The Siamese Network model with a custom training and testing loops.\n",
    "\n",
    "    Computes the triplet loss using the three embeddings produced by the\n",
    "    Siamese Network.\n",
    "\n",
    "    The triplet loss is defined as:\n",
    "       L(A, P, N) = max(‖f(A) - f(P)‖² - ‖f(A) - f(N)‖² + margin, 0)\n",
    "    \"\"\"\n",
    "\n",
    "    def __init__(self, siamese_network, margin=0.2):\n",
    "        super().__init__()\n",
    "        self.siamese_network = siamese_network\n",
    "        self.margin = margin\n",
    "        self.loss_tracker = metrics.Mean(name=\"loss\")\n",
    "        self.acc_tracker = metrics.Mean(name=\"accuracy\")\n",
    "\n",
    "    def call(self, inputs):\n",
    "        return self.siamese_network(inputs)\n",
    "\n",
    "    def train_step(self, data):\n",
    "        # GradientTape is a context manager that records every operation that\n",
    "        # you do inside. We are using it here to compute the loss so we can get\n",
    "        # the gradients and apply them using the optimizer specified in\n",
    "        # `compile()`.\n",
    "        with tf.GradientTape() as tape:\n",
    "            loss = self._compute_loss(data)\n",
    "\n",
    "        # Storing the gradients of the loss function with respect to the\n",
    "        # weights/parameters.\n",
    "        gradients = tape.gradient(loss, self.siamese_network.trainable_weights)\n",
    "\n",
    "        # Applying the gradients on the model using the specified optimizer\n",
    "        self.optimizer.apply_gradients(\n",
    "            zip(gradients, self.siamese_network.trainable_weights)\n",
    "        )\n",
    "\n",
    "        # Let's update and return the training loss metric.\n",
    "        self.loss_tracker.update_state(loss)\n",
    "        return {\"loss\": self.loss_tracker.result()}\n",
    "\n",
    "    def test_step(self, data):\n",
    "        loss = self._compute_loss(data)\n",
    "\n",
    "        # Let's update and return the loss metric.\n",
    "        self.loss_tracker.update_state(loss)\n",
    "        self.acc_tracker.update_state(\n",
    "            self._compute_accuracy(data)\n",
    "        )\n",
    "        return {\"loss\": self.loss_tracker.result(), \"accuracy\": self.acc_tracker.result()}\n",
    "\n",
    "    def _compute_loss(self, data):\n",
    "        # The output of the network is a tuple containing the distances\n",
    "        # between the anchor and the positive example, and the anchor and\n",
    "        # the negative example.\n",
    "        ap_distance, an_distance = self.siamese_network(data)\n",
    "\n",
    "        # Computing the Triplet Loss by subtracting both distances and\n",
    "        # making sure we don't get a negative value.\n",
    "        loss = ap_distance - an_distance\n",
    "        loss = tf.maximum(loss + self.margin, 0.0)\n",
    "        return loss\n",
    "    \n",
    "    def _compute_accuracy(self, data):\n",
    "        ap_distance, an_distance = self.siamese_network(data)\n",
    "        return tf.cast(ap_distance < an_distance, tf.float32)\n",
    "\n",
    "    @property\n",
    "    def metrics(self):\n",
    "        # We need to list our metrics here so the `reset_states()` can be\n",
    "        # called automatically.\n",
    "        return [self.loss_tracker, self.acc_tracker]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Fitting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "1/1 [==============================] - 1010s 1010s/step - loss: 0.6925 - val_loss: 0.4313 - val_accuracy: 0.5041\n",
      "Epoch 2/40\n",
      "1/1 [==============================] - ETA: 0s - loss: 0.3325"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m/var/folders/9f/9272h8p951zfstxdwzn0cj2h0000gn/T/ipykernel_61590/1150799400.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0msiamese_model\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mSiameseModel\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msiamese_network\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0msiamese_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcompile\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moptimizer\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0moptimizers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mAdam\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0.0005\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweighted_metrics\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"accuracy\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0msiamese_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_dataset\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m40\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalidation_data\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mval_dataset\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     63\u001b[0m         \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     64\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 65\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     66\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     67\u001b[0m             \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1703\u001b[0m                         \u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1704\u001b[0m                         \u001b[0mreturn_dict\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1705\u001b[0;31m                         \u001b[0m_use_cached_eval_dataset\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1706\u001b[0m                     )\n\u001b[1;32m   1707\u001b[0m                     val_logs = {\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     63\u001b[0m         \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     64\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 65\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     66\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     67\u001b[0m             \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mevaluate\u001b[0;34m(self, x, y, batch_size, verbose, sample_weight, steps, callbacks, max_queue_size, workers, use_multiprocessing, return_dict, **kwargs)\u001b[0m\n\u001b[1;32m   2038\u001b[0m                         ):\n\u001b[1;32m   2039\u001b[0m                             \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_test_batch_begin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2040\u001b[0;31m                             \u001b[0mtmp_logs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtest_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0miterator\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2041\u001b[0m                             \u001b[0;32mif\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshould_sync\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2042\u001b[0m                                 \u001b[0mcontext\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masync_wait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/tensorflow/python/util/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    148\u001b[0m     \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    149\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 150\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    151\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    152\u001b[0m       \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    878\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    879\u001b[0m       \u001b[0;32mwith\u001b[0m \u001b[0mOptionalXlaContext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_jit_compile\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 880\u001b[0;31m         \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    881\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    882\u001b[0m       \u001b[0mnew_tracing_count\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexperimental_get_tracing_count\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py\u001b[0m in \u001b[0;36m_call\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    917\u001b[0m       \u001b[0;31m# In this case we have not created variables on the first call. So we can\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    918\u001b[0m       \u001b[0;31m# run the first trace but we should fail if variables are created.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 919\u001b[0;31m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_variable_creation_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    920\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_created_variables\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mALLOW_DYNAMIC_VARIABLE_CREATION\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    921\u001b[0m         raise ValueError(\"Creating variables on a non-first call to a function\"\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compiler.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m    133\u001b[0m        filtered_flat_args) = self._maybe_define_function(args, kwargs)\n\u001b[1;32m    134\u001b[0m     return concrete_function._call_flat(\n\u001b[0;32m--> 135\u001b[0;31m         filtered_flat_args, captured_inputs=concrete_function.captured_inputs)  # pylint: disable=protected-access\n\u001b[0m\u001b[1;32m    136\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    137\u001b[0m   \u001b[0;34m@\u001b[0m\u001b[0mproperty\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/tensorflow/python/eager/polymorphic_function/monomorphic_function.py\u001b[0m in \u001b[0;36m_call_flat\u001b[0;34m(self, args, captured_inputs, cancellation_manager)\u001b[0m\n\u001b[1;32m   1744\u001b[0m       \u001b[0;31m# No tape is watching; skip to running the function.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1745\u001b[0m       return self._build_call_outputs(self._inference_function.call(\n\u001b[0;32m-> 1746\u001b[0;31m           ctx, args, cancellation_manager=cancellation_manager))\n\u001b[0m\u001b[1;32m   1747\u001b[0m     forward_backward = self._select_forward_and_backward_functions(\n\u001b[1;32m   1748\u001b[0m         \u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/tensorflow/python/eager/polymorphic_function/monomorphic_function.py\u001b[0m in \u001b[0;36mcall\u001b[0;34m(self, ctx, args, cancellation_manager)\u001b[0m\n\u001b[1;32m    381\u001b[0m               \u001b[0minputs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    382\u001b[0m               \u001b[0mattrs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mattrs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 383\u001b[0;31m               ctx=ctx)\n\u001b[0m\u001b[1;32m    384\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    385\u001b[0m           outputs = execute.execute_with_cancellation(\n",
      "\u001b[0;32m~/Git/FacialExpressionSyncService/.venv/lib/python3.7/site-packages/tensorflow/python/eager/execute.py\u001b[0m in \u001b[0;36mquick_execute\u001b[0;34m(op_name, num_outputs, inputs, attrs, ctx, name)\u001b[0m\n\u001b[1;32m     51\u001b[0m     \u001b[0mctx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mensure_initialized\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     52\u001b[0m     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,\n\u001b[0;32m---> 53\u001b[0;31m                                         inputs, attrs, num_outputs)\n\u001b[0m\u001b[1;32m     54\u001b[0m   \u001b[0;32mexcept\u001b[0m \u001b[0mcore\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_NotOkStatusException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     55\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "siamese_model = SiameseModel(siamese_network)\n",
    "siamese_model.compile(optimizer=optimizers.Adam(0.0005), weighted_metrics=[\"accuracy\"])\n",
    "siamese_model.fit(train_dataset, epochs=40, validation_data=val_dataset)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluate Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_df = fecnet_dataset_loader(\"data/faceexp-comparison-data-test-public.csv\")\n",
    "\n",
    "test_dataset = tf.data.Dataset.from_tensor_slices(\n",
    "    test_df\n",
    ")\n",
    "\n",
    "test_dataset = test_dataset.map(preprocess_triplets)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "siamese_model.evaluate(test_dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Positive similarity: 0.96082336\n",
      "Negative similarity 0.85784876\n",
      "Positive-Negative similarity 0.8386482\n"
     ]
    }
   ],
   "source": [
    "sample = next(iter(train_dataset))\n",
    "# visualise_face_mesh_triplet(*sample)\n",
    "\n",
    "anchor, positive, negative = sample\n",
    "anchor_embedding, positive_embedding, negative_embedding = (\n",
    "    embedding(tf.reshape(anchor, (-1, 478,3))),\n",
    "    embedding(tf.reshape(positive, (-1, 478,3))),\n",
    "    embedding(tf.reshape(negative, (-1, 478,3))),\n",
    ")\n",
    "cosine_similarity = metrics.CosineSimilarity()\n",
    "\n",
    "positive_similarity = cosine_similarity(anchor_embedding, positive_embedding)\n",
    "print(\"Positive similarity:\", positive_similarity.numpy())\n",
    "\n",
    "negative_similarity = cosine_similarity(anchor_embedding, negative_embedding)\n",
    "print(\"Negative similarity\", negative_similarity.numpy())\n",
    "\n",
    "positive_negative_similarity = cosine_similarity(positive_embedding, negative_embedding)\n",
    "print(\"Positive-Negative similarity\", positive_negative_similarity.numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO: Created TensorFlow Lite XNNPACK delegate for CPU.\n"
     ]
    }
   ],
   "source": [
    "# tf.data.Dataset.save(dataset, \"training_dataset\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".venv",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.9"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}