{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pathlib import Path\n",
    "from time import time as now\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.cluster import MiniBatchKMeans\n",
    "from imageio import imread, imsave\n",
    "import cv2\n",
    "import dlclib\n",
    "import deeplabcut as dlc\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "videoroot = Path(\"../videos/train_2020-09-05\")\n",
    "annoroot  = Path(\"../tests/labeled-frames\")\n",
    "scaleroot = Path(\"../tests/labeled-frames-scaled\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Frame extraction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def animal_name(path):\n",
    "    name = path.stem\n",
    "    bar  = name.index(\"_\")\n",
    "    return name[:bar]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "animals = (\"S005-19\", \"S006-19\", \"MLA-041630\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_frames(path, destdir=annoroot, n_picked=20, digits=6):\n",
    "    print(f\"from {path.name}: loading...\", end=\"\", flush=True)\n",
    "    frames = np.load(str(path))[\"frames\"]\n",
    "    print(\"resizing...\", end=\"\", flush=True)\n",
    "    data = []\n",
    "    for idx in range(frames.shape[0]):\n",
    "        data.append(cv2.resize(frames[idx], (320, 240)))\n",
    "    data = np.stack(data, axis=0)\n",
    "    print(\"clustering...\", end=\"\", flush=True)\n",
    "    anno   = MiniBatchKMeans(n_clusters=n_picked).fit_predict(data.reshape((data.shape[0], -1)))\n",
    "    print(\"saving...\", end=\"\", flush=True)\n",
    "    for clu in set(anno):\n",
    "        indexes  = np.where(anno == clu)[0]\n",
    "        picked   = np.random.randint(0, len(indexes))\n",
    "        idx  = indexes[picked]\n",
    "        dest = destdir / path.stem / f\"frame{str(idx).zfill(digits)}.png\"\n",
    "        if not dest.parent.exists():\n",
    "            dest.parent.mkdir(parents=True)\n",
    "        imsave(str(dest), frames[idx])\n",
    "    print(\"done.\", flush=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "from MLA-041630_2020-09-05_run121738.npz: loading...resizing...clustering...saving...done.\n",
      "from MLA-041630_2020-09-05_run121841.npz: loading...resizing...clustering...saving...done.\n",
      "from MLA-041630_2020-09-05_run121942.npz: loading...resizing...clustering...saving...done.\n",
      "from S005-19_2020-09-05_run124403.npz: loading...resizing...clustering...saving...done.\n",
      "from S005-19_2020-09-05_run124440.npz: loading...resizing...clustering...saving...done.\n",
      "from S005-19_2020-09-05_run124520.npz: loading...resizing...clustering...saving...done.\n",
      "from S006-19_2020-09-05_run123918.npz: loading...resizing...clustering...saving...done.\n",
      "from S006-19_2020-09-05_run123959.npz: loading...resizing...clustering...saving...done.\n",
      "from S006-19_2020-09-05_run124103.npz: loading...resizing...clustering...saving...done.\n"
     ]
    }
   ],
   "source": [
    "for path in sorted(videoroot.glob(\"*.npz\")):\n",
    "    animal = animal_name(path)\n",
    "    if animal in animals:\n",
    "        extract_frames(path)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Scaling of images\n",
    "\n",
    "In 8 bit images: $[0, 255]$ to $[15, 45]$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def scale_image(img_16b, low=15, high=45):\n",
    "    \"\"\"returns a 8-bit image\"\"\"\n",
    "    conv = (((img_16b / 256) - low) / (high - low))\n",
    "    conv[conv < 0] = 0\n",
    "    conv[conv > 1] = 1\n",
    "    return (conv * 255).astype(np.uint8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "img = imread(\"../tests/labeled-frames/S005-19_2020-09-05_run124403/frame000125.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7fafd4f29390>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(scale_image(img), cmap=plt.get_cmap(\"Greys_r\"), vmin=0, vmax=255)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "for src_ani in sorted(annoroot.iterdir()):\n",
