pipeline.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Demo Notebook\n",
    "\n",
    "This notebook aims to represent the DeepLabCut output analysis, using example data. \n",
    "So we begin by importing our packages, called for now \"dlc\". "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "import dlc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Loading data\n",
    "As an example, we will first work on a single file. Later, we'll try batch processing. We define the path where our file is, and with the function\n",
    "`dlc.load_data.read_data` we will load our `.h5` file. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "ename": "SyntaxError",
     "evalue": "invalid syntax (1281446189.py, line 2)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;36m  Cell \u001b[0;32mIn[10], line 2\u001b[0;36m\u001b[0m\n\u001b[0;31m    data = dlc.load_data.read_data( ,'/Users/annateruel/ca2+img-anna-2023-01-30/videos/0_neutro2DLC_resnet50_ca2+imgJan30shuffle1_500000.h5' 5, 30)\u001b[0m\n\u001b[0m                                    ^\u001b[0m\n\u001b[0;31mSyntaxError\u001b[0m\u001b[0;31m:\u001b[0m invalid syntax\n"
     ]
    }
   ],
   "source": [
    "h5_path =  '/Users/annateruel/ca2+img-anna-2023-01-30/videos'\n",
    "data = dlc.load_data.read_data( ,'/Users/annateruel/ca2+img-anna-2023-01-30/videos/0_neutro2DLC_resnet50_ca2+imgJan30shuffle1_500000.h5' 5, 30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since on this data file we have performed a tracking on the cage, we will define a list of bodyparts we want to work with. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "bodyparts = ['nose', 'rightear', 'leftear', 'head', 'sp1', 'sp2', 'tail']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd \n",
    "data = pd.read_hdf('/Users/annateruel/ca2+img-anna-2023-01-30/videos/0_neutro2DLC_resnet50_ca2+imgJan30shuffle1_500000.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2 = data.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Get centroid from a list of bodyparts\n",
    "\n",
    "One of the functions we can perform from the `data.py` file is to calculate the centroid of specified bodyparts in each dataframe. We can add this centroid coordinates (x,y) to our current dataframe. Let's try that!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Calculate interpolation\n",
    "\n",
    "If the tracking of some bodyparts is not perfect, we can get the interpolation of those values with likelihood under a certain threshold. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2 = data.copy() #we get a copy of the dataframe and work from there\n",
    "df2 = dlc.data.get_interpolation()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "DLCscorer = df2.columns[0][0]\n",
    "threshold = 0.95\n",
    "likelihood_col = (DLCscorer, 'nose', 'likelihood')\n",
    "if df2[likelihood_col].lt(threshold).any():\n",
    "    mask = df2[likelihood_col] < threshold\n",
    "\n",
    "    df2.loc[mask, (DLCscorer, 'nose', 'x')] = np.nan\n",
    "    df2.loc[mask, (DLCscorer, 'nose', 'y')] = np.nan\n",
    "\n",
    "    df2[(DLCscorer, 'nose', 'x')] = df2[(DLCscorer, 'nose', 'x')].interpolate(method='linear')\n",
    "    df2[(DLCscorer, 'nose', 'y')] = df2[(DLCscorer, 'nose', 'y')].interpolate(method='linear')\n",
    "\n",
    "nose2 = df2.loc[:,(DLCscorer, 'nose', slice(None))]\n",
    "nose2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "nose = data.loc[:, (slice(None), 'nose', slice(None))]\n",
    "nose.loc[:,(slice(None), slice(None), 'x')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.collections import LineCollection\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10, 10))\n",
    "\n",
    "x = nose.loc[:, (slice(None), slice(None), 'x')].values\n",
    "y = nose.loc[:, (slice(None), slice(None), 'y')].values\n",
    "likelihood = nose.loc[:, (slice(None), slice(None), 'likelihood')].values.flatten()  # Flatten the array\n",
    "\n",
    "# Create segments for LineCollection\n",
    "points = np.array([x, y]).T.reshape(-1, 1, 2)\n",
    "segments = np.concatenate([points[:-1], points[1:]], axis=1)\n",
    "\n",
    "# Create a LineCollection object with segments and colormap\n",
    "colormap = plt.cm.RdYlGn\n",
    "lc = LineCollection(segments, cmap=colormap, norm=plt.Normalize(likelihood.min(), likelihood.max()), linewidth=0.5, alpha=0.5)\n",
    "lc.set_array(likelihood)\n",
    "\n",
    "ax.add_collection(lc)\n",
    "\n",
    "# Add a colorbar to the plot to show what each color represents\n",
    "cbar = plt.colorbar(lc, ax=ax)\n",
    "cbar.set_label('Likelihood Value')\n",
    "\n",
    "ax.set_xlim(x.min(), x.max())\n",
    "ax.set_ylim(y.min(), y.max())\n",
    "\n",
    "plt.show()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.collections import LineCollection\n",
    "\n",
    "def plot_data(ax, data, title):\n",
    "    x = data.loc[:, (slice(None), slice(None), 'x')].values\n",
    "    y = data.loc[:, (slice(None), slice(None), 'y')].values\n",
    "    likelihood = data.loc[:, (slice(None), slice(None), 'likelihood')].values.flatten()\n",
    "\n",
    "    # Create segments for LineCollection\n",
    "    points = np.array([x, y]).T.reshape(-1, 1, 2)\n",
    "    segments = np.concatenate([points[:-1], points[1:]], axis=1)\n",
    "\n",
    "    # Create a LineCollection object with segments and colormap\n",
    "    colormap = plt.cm.RdYlGn\n",
    "    lc = LineCollection(segments, cmap=colormap, norm=plt.Normalize(likelihood.min(), likelihood.max()), linewidth=0.5, alpha=0.5)\n",
    "    lc.set_array(likelihood)\n",
    "\n",
    "    ax.add_collection(lc)\n",
    "    ax.set_xlim(x.min(), x.max())\n",
    "    ax.set_ylim(y.min(), y.max())\n",
    "    ax.set_title(title)\n",
    "    \n",
    "    return lc\n",
    "\n",
    "fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(20, 10))\n",
    "\n",
    "lc1 = plot_data(ax1, nose, \"Before Interpolation\")\n",
    "lc2 = plot_data(ax2, nose2, \"After Interpolation\")\n",
    "\n",
    "cbar_ax = fig.add_axes([0.92, 0.15, 0.02, 0.7])\n",
    "cbar = fig.colorbar(lc2, cax=cbar_ax)\n",
    "cbar.set_label('Likelihood Value')\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(20, 10))\n",
    "lc1 = plot_data(ax1, nose, \"Before Interpolation\")\n",
    "lc2 = plot_data(ax2, nose2, \"After Interpolation\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ROI Drawer\n",
    "\n",
    "Testing ROI Drawer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "from dlc.analysis.time_roi import ROIDrawer\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Drawing ROIs for video: /Users/annateruel/Desktop/videos/7_WIN_20231010_094516_1.mp4\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROIs saved to /Users/annateruel/Desktop/videos/7_WIN_20231010_094516_1_roi.h5\n",
      "Frame saved to /Users/annateruel/Desktop/videos/7_WIN_20231010_094516_1_roi.png\n"
     ]
    }
   ],
   "source": [
    "roi_drawer = ROIDrawer(video_path='/Users/annateruel/Desktop/videos/7_WIN_20231010_094516_1.mp4',save_dir='/Users/annateruel/Desktop/videos/', num_rois=4)\n",
    "r = roi_drawer.draw_rois()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Testing time in ROI\n",
    "\n",
    "1. Define ROIs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import dlc.analysis.time_roi as time_roi\n",
    "import pandas as pd\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processing group:    index   shape-type  vertex-index      axis-0      axis-1\n",
      "0      0  add_polygon             0   70.742562   25.991654\n",
      "1      0  add_polygon             1   93.744196   10.990588\n",
      "2      0  add_polygon             2  178.250200  117.998191\n",
      "3      0  add_polygon             3  155.748601  136.499505\n",
      "Processing group:    index   shape-type  vertex-index      axis-0      axis-1\n",
      "4      1  add_polygon             0  187.250839  175.002241\n",
      "5      1  add_polygon             1  209.252403  157.500997\n",
      "6      1  add_polygon             2  290.258158  264.508600\n",
      "7      1  add_polygon             3  268.256595  280.009701\n",
      "Processing group:     index   shape-type  vertex-index      axis-0      axis-1\n",
      "8       2  add_polygon             0  156.248637  137.999612\n",
      "9       2  add_polygon             1  187.250839  176.002312\n",
      "10      2  add_polygon             2   84.743556  253.507819\n",
      "11      2  add_polygon             3   51.741212  215.005083\n",
      "Processing group:     index   shape-type  vertex-index      axis-0      axis-1\n",
      "12      3  add_polygon             0  182.250484  116.998120\n",
      "13      3  add_polygon             1  290.258158   33.492187\n",
      "14      3  add_polygon             2  319.260219   73.995064\n",
      "15      3  add_polygon             3  212.752651  156.000891\n",
      "Extracted polygons: [<dlc.analysis.time_roi.PolygonROI object at 0x16ac9bc40>, <dlc.analysis.time_roi.PolygonROI object at 0x16a90d240>, <dlc.analysis.time_roi.PolygonROI object at 0x169f181c0>, <dlc.analysis.time_roi.PolygonROI object at 0x2834cbdf0>]\n"
     ]
    }
   ],
   "source": [
    "r = pd.read_hdf('/Users/annateruel/Desktop/videos/11_WIN_20231010_111425_1_roi.h5')\n",
    "roi_groups = r.groupby(['index', 'shape-type'])\n",
    "polygons, _ = time_roi.PolygonROI.extract_polygons(roi_groups)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "2. Get time (s) in each ROI for a given dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "time_in_roi = time_roi.TimeinRoi(fps=15)\n",
    "for poly in polygons:\n",
    "    time_in_roi.add_roi(poly)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "file_path = '/Users/annateruel/Desktop/videos/11_WIN_20231010_111425_1DLC_resnet50_capsaicin_malesOct20shuffle1_200000_filtered.h5'\n",
    "directory = '/Users/annateruel/Desktop/videos/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "scorer = 'DLC_resnet50_capsaicin_malesOct20shuffle1_200000'\n",
    "bodypart = 'Neck'\n",
    "\n",
    "tracking_data = time_in_roi.extract_tracking_data(file_path, scorer, bodypart)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[160.2, 134.26666666666668, 83.66666666666667, 82.86666666666666]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "time_spent = time_in_roi.time_in_rois(tracking_data)\n",
    "time_spent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "time_spent = time_in_roi.time_in_rois_dir(directory, scorer, bodypart)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "video_dir = '/Users/annateruel/Desktop/videos/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processing ROI file: 7_WIN_20231010_094516_1_roi.h5\n",
      "Processing group:    index   shape-type  vertex-index      axis-0      axis-1\n",
      "0      0  add_polygon             0   79.743201   25.491618\n",
      "1      0  add_polygon             1  103.744906    8.490410\n",
      "2      0  add_polygon             2  185.250697  116.498084\n",
      "3      0  add_polygon             3  160.248921  133.499292\n",
      "Processing group:    index   shape-type  vertex-index      axis-0      axis-1\n",
      "4      1  add_polygon             0  189.250982  171.501992\n",
      "5      1  add_polygon             1  214.252758  155.000820\n",
      "6      1  add_polygon             2  292.758336  260.508316\n",
      "7      1  add_polygon             3  268.256595  276.009417\n",
      "Processing group:     index   shape-type  vertex-index      axis-0      axis-1\n",
      "8       2  add_polygon             0  159.748885  135.999470\n",
      "9       2  add_polygon             1  189.250982  173.002099\n",
      "10      2  add_polygon             2   88.743841  246.507321\n",
      "11      2  add_polygon             3   57.741638  209.504692\n",
      "Processing group:     index   shape-type  vertex-index      axis-0      axis-1\n",
      "12      3  add_polygon             0  188.750946  114.497942\n",
      "13      3  add_polygon             1  292.758336   36.492400\n",
      "14      3  add_polygon             2  323.760538   76.995278\n",
      "15      3  add_polygon             3  216.752935  153.500713\n",
      "Extracted polygons: [<dlc.analysis.time_roi.PolygonROI object at 0x16a8f4a00>, <dlc.analysis.time_roi.PolygonROI object at 0x16a8f4970>, <dlc.analysis.time_roi.PolygonROI object at 0x16a8f4eb0>, <dlc.analysis.time_roi.PolygonROI object at 0x16a8f4d90>]\n",
      "Extracted polygons from 7_WIN_20231010_094516_1_roi.h5: 4\n",
      "Processing ROI file: 18_WIN_20231012_101024_1_roi.h5\n",
      "Processing group:    index   shape-type  vertex-index      axis-0      axis-1\n",
      "0      0  add_polygon             0   70.742562   25.491618\n",
      "1      0  add_polygon             1   96.244374    7.490339\n",
      "2      0  add_polygon             2  181.750449  118.498226\n",
      "3      0  add_polygon             3  158.748814  136.999541\n",
      "Processing group:    index   shape-type  vertex-index      axis-0      axis-1\n",
      "4      1  add_polygon             0  188.750946  174.502205\n",
      "5      1  add_polygon             1  210.752509  156.500926\n",
      "6      1  add_polygon             2  294.258442  264.508600\n",
      "7      1  add_polygon             3  272.256879  280.509737\n",
      "Processing group:     index   shape-type  vertex-index      axis-0      axis-1\n",
      "8       2  add_polygon             0  157.248708  138.499647\n",
      "9       2  add_polygon             1  190.251053  176.002312\n",
      "10      2  add_polygon             2   84.243521  255.007925\n",
      "11      2  add_polygon             3   49.741070  218.505332\n",
      "Processing group:     index   shape-type  vertex-index      axis-0      axis-1\n",
      "12      3  add_polygon             0  184.750662  116.998120\n",
      "13      3  add_polygon             1  291.258229   33.492187\n",
      "14      3  add_polygon             2  321.760396   73.495029\n",
      "15      3  add_polygon             3  215.252829  155.000820\n",
      "Extracted polygons: [<dlc.analysis.time_roi.PolygonROI object at 0x16a8f4ca0>, <dlc.analysis.time_roi.PolygonROI object at 0x16a8f52d0>, <dlc.analysis.time_roi.PolygonROI object at 0x16a8f5ed0>, <dlc.analysis.time_roi.PolygonROI object at 0x16a8f5e40>]\n",
      "Extracted polygons from 18_WIN_20231012_101024_1_roi.h5: 4\n",
      "Processing ROI file: 11_WIN_20231010_111425_1_roi.h5\n",
      "Processing group:    index   shape-type  vertex-index      axis-0      axis-1\n",
      "0      0  add_polygon             0   70.742562   25.991654\n",
      "1      0  add_polygon             1   93.744196   10.990588\n",
      "2      0  add_polygon             2  178.250200  117.998191\n",
      "3      0  add_polygon             3  155.748601  136.499505\n",
      "Processing group:    index   shape-type  vertex-index      axis-0      axis-1\n",
      "4      1  add_polygon             0  187.250839  175.002241\n",
      "5      1  add_polygon             1  209.252403  157.500997\n",
      "6      1  add_polygon             2  290.258158  264.508600\n",
      "7      1  add_polygon             3  268.256595  280.009701\n",
      "Processing group:     index   shape-type  vertex-index      axis-0      axis-1\n",
      "8       2  add_polygon             0  156.248637  137.999612\n",
      "9       2  add_polygon             1  187.250839  176.002312\n",
      "10      2  add_polygon             2   84.743556  253.507819\n",
      "11      2  add_polygon             3   51.741212  215.005083\n",
      "Processing group:     index   shape-type  vertex-index      axis-0      axis-1\n",
      "12      3  add_polygon             0  182.250484  116.998120\n",
      "13      3  add_polygon             1  290.258158   33.492187\n",
      "14      3  add_polygon             2  319.260219   73.995064\n",
      "15      3  add_polygon             3  212.752651  156.000891\n",
      "Extracted polygons: [<dlc.analysis.time_roi.PolygonROI object at 0x16a8f47c0>, <dlc.analysis.time_roi.PolygonROI object at 0x16a8f4730>, <dlc.analysis.time_roi.PolygonROI object at 0x16a8f6ad0>, <dlc.analysis.time_roi.PolygonROI object at 0x16a8f6a40>]\n",
      "Extracted polygons from 11_WIN_20231010_111425_1_roi.h5: 4\n"
     ]
    }
   ],
   "source": [
    "roi_polygons = {}\n",
    "for file in os.listdir(video_dir):\n",
    "    if file.endswith('roi.h5'):\n",
    "        roi_file_path = os.path.join(video_dir, file)\n",
    "        print(f\"Processing ROI file: {file}\")\n",
    "\n",
    "        try:\n",
    "            roi_data = pd.read_hdf(roi_file_path)\n",
    "            roi_groups = roi_data.groupby(['index', 'shape-type'])\n",
    "            polygons, _ = time_roi.PolygonROI.extract_polygons(roi_groups)\n",
    "            # Create the key by removing '_roi.h5' from the file name\n",
    "            key = file.replace('_roi.h5', '')\n",
    "            roi_polygons[key] = polygons  # Store polygons with the modified key\n",
    "            print(f\"Extracted polygons from {file}: {len(polygons)}\")\n",
    "        except Exception as e:\n",
    "            print(f\"Error processing {file}: {e}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "scorer = 'DLC_resnet50_capsaicin_malesOct20shuffle1_200000'\n",
    "bodypart = 'Neck'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>file</th>\n",
       "      <th>roi_index</th>\n",
       "      <th>time_in_roi</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>11_WIN_20231010_111425_1DLC_resnet50_capsaicin...</td>\n",
       "      <td>0</td>\n",
       "      <td>160.200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11_WIN_20231010_111425_1DLC_resnet50_capsaicin...</td>\n",
       "      <td>1</td>\n",
       "      <td>134.266667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>11_WIN_20231010_111425_1DLC_resnet50_capsaicin...</td>\n",
       "      <td>2</td>\n",
       "      <td>83.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>11_WIN_20231010_111425_1DLC_resnet50_capsaicin...</td>\n",
       "      <td>3</td>\n",
       "      <td>82.866667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7_WIN_20231010_094516_1DLC_resnet50_capsaicin_...</td>\n",
       "      <td>0</td>\n",
       "      <td>189.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>7_WIN_20231010_094516_1DLC_resnet50_capsaicin_...</td>\n",
       "      <td>1</td>\n",
       "      <td>213.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>7_WIN_20231010_094516_1DLC_resnet50_capsaicin_...</td>\n",
       "      <td>2</td>\n",
       "      <td>35.066667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7_WIN_20231010_094516_1DLC_resnet50_capsaicin_...</td>\n",
       "      <td>3</td>\n",
       "      <td>78.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>18_WIN_20231012_101024_1DLC_resnet50_capsaicin...</td>\n",
       "      <td>0</td>\n",
       "      <td>168.200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>18_WIN_20231012_101024_1DLC_resnet50_capsaicin...</td>\n",
       "      <td>1</td>\n",
       "      <td>222.466667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>18_WIN_20231012_101024_1DLC_resnet50_capsaicin...</td>\n",
       "      <td>2</td>\n",
       "      <td>49.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>18_WIN_20231012_101024_1DLC_resnet50_capsaicin...</td>\n",
       "      <td>3</td>\n",
       "      <td>76.266667</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                 file  roi_index  time_in_roi\n",
       "0   11_WIN_20231010_111425_1DLC_resnet50_capsaicin...          0   160.200000\n",
       "1   11_WIN_20231010_111425_1DLC_resnet50_capsaicin...          1   134.266667\n",
       "2   11_WIN_20231010_111425_1DLC_resnet50_capsaicin...          2    83.666667\n",
       "3   11_WIN_20231010_111425_1DLC_resnet50_capsaicin...          3    82.866667\n",
       "4   7_WIN_20231010_094516_1DLC_resnet50_capsaicin_...          0   189.333333\n",
       "5   7_WIN_20231010_094516_1DLC_resnet50_capsaicin_...          1   213.800000\n",
       "6   7_WIN_20231010_094516_1DLC_resnet50_capsaicin_...          2    35.066667\n",
       "7   7_WIN_20231010_094516_1DLC_resnet50_capsaicin_...          3    78.800000\n",
       "8   18_WIN_20231012_101024_1DLC_resnet50_capsaicin...          0   168.200000\n",
       "9   18_WIN_20231012_101024_1DLC_resnet50_capsaicin...          1   222.466667\n",
       "10  18_WIN_20231012_101024_1DLC_resnet50_capsaicin...          2    49.800000\n",
       "11  18_WIN_20231012_101024_1DLC_resnet50_capsaicin...          3    76.266667"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "time_in_roi = time_roi.TimeinRoi(fps=15)\n",
    "results_df = time_in_roi.time_in_rois_dir(directory=video_dir, rois=roi_polygons, scorer=scorer, body_part=bodypart)\n",
    "results_df            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "output_path = os.path.join(video_dir, \"time_in_rois_results.csv\")\n",
    "results_df.to_csv(output_path, index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Keys in roi_polygons dictionary:\n",
      "7_WIN_20231010_094516_1_roi.h5\n",
      "18_WIN_20231012_101024_1_roi.h5\n",
      "11_WIN_20231010_111425_1_roi.h5\n",
      "\n",
      "Expected keys based on tracking data files:\n",
      "11_WIN_20231010_111425_1DLC_resnet50_capsaicin_malesOct20shuffle1_200000\n",
      "7_WIN_20231010_094516_1DLC_resnet50_capsaicin_malesOct20shuffle1_200000\n",
      "18_WIN_20231012_101024_1DLC_resnet50_capsaicin_malesOct20shuffle1_200000\n"
     ]
    }
   ],
   "source": [
    "# Check the keys in the roi_polygons dictionary\n",
    "print(\"Keys in roi_polygons dictionary:\")\n",
    "for key in roi_polygons.keys():\n",
    "    print(key)\n",
    "\n",
    "# Check if the keys match the expected format\n",
    "print(\"\\nExpected keys based on tracking data files:\")\n",
    "for filename in os.listdir(video_dir):\n",
    "    if filename.endswith('filtered.h5'):\n",
    "        expected_key = filename.replace('_filtered.h5', '')\n",
    "        print(expected_key)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## TRIALS\n",
    "\n",
    "Testing some functions here: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import dlc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "bodyparts = ['nose', 'rightear', 'leftear', 'head']\n",
    "file_path = '/Users/annateruel/sr-ca2+img-anna-2023-11-17/videos/AD22-56-neutro2-1DLC_dlcrnetms5_sr-ca2+imgNov17shuffle1_500000_filtered.h5'\n",