    "    if not src_ani.is_dir():\n",
    "        continue\n",
    "    dst_ani = scaleroot / src_ani.name\n",
    "    for src_img in sorted(src_ani.glob(\"*.png\")):\n",
    "        dst_img = dst_ani / src_img.name\n",
    "        if not dst_img.parent.exists():\n",
    "            dst_img.parent.mkdir(parents=True)\n",
    "        imsave(str(dst_img), scale_image(imread(str(src_img))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Annotation was performed using ImageJ"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Profiling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "configs = {\n",
    "    \"S005-19\": Path(\"../dlc_projects/S005-tips-Keisuke-2020-09-05/config.yaml\").resolve(),\n",
    "    \"S006-19\": Path(\"../dlc_projects/S006-tips-Keisuke-2020-09-07/config.yaml\").resolve(),\n",
    "    \"MLA-041630\": Path(\"../dlc_projects/MLA-041630-tips-Keisuke-2020-09-09/config.yaml\").resolve()\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "WIDTHS = (10, 80, 160, 240, 320, 400, 480, 560, 640)\n",
    "\n",
    "def profile_session(animalname, session, testdir, widths=WIDTHS, px_per_mm=8.72):\n",
    "    manualpath = Path(testdir) / \"coords.tsv\"\n",
    "    if not manualpath.exists():\n",
    "        raise FileNotFoundError(f\"'coords.tsv' not found in {testdir.name}\")\n",
    "    manual   = pd.read_csv(str(Path(testdir) / \"coords.tsv\"), sep=\"\\t\")\n",
    "    names    = []\n",
    "    subsample= []\n",
    "    xpos     = []\n",
    "    ypos     = []\n",
    "    xann     = []\n",
    "    yann     = []\n",
    "    part     = []\n",
    "    estimate = []\n",
    "    \n",
    "    shape    = None\n",
    "    buf      = None\n",
    "    reshaped = None\n",
    "    rdims    = None\n",
    "    factor   = None\n",
    "\n",
    "    resize   = cv2.resize\n",
    "    session.get_pose(np.empty((640,480,3), dtype=np.uint8)) # let GPU kernel warm-start\n",
    "    \n",
    "    for frame, imgpath in enumerate(sorted(p for p in testdir.glob(\"*.png\"))):\n",
    "        name = testdir.name + \"_\" + imgpath.stem\n",
    "        img  = imread(str(imgpath)).astype(np.uint16) # emulate input format to be uint16\n",
    "\n",
    "        if shape is None:\n",
    "            # initialize parameters\n",
    "            shape = img.shape\n",
    "            reshaped = shape + (1,)\n",
    "            buf      = np.empty(shape+(3,), dtype=np.uint8)\n",
    "            \n",
    "        buf[:] = (img/256).reshape(reshaped)\n",
    "        \n",
    "        for width in widths:\n",
    "            \n",
    "            if width != shape[1]:\n",
    "                factor = shape[1]/width\n",
    "                height = int(shape[0]/factor)\n",
    "                rdims  = (width,height)\n",
    "            else:\n",
    "                rdims  = None\n",
    "                factor = 1\n",
    "\n",
    "            start  = now()\n",
    "            if rdims is not None:\n",
    "                pose   = session.get_pose(resize(buf, rdims, interpolation=cv2.INTER_NEAREST))\n",
    "            else:\n",
    "                pose   = session.get_pose(buf)\n",
    "            end    = now()\n",
    "            for ipart in range(pose.shape[0]):\n",
    "                x_est = pose[ipart,0]*factor\n",
    "                y_est = pose[ipart,1]*factor\n",
    "                x_anno = manual.X[3*frame+ipart]\n",
    "                y_anno = manual.Y[3*frame+ipart]\n",
    "                names.append(name)\n",
    "                subsample.append(width)\n",
    "                part.append(ipart)\n",
    "                xpos.append(x_est)\n",
    "                ypos.append(y_est)\n",
    "                xann.append(x_anno)\n",
    "                yann.append(y_anno)\n",