    "title = 'Demo Plot'\n",
    "path = '/Users/annateruel/sr-ca2+img-anna-2023-11-17/videos/'  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "density_plot = dlc.plotting.TrackingPlot(style='light')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "density_plot.plot_directory(path, bodyparts, title)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Video clustering for labelling\n",
    "\n",
    "I want to create a function that can analyze a set of videos, perform a form of dimensionality reduction or clustering on them, and then identify which videos are most distinct from each other. This process will involve comparing videos based on the similarity of their pixel values, possibly in grayscale, to group similar videos into clusters. From each cluster, you can then select one representative video for labeling in your DeepLabCut model.\n",
    "\n",
    "This is a test code to do that. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2\n",
    "import numpy as np\n",
    "import os\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.cluster import KMeans\n",
    "from scipy.spatial.distance import cdist\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_and_preprocess_video(video_path, frame_count=100):\n",
    "    \"\"\"\n",
    "    This function takes a video path and extracts a fixed number of frames, converts them to grayscale, resizes, and flattens them.\n",
    "    \"\"\"    \n",
    "    cap = cv2.VideoCapture(video_path)\n",
    "    frames = []\n",
    "    try:\n",
    "        while True:\n",
    "            ret, frame = cap.read()\n",
    "            if not ret:\n",
    "                break\n",
    "            gray_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)\n",
    "            gray_frame = cv2.resize(gray_frame, (100, 100))  # Resize for consistency\n",
    "            frames.append(gray_frame.flatten())\n",
    "            if len(frames) == frame_count:\n",
    "                break\n",
    "    finally:\n",
    "        cap.release()\n",
    "    return np.array(frames)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " The function load_and_preprocess_video you have is designed to process a single video. To create the video_features array that is needed for the rest of the PCA and clustering steps, you will need to call this function on each video in your directory and then average the frames to get a single feature vector for each video.\n",
    "\n",
    "Here's how you can create a function to load and preprocess features for all videos in a directory, using your existing function:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_video_features_from_directory(video_directory, frame_count=100):\n",
    "    \"\"\"\n",
    "    Loads and preprocesses video features for all videos in a directory.\n",
    "    \n",
    "    :param video_directory: The path to the directory containing the videos.\n",
    "    :param frame_count: The number of frames to extract from each video.\n",
    "    :return: A tuple of an array of video features and a list of video paths.\n",
    "    \"\"\"\n",
    "    video_paths = [os.path.join(video_directory, f) for f in os.listdir(video_directory) if f.endswith('.mp4')]\n",
    "    video_features = []\n",
    "\n",
    "    for video_path in video_paths:\n",
    "        # Use the provided function to load and preprocess the video\n",
    "        preprocessed_frames = load_and_preprocess_video(video_path, frame_count)\n",
    "        # Calculate the mean across all frames to get a single feature vector per video\n",
    "        video_feature = np.mean(preprocessed_frames, axis=0)\n",
    "        video_features.append(video_feature)\n",
    "\n",
    "    return np.array(video_features), video_paths\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "video_directory = \"/Users/annateruel/sdt_videos/\"\n",
    "video_features, video_paths = load_video_features_from_directory(video_directory)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The choice of setting the number of clusters to 5 was arbitrary and provided as a starting point for the example. In practice, the optimal number of clusters depends on the specific characteristics and distribution of your data.\n",
    "\n",
    "To determine the most appropriate number of clusters, you can use several methods:\n",
    "\n",
    "1. *Elbow Method*: Plot the sum of squared distances of samples to their closest cluster center for a range of number of clusters. Look for the “elbow” where the rate of decrease sharply changes, which can be considered an indicator of the optimal number of clusters.\n",
    "2. *Silhouette Score*: Calculate the mean silhouette coefficient over all samples. This gives a perspective into the density and separation of the formed clusters. The silhouette score ranges from -1 to 1, where a high value indicates that the object is well matched to its own cluster and poorly matched to neighboring clusters.\n",
    "3. *Gap Statistic*: Compares the total within intra-cluster variation for different values of k with their expected values under null reference distribution of the data. The optimal k is the one that maximizes the gap statistic.\n",
    "4. *Domain Knowledge*: Sometimes the optimal number of clusters is suggested by the context of the problem or domain expertise.\n",
    "\n",
    "Determining the best number of clusters for a dataset without domain knowledge can be challenging because it often depends on the context of the data and the goal of the clustering. However, there are several statistical methods that can help you decide. I will outline two of the most commonly used methods:\n",
    "\n",
    "1. **The elbow method**: This method involves plotting the explained variance as a function of the number of clusters, and picking the elbow of the curve as the number of clusters to use. The idea is to choose a small value of k that still has a low sum of squared distances (inertia).\n",
    "\n",
    "2. **The shilhouette method**: The silhouette value measures how similar an object is to its own cluster (cohesion) compared to other clusters (separation). The silhouette ranges from -1 to +1, where a high value indicates that the object is well matched to its own cluster and poorly matched to neighboring clusters. If many points have a high value, the clustering configuration is appropriate. If many points have a low or negative value, the clustering configuration may have too many or too few clusters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import silhouette_score\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Function to calculate the Sum of Squared Distances (SSD) for different values of k\n",
    "def calculate_ssd_for_k(reduced_features, k_range):\n",
    "    ssd = []\n",
    "    for k in k_range:\n",
    "        kmeans = KMeans(n_clusters=k, random_state=42).fit(reduced_features)\n",
    "        ssd.append(kmeans.inertia_)\n",
    "    return ssd\n",
    "\n",
    "# Function to calculate silhouette scores for different values of k\n",
    "def calculate_silhouette_for_k(reduced_features, k_range):\n",
    "    silhouette_scores = []\n",
    "    for k in k_range:\n",
    "        kmeans = KMeans(n_clusters=k, random_state=42).fit(reduced_features)\n",
    "        score = silhouette_score(reduced_features, kmeans.labels_)\n",
    "        silhouette_scores.append(score)\n",
    "    return silhouette_scores\n",
    "\n",
    "# Define the range of k you want to test\n",
    "k_range = range(2, 11)\n",
    "\n",
    "# Calculate SSD and silhouette scores\n",
    "ssd = calculate_ssd_for_k(reduced_features, k_range)\n",
    "silhouette_scores = calculate_silhouette_for_k(reduced_features, k_range)\n",
    "\n",
    "# Plotting the Elbow Method\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.plot(k_range, ssd, 'bx-')\n",
    "plt.xlabel('k')\n",
    "plt.ylabel('Sum of squared distances')\n",
    "plt.title('Elbow Method For Optimal k')\n",
    "\n",
    "# Plotting the Silhouette Method\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.plot(k_range, silhouette_scores, 'bx-')\n",
    "plt.xlabel('k')\n",
    "plt.ylabel('Silhouette score')\n",
    "plt.title('Silhouette Method For Optimal k')\n",
    "plt.tight_layout()\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/annateruel/miniconda3/envs/dlc_analysis/lib/python3.10/site-packages/sklearn/cluster/_kmeans.py:1412: FutureWarning:\n",
      "\n",
      "The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "plotlyServerURL": "https://plot.ly"
       },
       "data": [
        {
         "hoverinfo": "text",
         "hovertemplate": "PCA Component 1=%{x}<br>PCA Component 2=%{y}<br>PCA Component 3=%{z}<br>color=%{marker.color}<extra></extra>",
         "legendgroup": "",
         "marker": {
          "color": [
           1,
           0,
           7,
           7,
           14,
           10,
           2,
           4,
           1,
           11,
           13,
           14,
           7,
           3,
           9,
           1,
           8,
           7,
           10,
           1,
           8,
           3,
           7,
           0,
           10,
           2,
           5,
           11,
           0,
           14,
           12,
           0,
           1,
           5,
           1,
           2,
           1,
           12,
           4,
           1,
           8,
           6,
           1,
           2,
           10,
           1,
           9,
           0,
           2,
           1,
           4,
           0,
           11,
           13,
           14,
           3,
           3,
           4,
           14,
           1,
           1,
           7,
           8,
           5,
           10,
           6,
           5,
           0,
           9,
           1,
           7,
           1,
           11,
           5,
           10,
           3,
           7,
           9,
           10,
           10,
           7,
           2,
           3,
           1,
           9,
           11,
           2,
           10,
           2,
           12,
           7,
           12,
           0,
           7,
           9,
           1,
           3,
           13,
           7,
           1,
           13,
           1,
           8,
           13,
           4
          ],
          "coloraxis": "coloraxis",
          "size": 8,
          "symbol": "circle"
         },
         "mode": "markers",
         "name": "",
         "scene": "scene",
         "showlegend": false,
         "text": [
          "AD23-054_S1_bw_batch2.mp4",
          "AD23-081-S1_3_bw_batch1.mp4",
          "AD23-053-S2_3_bw_batch1.mp4",
          "AD23-101-S1_3_bw_batch1.mp4",
          "AD23-054_S5_bw_batch2.mp4",
          "AD22-128-S3_3_bw_batch1.mp4",
          "AD23-085-S2_3_bw_batch1.mp4",
          "AD23-081-S3_3__bw_batch1.mp4",
          "AD23-087_S1_bw_batch2.mp4",
          "AD23-100_S3_bw_batch2.mp4",
          "AD23-080-S3_3__bw_batch1.mp4",
          "AD23-087_S5_bw_batch2.mp4",
          "AD23-079-S1_3_bw_batch1.mp4",
          "AD23-086-S3_3__bw_batch1.mp4",
          "AD23-100_S4_bw_batch2.mp4",
          "AD23-087_S2_bw_batch2.mp4",
          "AD22-130-S2_3_bw_batch1.mp4",
          "AD22-129-S1_3_bw_batch1.mp4",
          "AD23-100-S3_3__bw_batch1.mp4",
          "AD23-054_S2_bw_batch2.mp4",
          "AD23-087_S3_2_bw_batch2.mp4",
          "AD23-086-S2_3_bw_batch1.mp4",
          "AD23-092-S1_3_bw_batch1.mp4",
          "AD23-054-S1_3_bw_batch1.mp4",
          "AD23-085-S3_3__bw_batch1.mp4",
          "AD23-050-S2_3_bw_batch1.mp4",
          "AD23-080-S2_3_bw_batch1.mp4",
          "AD23-092_S1_bw_batch2.mp4",
          "AD23-052-S1_3_bw_batch1.mp4",
          "AD23-055_S5_bw_batch2.mp4",
          "AD23-092_S5_bw_batch2.mp4",
          "AD23-084-S1_3_bw_batch1.mp4",
          "AD23-054_S4_2_bw_batch2.mp4",
          "AD23-100-S2_3_bw_batch1.mp4",
          "AD23-101_S3_2_bw_batch2.mp4",
          "AD23-055_S1_bw_batch2.mp4",
          "AD23-092_S4_2_bw_batch2.mp4",
          "AD23-086_S5_bw_batch2.mp4",
          "AD23-079-S3_3__bw_batch1.mp4",
          "AD23-101_S3_bw_batch2.mp4",
          "AD23-085_S3_bw_batch2.mp4",
          "AD23-87-S2_3_bw_batch1.mp4",
          "AD23-086_S1_bw_batch2.mp4",
          "AD23-086_S2_bw_batch2.mp4",
          "AD22-128-S2_3_bw_batch1.mp4",
          "AD23-101_S4_bw_batch2.mp4",
          "AD23-085_S4_bw_batch2.mp4",
          "AD23-087-S1_3_bw_batch1.mp4",
          "AD23-055-S2_3_bw_batch1.mp4",
          "AD23-092_S2_bw_batch2.mp4",
          "AD22-130-S3_3__bw_batch1.mp4",
          "AD23-051-S1_3_bw_batch1.mp4",
          "AD23-055_S2_bw_batch2.mp4",
          "AD22-130-S1_3_bw_batch1.mp4",
          "AD23-085_S5_bw_batch2.mp4",
          "AD23-084-S3_3_bw_batch1.mp4",
          "AD22-129-S2_3_bw_batch1.mp4",
          "AD23-054-S2_3_1_batch1.mp4",
          "AD23-101_S5_bw_batch2.mp4",
          "AD23-086_S3_bw_batch2.mp4",
          "AD23-085_S1_bw_batch2.mp4",
          "AD23-055-S3_3__bw_batch1.mp4",
          "AD23-101_S1_bw_batch2.mp4",
          "AD23-079-S2_3_bw_batch1.mp4",
          "AD23-050-S1_3_bw_batch1.mp4",
          "AD23-055_S3_bw_batch2.mp4",
          "AD23-092-S2_3_bw_batch1.mp4",
          "AD23-086-S1_3_bw_batch1.mp4",
          "AD23-100_S4_2_bw_batch2.mp4",
          "AD23-092_S3_bw_batch2.mp4",
          "AD23-085-S1_3_bw_batch1.mp4",
          "AD23-092_S4_bw_batch2.mp4",
          "AD23-100_S3_2_bw_batch2.mp4",
          "AD23-101-S2_3_bw_batch1.mp4",
          "AD23-051-S3_3__bw_batch1.mp4",
          "AD23-081-S2_3_bw_batch1.mp4",
          "AD23-053-S1_3_bw_batch1.mp4",
          "AD23-055_S4_bw_batch2.mp4",
          "AD23-050-S3_3__bw_batch1.mp4",
          "AD23-053-S3_3__bw_batch1.mp4",
          "AD23-087-S3_3_bw_batch1.mp4",
          "AD23-085_S2_bw_batch2.mp4",
          "AD23-052-S3_3__bw_batch1.mp4",
          "AD23-101_S2_bw_batch2.mp4",
          "AD23-086_S4_bw_batch2.mp4",
          "AD23-100_S1_bw_batch2.mp4",
          "AD23-101-S3_3_bw_batch1.mp4",
          "AD22-128-S1_3_bw_batch1.mp4",
          "AD23-087_S3_bw_batch2.mp4",
          "AD23-100_S5_bw_batch2.mp4",
          "AD23-051-S2_3_bw_batch1.mp4",
          "AD23-054_S3_bw_batch2.mp4",
          "AD23-055-S1_3_bw_batch1.mp4",
          "AD23-100-S1_3_bw_batch1.mp4",
          "AD23-086_S4_2_bw_batch2.mp4",
          "AD23-054_S4_bw_batch2.mp4",
          "AD23-084-S2_3_bw_batch1.mp4",
          "AD22-129-S3_3_bw_batch1.mp4",
          "AD23-080-S1_3_bw_batch1.mp4",
          "AD23-086_S3_1_bw_batch2.mp4",
          "AD23-052-S2_3_bw_batch1.mp4",
          "AD23-087_S4_bw_batch2.mp4",
          "AD23-100_S2_bw_batch2.mp4",
          "AD23-092-S3_3_bw_batch1.mp4",
          "AD23-054-S3_3_bw_batch1.mp4"
         ],
         "type": "scatter3d",
         "x": [
          5376.752748823752,
          -2537.079462360795,
          -1926.946756730366,
          -1663.6776779167537,
          6591.433766601307,
          -3089.901353059174,
          -481.9178784703933,
          -4580.930247057205,
          4542.571850183632,
          -4625.364279399827,
          -3244.2358919171,
          6734.747013986192,
          -1469.072320388057,
          -1708.1640844463116,
          2710.1313818779176,
          4697.788790637575,
          -964.6917775351196,
          -1285.444869153662,
          -4603.116308238724,
          4717.817453899894,
          -1550.6232630819432,
          -1415.1614411205765,
          -1419.0725476275309,
          -4843.127930678985,
          -2973.416769241281,
          152.19032557653026,
          1963.930774920341,
          -4073.4622680008424,
          -2817.965155366861,
          5916.150316312478,
          6988.949141463371,
          -2052.6201443554824,
          4785.272732557183,
          343.35610020111704,
          5368.560349488817,
          -354.05065683294987,
          5426.252102581415,
          6832.762342293843,
          -4328.882697153717,
          5120.29916990244,
          -412.93356268122733,
          -6783.839428218502,
          4444.044081924753,
          -362.3781516167533,
          -3699.1341734228986,
          5294.772022969203,
          3144.5416690450447,
          -3879.701432235351,
          -370.89313640206393,
          4943.237619636131,
          -4248.93195120294,
          -2869.5970186267837,
          -4241.868788391173,
          -2858.9806373106667,
          6572.801174384918,
          -1103.6224681089575,
          -2624.6483257165632,
          -5226.812463038203,
          6323.573680023059,
          5313.298218709046,
          4410.134169508319,
          -2370.9016663653792,
          -1460.1465713139185,
          1185.1771018774932,
          -3183.0954798304942,
          -7063.6507844015105,
          977.2007869114766,
          -3462.682000935804,
          2028.5160912611607,
          4928.828873093349,
          -1245.0195984392562,
          4901.2834546757185,
          -3806.4132648670693,
          784.4694622610955,
          -3316.5695276904453,
          -873.1425980313061,
          -2332.418103985683,
          2312.9377314484614,
          -4499.4330866310665,
          -3820.6296829093876,
          -528.5977628962812,
          -1211.9080553322337,
          -872.3681775997624,
          4819.566653273979,
          1039.2911224219547,
          -5041.301224436049,
          -1242.8106890550478,
          -3822.517506076877,
          -450.80059018662723,
          7542.0211398667,
          -1114.589219884747,
          6284.123279921652,
          -4629.213237342748,
          -1754.0543866811468,
          1754.7637483437559,
          4121.92983991171,
          -2355.646745667328,
          -1984.1743723982934,
          471.7586681961264,
          5012.463310616463,
          -3664.882447180801,
          4702.204275512323,
          -1874.0848128543807,
          -1981.826241363517,
          -4926.759385638792
         ],
         "y": [
          -594.5984013952109,
          -1162.2751907071415,
          -704.0395059333237,
          -1459.4443427491135,
          463.7036830033256,
          -585.2140889877188,
          807.8236593059133,
          -1695.2948750489647,
          -306.79160536627353,
          4851.098193410676,
          -2596.623343777452,
          18.85059840890823,
          -1046.5210460393698,
          -3671.407916859134,
          3601.220758477944,
          -854.8164517096765,
          5051.979979212979,
          -1904.7515078982262,
          -243.77077459719686,
          -644.9721464067459,
          4387.602257514306,
          -3383.13032314782,
          -932.7902428781515,
          -745.3003720172507,
          -396.4285468229258,
          2635.4457598348718,
          -4640.364945690202,
          4435.339200363351,
          -551.3543422444134,
          2063.186868553816,
          -667.770372589642,
          -1343.0082924708672,
          84.12170170770189,
          -3653.7776743892314,
          -371.97832333987,
          3083.357436262664,
          209.1391494351535,
          -406.594680800688,
          -1796.543364313982,
          -292.25992462216703,
          4026.649640682954,
          1124.2386260534088,
          -255.80046651830895,
          3339.6680490012186,
          -1019.102661570249,
          100.12759200668388,
          3275.5009788207885,
          338.1565670855411,
          1951.0772002817514,
          -333.0904590212709,
          -1510.7869280839393,
          -916.9502048047241,
          4592.994486083146,
          -3242.0072913115923,
          662.1961430445463,
          -3133.2833943500164,
          -3481.7534205883235,
          -390.36664610925277,
          906.3244653896563,
          -698.0878145925784,
          -714.9433313585888,
          -1115.2800061833032,
          4261.499584758916,
          -4958.411801887069,
          -872.3677845880158,
          771.8027001414929,
          -3981.2689200350387,
          -1783.139424125735,
          3482.4590583938575,
          -434.9879575716131,
          -1521.4682542650198,
          937.1174989573725,
          5214.614060393093,
          -2842.8670447790014,
          -273.25398689324436,
          -3805.5253934145185,
          -1162.0055885939169,
          5195.848604165498,
          -401.9302099428684,
          -566.8187422116682,
          -437.86891196665186,
          2148.1648175949235,
          -3633.4290083426235,
          -792.2068734043787,
          5644.032581485481,
          3602.2552732146587,
          2123.2317640464903,
          -1603.0413782574647,
          3525.8074331912594,
          -3462.5771923737307,
          692.0544217282423,
          -1852.217724733277,
          -358.9759970114803,
          -2397.891315649079,
          4962.86251786979,
          807.4278100808898,
          -2447.9532617895024,
          -1303.4637199122533,
          -2443.0453699032305,
          -502.0853433564112,
          -2865.5590140135396,
          1029.0296960482076,
          5520.486792430696,
          -772.2519236685846,
          -988.6102384573505
         ],
         "z": [
          69.81649700923971,
          3166.5688745593625,
          1984.4607093006834,
          3020.3643609196824,
          -248.60356799512596,
          -309.77483233343696,
          1198.3020745517877,
          -1660.860786803463,
          261.6454916607038,
          -1093.7337667621548,
          -1658.9580003268604,
          -219.00270105727625,
          1420.7108898406982,
          -2471.5969169602818,
          -841.245558705013,
          597.7026384270221,
          -438.8296370405381,
          2493.35368979736,
          950.1258280928428,
          374.7346866231669,
          -595.4370376118375,
          -3396.7641707579887,
          2456.418546037136,
          2257.8339645138353,
          751.6315901570018,
          458.13057335272043,
          -1965.5423175858814,
          -896.2306765929196,
          2596.786449204002,
          -881.2317614638804,
          -5.266571881469827,
          3496.7098981740774,
          808.5182672517019,
          -1440.9939187798668,
          335.472814282631,
          274.5974838644635,
          103.54729934438619,
          -81.38547930074968,
          -1121.6371126456604,
          322.8181902380762,
          -1144.0627317791175,
          109.72688030819343,
          189.9416414833527,
          261.4540918786065,
          1391.7901173793318,