    "                estimate.append(np.nan if ipart>0 else (end-start))\n",
    "    xpos = np.array(xpos)\n",
    "    ypos = np.array(ypos)\n",
    "    xann = np.array(xann)\n",
    "    yann = np.array(yann)\n",
    "    dev  = np.sqrt((xpos - xann)**2 + (ypos - yann)**2)\n",
    "    \n",
    "    return pd.DataFrame(data=dict(Animal=[animalname]*len(names),\n",
    "                                  Image=names,\n",
    "                                  Subsample_px=subsample,\n",
    "                                  Part=part,\n",
    "                                  X_anno=xann / px_per_mm,\n",
    "                                  Y_anno=yann / px_per_mm, \n",
    "                                  X_pred=xpos / px_per_mm, \n",
    "                                  Y_pred=ypos / px_per_mm,\n",
    "                                  Deviation=dev / px_per_mm,\n",
    "                                  Latency=estimate))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "TFSession = dlclib.estimate.TFSession"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S005-tips-Keisuke-2020-09-05/dlc-models/iteration-1/S005-tipsSep5-trainset95shuffle1/train/snapshot-1030000 for model /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S005-tips-Keisuke-2020-09-05/dlc-models/iteration-1/S005-tipsSep5-trainset95shuffle1\n",
      "num_outputs =  1\n",
      "Initializing ResNet\n",
      "INFO:tensorflow:Restoring parameters from /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S005-tips-Keisuke-2020-09-05/dlc-models/iteration-1/S005-tipsSep5-trainset95shuffle1/train/snapshot-1030000\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S005-tips-Keisuke-2020-09-05/dlc-models/iteration-1/S005-tipsSep5-trainset95shuffle1/train/snapshot-1030000\n"
     ]
    }
   ],
   "source": [
    "cfg = configs[\"S005-19\"]\n",
    "session = TFSession.from_config(cfg, locate_on_gpu=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/MLA-041630-tips-Keisuke-2020-09-09/dlc-models/iteration-1/MLA-041630-tipsSep9-trainset95shuffle1/train/snapshot-1030000 for model /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/MLA-041630-tips-Keisuke-2020-09-09/dlc-models/iteration-1/MLA-041630-tipsSep9-trainset95shuffle1\n",
      "num_outputs =  1\n",
      "Initializing ResNet\n",
      "INFO:tensorflow:Restoring parameters from /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/MLA-041630-tips-Keisuke-2020-09-09/dlc-models/iteration-1/MLA-041630-tipsSep9-trainset95shuffle1/train/snapshot-1030000\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/MLA-041630-tips-Keisuke-2020-09-09/dlc-models/iteration-1/MLA-041630-tipsSep9-trainset95shuffle1/train/snapshot-1030000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "running: MLA-041630_2020-09-05_run121738...done.\n",
      "running: MLA-041630_2020-09-05_run121841...done.\n",
      "running: MLA-041630_2020-09-05_run121942...done.\n",
      "Using /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S005-tips-Keisuke-2020-09-05/dlc-models/iteration-1/S005-tipsSep5-trainset95shuffle1/train/snapshot-1030000 for model /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S005-tips-Keisuke-2020-09-05/dlc-models/iteration-1/S005-tipsSep5-trainset95shuffle1\n",
      "num_outputs =  1\n",
      "Initializing ResNet\n",
      "INFO:tensorflow:Restoring parameters from /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S005-tips-Keisuke-2020-09-05/dlc-models/iteration-1/S005-tipsSep5-trainset95shuffle1/train/snapshot-1030000\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S005-tips-Keisuke-2020-09-05/dlc-models/iteration-1/S005-tipsSep5-trainset95shuffle1/train/snapshot-1030000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "running: S005-19_2020-09-05_run124403...done.\n",
      "running: S005-19_2020-09-05_run124440...done.\n",
      "running: S005-19_2020-09-05_run124520...done.\n",