          414.22563576093165,
          -910.9579993251866,
          2567.3940842959983,
          951.4026708301658,
          253.69424488377308,
          -352.87612158218934,
          3200.1782291849513,
          -540.1917433867995,
          -715.1628803255527,
          -397.83106691032987,
          -2190.2459092842,
          -2216.0735500357523,
          -1247.230687208826,
          -438.3870890449559,
          403.2101496341354,
          401.24492868728663,
          1690.031380275457,
          -80.69381285826579,
          -2386.9251820167306,
          834.2158930794137,
          -2425.2193107987223,
          -2423.3886488317175,
          4106.041198128136,
          -520.5416929514083,
          530.3198227407122,
          1853.6061468434866,
          -208.4976550942919,
          -1186.3117550417207,
          -802.1016144856078,
          753.5026020337799,
          -1491.2057505913538,
          2204.794748583156,
          -1615.073869397015,
          289.6151737228421,
          479.32103808377457,
          920.2681533883975,
          518.8807534276025,
          -3114.0414771303886,
          481.8410122894228,
          -1632.135379420407,
          -527.4371321011291,
          37.486287545991814,
          1044.4145734718882,
          -436.67413735516885,
          1017.8177164359673,
          1329.862790320984,
          649.3136027527092,
          3477.2724149419814,
          1226.7924597952672,
          -1295.24073585057,
          567.4913272315118,
          -3539.3939583557944,
          -620.8361598459609,
          1582.862743289436,
          381.7760858521518,
          -3298.568846855756,
          54.1401357321044,
          -1124.0108576098487,
          -3318.650413906779,
          -4049.1185675095426
         ]
        }
       ],
       "layout": {
        "coloraxis": {
         "colorbar": {
          "title": {
           "text": "color"
          }
         },
         "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"
          ]
         ]
        },
        "legend": {
         "tracegroupgap": 0
        },
        "scene": {
         "domain": {
          "x": [
           0,
           1
          ],
          "y": [
           0,
           1
          ]
         },
         "xaxis": {
          "title": {
           "text": "PCA Component 1"
          }
         },
         "yaxis": {
          "title": {
           "text": "PCA Component 2"
          }
         },
         "zaxis": {
          "title": {
           "text": "PCA Component 3"
          }
         }
        },
        "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
          }
         }
        },
        "title": {
         "text": "PCA Clustering of Videos"
        }
       }
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Representative Videos: ['/Users/annateruel/sdt_videos/AD23-081-S1_3_bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD23-054_S1_bw_batch2.mp4', '/Users/annateruel/sdt_videos/AD23-085-S2_3_bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD23-086-S3_3__bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD23-081-S3_3__bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD23-080-S2_3_bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD23-87-S2_3_bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD23-053-S2_3_bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD22-130-S2_3_bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD23-100_S4_bw_batch2.mp4', '/Users/annateruel/sdt_videos/AD22-128-S3_3_bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD23-100_S3_bw_batch2.mp4', '/Users/annateruel/sdt_videos/AD23-092_S5_bw_batch2.mp4', '/Users/annateruel/sdt_videos/AD23-080-S3_3__bw_batch1.mp4', '/Users/annateruel/sdt_videos/AD23-054_S5_bw_batch2.mp4']\n"
     ]
    }
   ],
   "source": [
    "import plotly.express as px \n",
    "\n",
    "def main(video_directory, n_clusters=15):\n",
    "    video_paths = [os.path.join(video_directory, f) for f in os.listdir(video_directory) if f.endswith('.mp4')]\n",
    "    video_features = []\n",
    "\n",
    "    for video_path in video_paths:\n",
    "        features = load_and_preprocess_video(video_path)\n",
    "        video_features.append(np.mean(features, axis=0))  # Mean of frames as the video feature\n",
    "\n",
    "    video_features = np.array(video_features)\n",
    "\n",
    "    # Use 3 PCA components for 3D visualization\n",
    "    pca = PCA(n_components=3)\n",
    "    reduced_features = pca.fit_transform(video_features)\n",
    "\n",
    "    # KMeans clustering\n",
    "    kmeans = KMeans(n_clusters=n_clusters)\n",
    "    kmeans.fit(reduced_features)\n",
    "    labels = kmeans.labels_\n",
    "\n",
    "    # Create a 3D scatter plot using Plotly\n",
    "    fig = px.scatter_3d(\n",
    "        reduced_features, x=0, y=1, z=2,\n",
    "        color=labels,\n",
    "        labels={'0': 'PCA Component 1', '1': 'PCA Component 2', '2': 'PCA Component 3'},\n",
    "        title='PCA Clustering of Videos'\n",
    "    )\n",
    "\n",
    "    # Customize the plotly figure to display video names on hover\n",
    "    hover_texts = [os.path.basename(video_path) for video_path in video_paths]\n",
    "    fig.update_traces(marker_size=8, hoverinfo='text', text=hover_texts)\n",
    "\n",
    "    fig.show()\n",
    "\n",
    "    # Select one video from each cluster\n",
    "    representative_videos = []\n",
    "    for i in range(n_clusters):\n",
    "        cluster_indices = np.where(labels == i)[0]\n",
    "        representative_videos.append(video_paths[cluster_indices[0]])  # Select the first video of each cluster for simplicity\n",
    "\n",
    "    print(\"Representative Videos:\", representative_videos)\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main(\"/Users/annateruel/sdt_videos/\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## trying lr and loss plotting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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>iterations</th>\n",
       "      <th>loss</th>\n",
       "      <th>lr</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000</td>\n",
       "      <td>0.02260</td>\n",
       "      <td>0.0100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2000</td>\n",
       "      <td>0.01550</td>\n",
       "      <td>0.0100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3000</td>\n",
       "      <td>0.01400</td>\n",
       "      <td>0.0100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4000</td>\n",
       "      <td>0.01310</td>\n",
       "      <td>0.0100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5000</td>\n",
       "      <td>0.01270</td>\n",
       "      <td>0.0100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>363</th>\n",
       "      <td>414000</td>\n",
       "      <td>0.00322</td>\n",
       "      <td>0.0005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>364</th>\n",
       "      <td>415000</td>\n",
       "      <td>0.00312</td>\n",
       "      <td>0.0005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>365</th>\n",
       "      <td>416000</td>\n",
       "      <td>0.00325</td>\n",
       "      <td>0.0005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>366</th>\n",
       "      <td>417000</td>\n",
       "      <td>0.00330</td>\n",
       "      <td>0.0005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>367</th>\n",
       "      <td>418000</td>\n",
       "      <td>0.00313</td>\n",
       "      <td>0.0005</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>368 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     iterations     loss      lr\n",
       "0          1000  0.02260  0.0100\n",
       "1          2000  0.01550  0.0100\n",
       "2          3000  0.01400  0.0100\n",
       "3          4000  0.01310  0.0100\n",
       "4          5000  0.01270  0.0100\n",
       "..          ...      ...     ...\n",
       "363      414000  0.00322  0.0005\n",
       "364      415000  0.00312  0.0005\n",
       "365      416000  0.00325  0.0005\n",
       "366      417000  0.00330  0.0005\n",
       "367      418000  0.00313  0.0005\n",
       "\n",
       "[368 rows x 3 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_path = '/Users/annateruel/Desktop/learning_stats2.csv' \n",
    "df = pd.read_csv(data_path, sep=';', decimal=',', engine='python')\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.style.use(\"dark_background\")\n",
    "\n",
    "# Create a figure and a set of subplots\n",
    "fig, ax1 = plt.subplots()\n",
    "\n",
    "# Plot the loss on the primary y-axis\n",
    "ax1.plot(df['iterations'], df['loss'], color='orange', label='Loss')\n",
    "ax1.set_xlabel('Iterations')\n",
    "ax1.set_ylabel('Loss', color='orange')\n",
    "ax1.tick_params(axis='y', labelcolor='orange')\n",
    "\n",
    "# Create a second y-axis for the learning rate\n",
    "ax2 = ax1.twinx()\n",
    "ax2.plot(df['iterations'], df['lr'], color='purple', label='Learning Rate')\n",
    "ax2.set_ylabel('Learning Rate', color='purple')\n",
    "ax2.tick_params(axis='y', labelcolor='purple')\n",
    "\n",
    "# Show the legend\n",
    "ax1.legend(loc='upper left')\n",
    "ax2.legend(loc='upper right')\n",
    "\n",
    "# Show the plot\n",
    "plt.title('Loss and Learning Rate Through Number of Iterations')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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 tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>scorer</th>\n",
       "      <th colspan=\"21\" halign=\"left\">Paola</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>individuals</th>\n",
       "      <th colspan=\"10\" halign=\"left\">individual1</th>\n",
       "      <th>...</th>\n",
       "      <th colspan=\"10\" halign=\"left\">individual2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>bodyparts</th>\n",
       "      <th colspan=\"2\" halign=\"left\">beak</th>\n",
       "      <th colspan=\"2\" halign=\"left\">left flipper base</th>\n",
       "      <th colspan=\"2\" halign=\"left\">left flipper tip</th>\n",
       "      <th colspan=\"2\" halign=\"left\">right flipper base</th>\n",
       "      <th colspan=\"2\" halign=\"left\">right flipper tip</th>\n",
       "      <th>...</th>\n",
       "      <th colspan=\"2\" halign=\"left\">dorsal base</th>\n",
       "      <th colspan=\"2\" halign=\"left\">dorsal tip</th>\n",
       "      <th colspan=\"2\" halign=\"left\">tail</th>\n",
       "      <th colspan=\"2\" halign=\"left\">left fluke</th>\n",
       "      <th colspan=\"2\" halign=\"left\">right fluke</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>coords</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>...</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"16\" valign=\"top\">labeled-data</th>\n",
       "      <th rowspan=\"16\" valign=\"top\">DJI_20231011121723_0076_D</th>\n",
       "      <th>img0056.png</th>\n",
       "      <td>1377.889896</td>\n",
       "      <td>1343.359701</td>\n",
       "      <td>1255.117834</td>\n",
       "      <td>1497.327944</td>\n",
       "      <td>1214.864698</td>\n",
       "      <td>1536.574751</td>\n",
       "      <td>1341.662075</td>\n",
       "      <td>1523.492482</td>\n",
       "      <td>1344.681060</td>\n",
       "      <td>1582.865857</td>\n",
       "      <td>...</td>\n",
       "      <td>1500.661959</td>\n",
       "      <td>1388.644478</td>\n",
       "      <td>1513.744228</td>\n",
       "      <td>1437.954569</td>\n",
       "      <td>1474.497421</td>\n",
       "      <td>1685.511352</td>\n",
       "      <td>1383.927867</td>\n",
       "      <td>1752.935354</td>\n",
       "      <td>1532.864468</td>\n",
       "      <td>1745.891055</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0082.png</th>\n",
       "      <td>1404.054496</td>\n",
       "      <td>1213.543339</td>\n",
       "      <td>1336.630494</td>\n",
       "      <td>1376.568538</td>\n",
       "      <td>1318.516583</td>\n",
       "      <td>1407.764718</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1434.244347</td>\n",
       "      <td>1418.834330</td>\n",
       "      <td>...</td>\n",
       "      <td>1476.510140</td>\n",
       "      <td>1326.252118</td>\n",
       "      <td>1485.567095</td>\n",
       "      <td>1383.612836</td>\n",
       "      <td>1499.655692</td>\n",
       "      <td>1607.017738</td>\n",
       "      <td>1436.257004</td>\n",
       "      <td>1648.277202</td>\n",
       "      <td>1580.161963</td>\n",
       "      <td>1638.213918</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0096.png</th>\n",
       "      <td>1404.054496</td>\n",
       "      <td>1108.885187</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1471.478498</td>\n",
       "      <td>1243.733191</td>\n",
       "      <td>1471.478498</td>\n",
       "      <td>1299.081252</td>\n",
       "      <td>1500.662021</td>\n",
       "      <td>1529.530452</td>\n",
       "      <td>1436.257004</td>\n",
       "      <td>1593.935469</td>\n",
       "      <td>1579.155635</td>\n",
       "      <td>1581.859528</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0117.png</th>\n",
       "      <td>1394.997540</td>\n",
       "      <td>924.727093</td>\n",
       "      <td>1372.858316</td>\n",
       "      <td>1075.676351</td>\n",
       "      <td>1361.788704</td>\n",
       "      <td>1126.999098</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1491.605065</td>\n",
       "      <td>1148.131994</td>\n",
       "      <td>1488.586080</td>\n",
       "      <td>1206.499041</td>\n",
       "      <td>1495.630379</td>\n",
       "      <td>1383.612836</td>\n",
       "      <td>1412.105123</td>\n",
       "      <td>1429.903942</td>\n",
       "      <td>1560.035396</td>\n",
       "      <td>1416.821673</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0268.png</th>\n",
       "      <td>1355.750733</td>\n",
       "      <td>568.486845</td>\n",
       "      <td>1367.826674</td>\n",
       "      <td>707.360162</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1419.149422</td>\n",
       "      <td>839.189180</td>\n",
       "      <td>1424.181064</td>\n",
       "      <td>893.530913</td>\n",
       "      <td>1438.269661</td>\n",
       "      <td>1050.518141</td>\n",
       "      <td>1385.940585</td>\n",
       "      <td>1106.872531</td>\n",
       "      <td>1529.845544</td>\n",
       "      <td>1086.745963</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0367.png</th>\n",
       "      <td>1330.592462</td>\n",
       "      <td>377.284436</td>\n",
       "      <td>1327.573477</td>\n",
       "      <td>517.164082</td>\n",
       "      <td>1303.421596</td>\n",
       "      <td>556.410889</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1410.092405</td>\n",
       "      <td>598.676681</td>\n",
       "      <td>1411.098733</td>\n",
       "      <td>646.980443</td>\n",
       "      <td>1447.326555</td>\n",
       "      <td>838.182837</td>\n",
       "      <td>1381.915210</td>\n",
       "      <td>906.613167</td>\n",
       "      <td>1507.706258</td>\n",
       "      <td>884.473942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0370.png</th>\n",
       "      <td>1335.624104</td>\n",
       "      <td>365.208496</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1311.472223</td>\n",
       "      <td>539.303306</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1410.092405</td>\n",
       "      <td>594.651367</td>\n",
       "      <td>1413.111390</td>\n",
       "      <td>635.910831</td>\n",
       "      <td>1434.244286</td>\n",
       "      <td>828.119553</td>\n",
       "      <td>1377.889896</td>\n",
       "      <td>897.556211</td>\n",
       "      <td>1505.693601</td>\n",
       "      <td>877.429644</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0433.png</th>\n",
       "      <td>1405.060824</td>\n",
       "      <td>206.208599</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1465.440527</td>\n",
       "      <td>451.752725</td>\n",
       "      <td>1471.478498</td>\n",
       "      <td>489.993204</td>\n",
       "      <td>1501.668349</td>\n",
       "      <td>679.182940</td>\n",
       "      <td>1462.421542</td>\n",
       "      <td>740.568972</td>\n",
       "      <td>1562.048052</td>\n",
       "      <td>707.360135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0518.png</th>\n",
       "      <td>1432.231691</td>\n",
       "      <td>97.525134</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1496.636707</td>\n",
       "      <td>307.847766</td>\n",
       "      <td>1485.567095</td>\n",
       "      <td>358.164186</td>\n",
       "      <td>1518.775932</td>\n",
       "      <td>536.284310</td>\n",
       "      <td>1463.427871</td>\n",
       "      <td>579.556430</td>\n",
       "      <td>1580.161963</td>\n",
       "      <td>580.562759</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0590.png</th>\n",
       "      <td>1536.889843</td>\n",
       "      <td>106.582089</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1545.946798</td>\n",
       "      <td>268.600959</td>\n",
       "      <td>1545.946798</td>\n",
       "      <td>302.816124</td>\n",
       "      <td>1582.174620</td>\n",
       "      <td>489.993204</td>\n",
       "      <td>1530.851872</td>\n",
       "      <td>556.410877</td>\n",
       "      <td>1637.522681</td>\n",
       "      <td>528.233683</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0780.png</th>\n",
       "      <td>1505.693663</td>\n",
       "      <td>354.138884</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1600.288531</td>\n",
       "      <td>589.619726</td>\n",
       "      <td>1593.244232</td>\n",
       "      <td>639.936145</td>\n",
       "      <td>1604.313844</td>\n",
       "      <td>804.974000</td>\n",
       "      <td>1551.984768</td>\n",
       "      <td>852.271434</td>\n",
       "      <td>1652.617607</td>\n",
       "      <td>852.271434</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0805.png</th>\n",
       "      <td>1508.712648</td>\n",
       "      <td>343.069271</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1586.199934</td>\n",
       "      <td>598.676681</td>\n",
       "      <td>1590.225247</td>\n",
       "      <td>651.005757</td>\n",
       "      <td>1624.440412</td>\n",
       "      <td>817.049940</td>\n",
       "      <td>1573.117665</td>\n",
       "      <td>873.404330</td>\n",
       "      <td>1678.782145</td>\n",
       "      <td>843.214478</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img0934.png</th>\n",
       "      <td>1497.643036</td>\n",
       "      <td>261.556672</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1588.212590</td>\n",
       "      <td>493.012201</td>\n",
       "      <td>1596.263217</td>\n",
       "      <td>532.259008</td>\n",
       "      <td>1613.370800</td>\n",
       "      <td>698.303191</td>\n",
       "      <td>1572.111336</td>\n",
       "      <td>754.657581</td>\n",
       "      <td>1665.699876</td>\n",
       "      <td>736.543670</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img1000.png</th>\n",
       "      <td>1505.693663</td>\n",
       "      <td>73.373249</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1545.946798</td>\n",
       "      <td>290.740180</td>\n",
       "      <td>1539.908828</td>\n",
       "      <td>335.018629</td>\n",
       "      <td>1574.123993</td>\n",
       "      <td>482.948901</td>\n",
       "      <td>1517.769603</td>\n",
       "      <td>556.410873</td>\n",
       "      <td>1642.554323</td>\n",
       "      <td>537.290634</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img1036.png</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1580.161963</td>\n",
       "      <td>133.752952</td>\n",
       "      <td>1571.105008</td>\n",
       "      <td>177.025072</td>\n",
       "      <td>1618.402442</td>\n",
       "      <td>356.151525</td>\n",
       "      <td>1583.180948</td>\n",
       "      <td>395.398332</td>\n",
       "      <td>1674.756831</td>\n",
       "      <td>402.442631</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>img1071.png</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1636.516353</td>\n",
       "      <td>178.031401</td>\n",
       "      <td>1590.225247</td>\n",
       "      <td>218.284536</td>\n",
       "      <td>1697.902384</td>\n",
       "      <td>206.208596</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>16 rows × 40 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "scorer                                                    Paola               \\\n",
       "individuals                                         individual1                \n",
       "bodyparts                                                  beak                \n",
       "coords                                                        x            y   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1377.889896  1343.359701   \n",
       "                                       img0082.png  1404.054496  1213.543339   \n",
       "                                       img0096.png  1404.054496  1108.885187   \n",
       "                                       img0117.png  1394.997540   924.727093   \n",
       "                                       img0268.png  1355.750733   568.486845   \n",
       "                                       img0367.png  1330.592462   377.284436   \n",
       "                                       img0370.png  1335.624104   365.208496   \n",
       "                                       img0433.png  1405.060824   206.208599   \n",
       "                                       img0518.png  1432.231691    97.525134   \n",
       "                                       img0590.png  1536.889843   106.582089   \n",
       "                                       img0780.png  1505.693663   354.138884   \n",
       "                                       img0805.png  1508.712648   343.069271   \n",
       "                                       img0934.png  1497.643036   261.556672   \n",
       "                                       img1000.png  1505.693663    73.373249   \n",
       "                                       img1036.png          NaN          NaN   \n",
       "                                       img1071.png          NaN          NaN   \n",
       "\n",
       "scorer                                                                \\\n",