      "Using /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S006-tips-Keisuke-2020-09-07/dlc-models/iteration-1/S006-tipsSep7-trainset95shuffle1/train/snapshot-1030000 for model /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S006-tips-Keisuke-2020-09-07/dlc-models/iteration-1/S006-tipsSep7-trainset95shuffle1\n",
      "num_outputs =  1\n",
      "Initializing ResNet\n",
      "INFO:tensorflow:Restoring parameters from /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S006-tips-Keisuke-2020-09-07/dlc-models/iteration-1/S006-tipsSep7-trainset95shuffle1/train/snapshot-1030000\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from /home/mouse/ks/Projects/whiskers-realtime/dlc_projects/S006-tips-Keisuke-2020-09-07/dlc-models/iteration-1/S006-tipsSep7-trainset95shuffle1/train/snapshot-1030000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "running: S006-19_2020-09-05_run123918...done.\n",
      "running: S006-19_2020-09-05_run123959...done.\n",
      "running: S006-19_2020-09-05_run124103...done.\n"
     ]
    }
   ],
   "source": [
    "tables = []\n",
    "dirfmt = \"{name}_*\"\n",
    "for animal in sorted(configs.keys()):\n",
    "    session = TFSession.from_config(configs[animal], locate_on_gpu=True)\n",
    "    for testdir in sorted(annoroot.glob(dirfmt.format(name=animal))):\n",
    "        print(f\"running: {testdir.name}...\", end=\"\", flush=True)\n",
    "        tables.append(profile_session(animal, session, testdir))\n",
    "        print(\"done.\")\n",
    "merged = pd.concat(tables)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Animal</th>\n",
       "      <th>Image</th>\n",
       "      <th>Subsample_px</th>\n",
       "      <th>Part</th>\n",
       "      <th>X_anno</th>\n",
       "      <th>Y_anno</th>\n",
       "      <th>X_pred</th>\n",
       "      <th>Y_pred</th>\n",
       "      <th>Deviation</th>\n",
       "      <th>Latency</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>MLA-041630</td>\n",
       "      <td>MLA-041630_2020-09-05_run121738_frame000167</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>34.403670</td>\n",
       "      <td>11.773739</td>\n",
       "      <td>26.060678</td>\n",
       "      <td>38.190525</td>\n",
       "      <td>27.702926</td>\n",
       "      <td>0.005150</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>MLA-041630</td>\n",
       "      <td>MLA-041630_2020-09-05_run121738_frame000167</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>45.871560</td>\n",
       "      <td>9.938876</td>\n",
       "      <td>14.638042</td>\n",
       "      <td>46.016737</td>\n",
       "      <td>47.719437</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MLA-041630</td>\n",
       "      <td>MLA-041630_2020-09-05_run121738_frame000167</td>\n",
       "      <td>10</td>\n",
       "      <td>2</td>\n",
       "      <td>50.688073</td>\n",
       "      <td>11.085665</td>\n",
       "      <td>24.960107</td>\n",
       "      <td>57.567238</td>\n",
       "      <td>53.126875</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>MLA-041630</td>\n",
       "      <td>MLA-041630_2020-09-05_run121738_frame000167</td>\n",
       "      <td>80</td>\n",
       "      <td>0</td>\n",
       "      <td>34.403670</td>\n",
       "      <td>11.773739</td>\n",
       "      <td>5.023658</td>\n",
       "      <td>58.184034</td>\n",
       "      <td>54.928140</td>\n",
       "      <td>0.004902</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>MLA-041630</td>\n",
       "      <td>MLA-041630_2020-09-05_run121738_frame000167</td>\n",
       "      <td>80</td>\n",
       "      <td>1</td>\n",
       "      <td>45.871560</td>\n",
       "      <td>9.938876</td>\n",
       "      <td>4.104537</td>\n",
       "      <td>58.171991</td>\n",
       "      <td>63.803742</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Animal                                        Image  Subsample_px  \\\n",