       "individuals                                                            \n",
       "bodyparts                                          left flipper base   \n",
       "coords                                                             x   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png       1255.117834   \n",
       "                                       img0082.png       1336.630494   \n",
       "                                       img0096.png               NaN   \n",
       "                                       img0117.png       1372.858316   \n",
       "                                       img0268.png       1367.826674   \n",
       "                                       img0367.png       1327.573477   \n",
       "                                       img0370.png               NaN   \n",
       "                                       img0433.png               NaN   \n",
       "                                       img0518.png               NaN   \n",
       "                                       img0590.png               NaN   \n",
       "                                       img0780.png               NaN   \n",
       "                                       img0805.png               NaN   \n",
       "                                       img0934.png               NaN   \n",
       "                                       img1000.png               NaN   \n",
       "                                       img1036.png               NaN   \n",
       "                                       img1071.png               NaN   \n",
       "\n",
       "scorer                                                           \\\n",
       "individuals                                                       \n",
       "bodyparts                                                         \n",
       "coords                                                        y   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1497.327944   \n",
       "                                       img0082.png  1376.568538   \n",
       "                                       img0096.png          NaN   \n",
       "                                       img0117.png  1075.676351   \n",
       "                                       img0268.png   707.360162   \n",
       "                                       img0367.png   517.164082   \n",
       "                                       img0370.png          NaN   \n",
       "                                       img0433.png          NaN   \n",
       "                                       img0518.png          NaN   \n",
       "                                       img0590.png          NaN   \n",
       "                                       img0780.png          NaN   \n",
       "                                       img0805.png          NaN   \n",
       "                                       img0934.png          NaN   \n",
       "                                       img1000.png          NaN   \n",
       "                                       img1036.png          NaN   \n",
       "                                       img1071.png          NaN   \n",
       "\n",
       "scorer                                                               \\\n",
       "individuals                                                           \n",
       "bodyparts                                          left flipper tip   \n",
       "coords                                                            x   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png      1214.864698   \n",
       "                                       img0082.png      1318.516583   \n",
       "                                       img0096.png              NaN   \n",
       "                                       img0117.png      1361.788704   \n",
       "                                       img0268.png              NaN   \n",
       "                                       img0367.png      1303.421596   \n",
       "                                       img0370.png      1311.472223   \n",
       "                                       img0433.png              NaN   \n",
       "                                       img0518.png              NaN   \n",
       "                                       img0590.png              NaN   \n",
       "                                       img0780.png              NaN   \n",
       "                                       img0805.png              NaN   \n",
       "                                       img0934.png              NaN   \n",
       "                                       img1000.png              NaN   \n",
       "                                       img1036.png              NaN   \n",
       "                                       img1071.png              NaN   \n",
       "\n",
       "scorer                                                           \\\n",
       "individuals                                                       \n",
       "bodyparts                                                         \n",
       "coords                                                        y   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1536.574751   \n",
       "                                       img0082.png  1407.764718   \n",
       "                                       img0096.png          NaN   \n",
       "                                       img0117.png  1126.999098   \n",
       "                                       img0268.png          NaN   \n",
       "                                       img0367.png   556.410889   \n",
       "                                       img0370.png   539.303306   \n",
       "                                       img0433.png          NaN   \n",
       "                                       img0518.png          NaN   \n",
       "                                       img0590.png          NaN   \n",
       "                                       img0780.png          NaN   \n",
       "                                       img0805.png          NaN   \n",
       "                                       img0934.png          NaN   \n",
       "                                       img1000.png          NaN   \n",
       "                                       img1036.png          NaN   \n",
       "                                       img1071.png          NaN   \n",
       "\n",
       "scorer                                                                 \\\n",
       "individuals                                                             \n",
       "bodyparts                                          right flipper base   \n",
       "coords                                                              x   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png        1341.662075   \n",
       "                                       img0082.png                NaN   \n",
       "                                       img0096.png                NaN   \n",
       "                                       img0117.png                NaN   \n",
       "                                       img0268.png                NaN   \n",
       "                                       img0367.png                NaN   \n",
       "                                       img0370.png                NaN   \n",
       "                                       img0433.png                NaN   \n",
       "                                       img0518.png                NaN   \n",
       "                                       img0590.png                NaN   \n",
       "                                       img0780.png                NaN   \n",
       "                                       img0805.png                NaN   \n",
       "                                       img0934.png                NaN   \n",
       "                                       img1000.png                NaN   \n",
       "                                       img1036.png                NaN   \n",
       "                                       img1071.png                NaN   \n",
       "\n",
       "scorer                                                           \\\n",
       "individuals                                                       \n",
       "bodyparts                                                         \n",
       "coords                                                        y   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1523.492482   \n",
       "                                       img0082.png          NaN   \n",
       "                                       img0096.png          NaN   \n",
       "                                       img0117.png          NaN   \n",
       "                                       img0268.png          NaN   \n",
       "                                       img0367.png          NaN   \n",
       "                                       img0370.png          NaN   \n",
       "                                       img0433.png          NaN   \n",
       "                                       img0518.png          NaN   \n",
       "                                       img0590.png          NaN   \n",
       "                                       img0780.png          NaN   \n",
       "                                       img0805.png          NaN   \n",
       "                                       img0934.png          NaN   \n",
       "                                       img1000.png          NaN   \n",
       "                                       img1036.png          NaN   \n",
       "                                       img1071.png          NaN   \n",
       "\n",
       "scorer                                                                \\\n",
       "individuals                                                            \n",
       "bodyparts                                          right flipper tip   \n",
       "coords                                                             x   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png       1344.681060   \n",
       "                                       img0082.png       1434.244347   \n",
       "                                       img0096.png               NaN   \n",
       "                                       img0117.png               NaN   \n",
       "                                       img0268.png               NaN   \n",
       "                                       img0367.png               NaN   \n",
       "                                       img0370.png               NaN   \n",
       "                                       img0433.png               NaN   \n",
       "                                       img0518.png               NaN   \n",
       "                                       img0590.png               NaN   \n",
       "                                       img0780.png               NaN   \n",
       "                                       img0805.png               NaN   \n",
       "                                       img0934.png               NaN   \n",
       "                                       img1000.png               NaN   \n",
       "                                       img1036.png               NaN   \n",
       "                                       img1071.png               NaN   \n",
       "\n",
       "scorer                                                           ...  \\\n",
       "individuals                                                      ...   \n",
       "bodyparts                                                        ...   \n",
       "coords                                                        y  ...   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1582.865857  ...   \n",
       "                                       img0082.png  1418.834330  ...   \n",
       "                                       img0096.png          NaN  ...   \n",
       "                                       img0117.png          NaN  ...   \n",
       "                                       img0268.png          NaN  ...   \n",
       "                                       img0367.png          NaN  ...   \n",
       "                                       img0370.png          NaN  ...   \n",
       "                                       img0433.png          NaN  ...   \n",
       "                                       img0518.png          NaN  ...   \n",
       "                                       img0590.png          NaN  ...   \n",
       "                                       img0780.png          NaN  ...   \n",
       "                                       img0805.png          NaN  ...   \n",
       "                                       img0934.png          NaN  ...   \n",
       "                                       img1000.png          NaN  ...   \n",
       "                                       img1036.png          NaN  ...   \n",
       "                                       img1071.png          NaN  ...   \n",
       "\n",
       "scorer                                                                        \\\n",
       "individuals                                         individual2                \n",
       "bodyparts                                           dorsal base                \n",
       "coords                                                        x            y   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1500.661959  1388.644478   \n",
       "                                       img0082.png  1476.510140  1326.252118   \n",
       "                                       img0096.png  1471.478498  1243.733191   \n",
       "                                       img0117.png  1491.605065  1148.131994   \n",
       "                                       img0268.png  1419.149422   839.189180   \n",
       "                                       img0367.png  1410.092405   598.676681   \n",
       "                                       img0370.png  1410.092405   594.651367   \n",
       "                                       img0433.png  1465.440527   451.752725   \n",
       "                                       img0518.png  1496.636707   307.847766   \n",
       "                                       img0590.png  1545.946798   268.600959   \n",
       "                                       img0780.png  1600.288531   589.619726   \n",
       "                                       img0805.png  1586.199934   598.676681   \n",
       "                                       img0934.png  1588.212590   493.012201   \n",
       "                                       img1000.png  1545.946798   290.740180   \n",
       "                                       img1036.png  1580.161963   133.752952   \n",
       "                                       img1071.png          NaN          NaN   \n",
       "\n",
       "scorer                                                                        \\\n",
       "individuals                                                                    \n",
       "bodyparts                                            dorsal tip                \n",
       "coords                                                        x            y   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1513.744228  1437.954569   \n",
       "                                       img0082.png  1485.567095  1383.612836   \n",
       "                                       img0096.png  1471.478498  1299.081252   \n",
       "                                       img0117.png  1488.586080  1206.499041   \n",
       "                                       img0268.png  1424.181064   893.530913   \n",
       "                                       img0367.png  1411.098733   646.980443   \n",
       "                                       img0370.png  1413.111390   635.910831   \n",
       "                                       img0433.png  1471.478498   489.993204   \n",
       "                                       img0518.png  1485.567095   358.164186   \n",
       "                                       img0590.png  1545.946798   302.816124   \n",
       "                                       img0780.png  1593.244232   639.936145   \n",
       "                                       img0805.png  1590.225247   651.005757   \n",
       "                                       img0934.png  1596.263217   532.259008   \n",
       "                                       img1000.png  1539.908828   335.018629   \n",
       "                                       img1036.png  1571.105008   177.025072   \n",
       "                                       img1071.png          NaN          NaN   \n",
       "\n",
       "scorer                                                                        \\\n",
       "individuals                                                                    \n",
       "bodyparts                                                  tail                \n",
       "coords                                                        x            y   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1474.497421  1685.511352   \n",
       "                                       img0082.png  1499.655692  1607.017738   \n",
       "                                       img0096.png  1500.662021  1529.530452   \n",
       "                                       img0117.png  1495.630379  1383.612836   \n",
       "                                       img0268.png  1438.269661  1050.518141   \n",
       "                                       img0367.png  1447.326555   838.182837   \n",
       "                                       img0370.png  1434.244286   828.119553   \n",
       "                                       img0433.png  1501.668349   679.182940   \n",
       "                                       img0518.png  1518.775932   536.284310   \n",
       "                                       img0590.png  1582.174620   489.993204   \n",
       "                                       img0780.png  1604.313844   804.974000   \n",
       "                                       img0805.png  1624.440412   817.049940   \n",
       "                                       img0934.png  1613.370800   698.303191   \n",
       "                                       img1000.png  1574.123993   482.948901   \n",
       "                                       img1036.png  1618.402442   356.151525   \n",
       "                                       img1071.png  1636.516353   178.031401   \n",
       "\n",
       "scorer                                                                        \\\n",
       "individuals                                                                    \n",
       "bodyparts                                            left fluke                \n",
       "coords                                                        x            y   \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1383.927867  1752.935354   \n",
       "                                       img0082.png  1436.257004  1648.277202   \n",
       "                                       img0096.png  1436.257004  1593.935469   \n",
       "                                       img0117.png  1412.105123  1429.903942   \n",
       "                                       img0268.png  1385.940585  1106.872531   \n",
       "                                       img0367.png  1381.915210   906.613167   \n",
       "                                       img0370.png  1377.889896   897.556211   \n",
       "                                       img0433.png  1462.421542   740.568972   \n",
       "                                       img0518.png  1463.427871   579.556430   \n",
       "                                       img0590.png  1530.851872   556.410877   \n",
       "                                       img0780.png  1551.984768   852.271434   \n",
       "                                       img0805.png  1573.117665   873.404330   \n",
       "                                       img0934.png  1572.111336   754.657581   \n",
       "                                       img1000.png  1517.769603   556.410873   \n",
       "                                       img1036.png  1583.180948   395.398332   \n",
       "                                       img1071.png  1590.225247   218.284536   \n",
       "\n",
       "scorer                                                                        \n",
       "individuals                                                                   \n",
       "bodyparts                                           right fluke               \n",
       "coords                                                        x            y  \n",
       "labeled-data DJI_20231011121723_0076_D img0056.png  1532.864468  1745.891055  \n",
       "                                       img0082.png  1580.161963  1638.213918  \n",
       "                                       img0096.png  1579.155635  1581.859528  \n",
       "                                       img0117.png  1560.035396  1416.821673  \n",
       "                                       img0268.png  1529.845544  1086.745963  \n",
       "                                       img0367.png  1507.706258   884.473942  \n",
       "                                       img0370.png  1505.693601   877.429644  \n",
       "                                       img0433.png  1562.048052   707.360135  \n",
       "                                       img0518.png  1580.161963   580.562759  \n",
       "                                       img0590.png  1637.522681   528.233683  \n",
       "                                       img0780.png  1652.617607   852.271434  \n",
       "                                       img0805.png  1678.782145   843.214478  \n",
       "                                       img0934.png  1665.699876   736.543670  \n",
       "                                       img1000.png  1642.554323   537.290634  \n",
       "                                       img1036.png  1674.756831   402.442631  \n",
       "                                       img1071.png  1697.902384   206.208596  \n",
       "\n",
       "[16 rows x 40 columns]"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "pd.read_hdf('/Users/annateruel/Downloads/CollectedData_Paola.h5')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "dlc_analysis",
   "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": "3.1.-1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}