       "0  MLA-041630  MLA-041630_2020-09-05_run121738_frame000167            10   \n",
       "1  MLA-041630  MLA-041630_2020-09-05_run121738_frame000167            10   \n",
       "2  MLA-041630  MLA-041630_2020-09-05_run121738_frame000167            10   \n",
       "3  MLA-041630  MLA-041630_2020-09-05_run121738_frame000167            80   \n",
       "4  MLA-041630  MLA-041630_2020-09-05_run121738_frame000167            80   \n",
       "\n",
       "   Part     X_anno     Y_anno     X_pred     Y_pred  Deviation   Latency  \n",
       "0     0  34.403670  11.773739  26.060678  38.190525  27.702926  0.005150  \n",
       "1     1  45.871560   9.938876  14.638042  46.016737  47.719437       NaN  \n",
       "2     2  50.688073  11.085665  24.960107  57.567238  53.126875       NaN  \n",
       "3     0  34.403670  11.773739   5.023658  58.184034  54.928140  0.004902  \n",
       "4     1  45.871560   9.938876   4.104537  58.171991  63.803742       NaN  "
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "merged.to_csv(\"../profiling_data/profiling-frame-annotation.csv\", index=False, header=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>mean_ms</th>\n",
       "      <th>std_ms</th>\n",
       "      <th>median_ms</th>\n",
       "      <th>lower_ms</th>\n",
       "      <th>upper_ms</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Subsample_px</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>6.430055</td>\n",
       "      <td>1.033394</td>\n",
       "      <td>6.493092</td>\n",
       "      <td>4.347724</td>\n",
       "      <td>8.341277</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>6.273692</td>\n",
       "      <td>1.031179</td>\n",
       "      <td>6.352544</td>\n",
       "      <td>4.354221</td>\n",
       "      <td>8.283639</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>160</th>\n",
       "      <td>6.668799</td>\n",
       "      <td>0.952479</td>\n",
       "      <td>6.723762</td>\n",
       "      <td>4.855174</td>\n",
       "      <td>8.328974</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>240</th>\n",
       "      <td>7.465666</td>\n",
       "      <td>1.085976</td>\n",
       "      <td>7.472873</td>\n",
       "      <td>5.325574</td>\n",
       "      <td>9.725088</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>320</th>\n",
       "      <td>8.083224</td>\n",
       "      <td>1.438565</td>\n",
       "      <td>7.856011</td>\n",
       "      <td>6.052393</td>\n",
       "      <td>12.397236</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>400</th>\n",
       "      <td>9.866638</td>\n",
       "      <td>1.839840</td>\n",
       "      <td>9.510040</td>\n",
       "      <td>7.702416</td>\n",
       "      <td>15.293729</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>480</th>\n",
       "      <td>11.114215</td>\n",
       "      <td>1.733747</td>\n",
       "      <td>10.564446</td>\n",
       "      <td>9.446007</td>\n",
       "      <td>16.810578</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>560</th>\n",
       "      <td>12.972216</td>\n",
       "      <td>1.390637</td>\n",
       "      <td>12.578368</td>\n",
       "      <td>11.446917</td>\n",
       "      <td>17.281127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>640</th>\n",
       "      <td>15.392843</td>\n",
       "      <td>1.230318</td>\n",
       "      <td>15.159607</td>\n",
       "      <td>13.664925</td>\n",
       "      <td>18.339157</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                mean_ms    std_ms  median_ms   lower_ms   upper_ms\n",
       "Subsample_px                                                      \n",
       "10             6.430055  1.033394   6.493092   4.347724   8.341277\n",
       "80             6.273692  1.031179   6.352544   4.354221   8.283639\n",
       "160            6.668799  0.952479   6.723762   4.855174   8.328974\n",
       "240            7.465666  1.085976   7.472873   5.325574   9.725088\n",
       "320            8.083224  1.438565   7.856011   6.052393  12.397236\n",
       "400            9.866638  1.839840   9.510040   7.702416  15.293729\n",
       "480           11.114215  1.733747  10.564446   9.446007  16.810578\n",
       "560           12.972216  1.390637  12.578368  11.446917  17.281127\n",
       "640           15.392843  1.230318  15.159607  13.664925  18.339157"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "alpha = 5\n",
    "latency = merged.groupby([\"Subsample_px\"]).agg(\n",
    "    mean_ms=pd.NamedAgg(column=\"Latency\", aggfunc=lambda values: np.nanmean(values)*1000),\n",
    "    std_ms=pd.NamedAgg(column=\"Latency\", aggfunc=lambda values: np.nanstd(values)*1000),\n",
    "    median_ms=pd.NamedAgg(column=\"Latency\", aggfunc=lambda values: np.nanmedian(values)*1000),\n",
    "    lower_ms=pd.NamedAgg(column=\"Latency\", aggfunc=lambda values: np.nanpercentile(values, alpha/2)*1000),\n",
    "    upper_ms=pd.NamedAgg(column=\"Latency\", aggfunc=lambda values: np.nanpercentile(values, 100-alpha/2)*1000)\n",
    ")\n",
    "latency"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>mean_mm</th>\n",
       "      <th>std_mm</th>\n",
       "      <th>median_mm</th>\n",
       "      <th>lower_mm</th>\n",
       "      <th>upper_mm</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Subsample_px</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>40.201071</td>\n",
       "      <td>16.148102</td>\n",
       "      <td>36.776045</td>\n",
       "      <td>5.014546</td>\n",
       "      <td>68.208691</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>42.840615</td>\n",
       "      <td>15.544551</td>\n",
       "      <td>47.190147</td>\n",
       "      <td>13.330581</td>\n",
       "      <td>67.033342</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>160</th>\n",
       "      <td>21.162369</td>\n",
       "      <td>17.673171</td>\n",
       "      <td>18.205371</td>\n",
       "      <td>0.664390</td>\n",
       "      <td>55.796006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>240</th>\n",
       "      <td>1.578684</td>\n",
       "      <td>3.420325</td>\n",
       "      <td>0.649698</td>\n",
       "      <td>0.083853</td>\n",
       "      <td>9.818834</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>320</th>\n",
       "      <td>1.395509</td>\n",
       "      <td>3.936289</td>\n",
       "      <td>0.480147</td>\n",
       "      <td>0.091430</td>\n",
       "      <td>7.544099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>400</th>\n",
       "      <td>1.242824</td>\n",
       "      <td>3.708638</td>\n",
       "      <td>0.429932</td>\n",
       "      <td>0.076032</td>\n",
       "      <td>7.171122</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>480</th>\n",
       "      <td>1.361796</td>\n",
       "      <td>4.378298</td>\n",
       "      <td>0.470524</td>\n",
       "      <td>0.065180</td>\n",
       "      <td>7.351714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>560</th>\n",
       "      <td>1.578295</td>\n",
       "      <td>5.136374</td>\n",
       "      <td>0.461341</td>\n",
       "      <td>0.071649</td>\n",
       "      <td>11.141607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>640</th>\n",
       "      <td>1.621084</td>\n",
       "      <td>4.794471</td>\n",
       "      <td>0.472125</td>\n",
       "      <td>0.110215</td>\n",
       "      <td>12.599740</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                mean_mm     std_mm  median_mm   lower_mm   upper_mm\n",
       "Subsample_px                                                       \n",
       "10            40.201071  16.148102  36.776045   5.014546  68.208691\n",
       "80            42.840615  15.544551  47.190147  13.330581  67.033342\n",
       "160           21.162369  17.673171  18.205371   0.664390  55.796006\n",
       "240            1.578684   3.420325   0.649698   0.083853   9.818834\n",
       "320            1.395509   3.936289   0.480147   0.091430   7.544099\n",
       "400            1.242824   3.708638   0.429932   0.076032   7.171122\n",
       "480            1.361796   4.378298   0.470524   0.065180   7.351714\n",
       "560            1.578295   5.136374   0.461341   0.071649  11.141607\n",
       "640            1.621084   4.794471   0.472125   0.110215  12.599740"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy = merged.groupby([\"Subsample_px\"]).agg(\n",
    "    mean_mm=pd.NamedAgg(column=\"Deviation\", aggfunc=lambda values: np.nanmean(values)),\n",
    "    std_mm=pd.NamedAgg(column=\"Deviation\", aggfunc=lambda values: np.nanstd(values)),\n",
    "    median_mm=pd.NamedAgg(column=\"Deviation\", aggfunc=lambda values: np.nanmedian(values)),\n",
    "    lower_mm=pd.NamedAgg(column=\"Deviation\", aggfunc=lambda values: np.nanpercentile(values, alpha/2)),\n",
    "    upper_mm=pd.NamedAgg(column=\"Deviation\", aggfunc=lambda values: np.nanpercentile(values, 100-alpha/2))\n",
    ")\n",
    "accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "latency.to_csv(\"../profiling_data/profiling-frame-annotation-latencysummary.csv\", header=True, index=True)\n",
    "accuracy.to_csv(\"../profiling_data/profiling-frame-annotation-accuracysummary.csv\", header=True, index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Figure generation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "figdir = Path(\"../profiling_figs/profiling-DLC_2020-09-23\")\n",
    "if not figdir.exists():\n",
    "    figdir.mkdir(parents=True)\n",
    "\n",
    "def save_fig(fig, basename, dpi=600):\n",
    "    basepath = figdir / basename\n",
    "    fig.savefig(str(basepath.with_suffix(\".png\")), dpi=dpi)\n",
    "    fig.savefig(str(basepath.with_suffix(\".svg\")))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 180x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fmt   = \"o-\"\n",
    "\n",
    "fig = plt.figure(figsize=(2.5,3))\n",
    "\n",
    "plt.errorbar(accuracy.index, accuracy.median_mm, fmt=fmt, yerr=(accuracy.median_mm - accuracy.lower_mm,\n",
    "                                                              accuracy.upper_mm - accuracy.median_mm),\n",
    "             color=\"mediumvioletred\", ms=8, lw=2)\n",
    "\n",
    "plt.xlim(0, 700)\n",
    "plt.ylim(0, 70)\n",
    "plt.xticks((100, 300, 500))\n",
    "plt.yticks((0, 30, 60))\n",
    "plt.xlabel(\"Image width (px)\", fontsize=16)\n",
    "plt.ylabel(\"Accuracy (mm)\", fontsize=16)\n",
    "plt.tick_params(labelsize=16)\n",
    "for side in (\"top\", \"right\"):\n",
    "    plt.gca().spines[side].set_visible(False)\n",
    "plt.subplots_adjust(bottom=.25, left=.28, right=.96)\n",
    "save_fig(fig, \"accuracy\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 180x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(2.5,3))\n",
    "plt.errorbar(latency.index, latency.median_ms, fmt=\"o-\",\n",
    "             yerr=(latency.median_ms - latency.lower_ms,\n",
    "                   latency.upper_ms - latency.median_ms),\n",
    "             color=\"mediumvioletred\", ms=8, lw=2)\n",
    "plt.xlim(0, 700)\n",
    "plt.ylim(0, 20)\n",
    "plt.xticks((100, 300, 500))\n",
    "plt.yticks((0, 5, 10, 15))\n",
    "plt.xlabel(\"Image width (px)\", fontsize=16)\n",
    "plt.ylabel(\"Latency (ms)\", fontsize=16)\n",
    "plt.tick_params(labelsize=16)\n",
    "for side in (\"top\", \"right\"):\n",
    "    plt.gca().spines[side].set_visible(False)\n",
    "plt.subplots_adjust(bottom=.25, left=.28, right=.96)\n",
    "save_fig(fig, \"latency\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:dlc2.1] *",
   "language": "python",
   "name": "conda-env-dlc2.1-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}