notebooks/02_user_guide/04_tuning_the_trackers.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tuning the trackers\n",
    "\n",
    "Tuning Trackers\n",
    "\n",
    "Tuning the tracking model's hyperparameters is key to good performance. In SportsLabKit, we use the track-by-detection method. This includes a detection model, image model, and motion model, each with its own hyperparameters. These are different from the training parameters or weights. For example:\n",
    "\n",
    "- Detection model: Image size is set to 640x480 pixels\n",
    "- Image model: Image size is set to 800x600 pixels\n",
    "- Motion model: Stride is set to 2 frames\n",
    "\n",
    "You can fine-tune these hyperparameters with optimization methods like Covariance Matrix Adaptation Evolution Strategy (CMA-ES) and Tree-structured Parzen Estimator (TPE).\n",
    "\n",
    "SportsLabKit provides functions to make tuning easier, which is important for better tracking performance. In this notebook, we'll learn how to use `tracker.tune_hparams()` and see how it improves the SORT Tracker."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare dataset\n",
    "\n",
    "First, we'll prepare the dataset. To perform tuning we also need the ground truth tracking data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sportslabkit as slk\n",
    "from sportslabkit.logger import set_log_level\n",
    "\n",
    "dataset_path = slk.datasets.get_path(\"top_view\")\n",
    "path_to_csv = sorted(dataset_path.glob(\"annotations/*.csv\"))[0]\n",
    "path_to_mp4 = sorted(dataset_path.glob(\"videos/*.mp4\"))[0]\n",
    "\n",
    "root = slk.utils.get_git_root()\n",
    "cam = slk.Camera(path_to_mp4)\n",
    "\n",
    "# For the sake of speed, we'll only use the first 10 frames\n",
    "n_frames = 10\n",
    "frames = cam[:n_frames]\n",
    "\n",
    "bbdf_gt = slk.load_df(path_to_csv)\n",
    "# TODO: Hopefully we can get rid of this \n",
    "if bbdf_gt.index[0] == 0:\n",
    "    bbdf_gt.index += 1\n",
    "bbdf_gt = bbdf_gt[:n_frames]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare SORT Tracker\n",
    "\n",
    "Next, we'll set up the tracker to be used, focusing on SORT, a simple yet effective Tracker in SportsLabKit. SORT relies on both a detection model and a motion model, each with specific configurations. Here are the details:\n",
    "\n",
    "* **Detection Model - `YOLOv8x`**\n",
    "    * `conf` - Confidence threshold for detecting objects (Default: 0.5)\n",
    "    * `iou` - Intersection-over-Union threshold for suppressing duplicate detections (Default: 0.3)\n",
    "    * `imgsz` - Image size to which the input is resized, affecting detection (Default: 2560x2560 pixels)\n",
    "\n",
    "\n",
    "* **Motion Model - `KalmanFilterMotionModel`**\n",
    "    * `dt` - Time step between consecutive measurements, crucial for predicting object position (Default: 1/30)\n",
    "    * `process_noise` - Noise in the process model, representing uncertainty in motion prediction (Default: 1e-4)\n",
    "    * `measurement_noise` - Noise in the measurements, representing sensor noise (Default: 1e-1)\n",
    "    * `confidence_scaler` - Factor to scale the confidence in prediction, adjusting the influence of measurements vs predictions (Default: 0.5)\n",
    "\n",
    "* **Tracking Algorithm - `SORTTracker`**\n",
    "    * `metric` - The cost metric to use for assignment\n",
    "    * `metric_gate` - The gate threshold for the cost metric\n",
    "    * `t_lost` - The number of frames to wait before removing a track\n",
    "    * `t_confirm` - The number of frames to wait before confirming a track\n",
    "\n",
    "These configurations are essential in setting up the SORT Tracker, and the given defaults provide a good starting point. They can be tuned further as needed to align with the specific use case and environment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sportslabkit.mot import SORTTracker\n",
    "\n",
    "slk.logger.set_log_level('INFO')\n",
    "det_model = slk.detection_model.load(\n",
    "    model_name='yolov8',\n",
    "    model=root/'models/yolov8/soccer_top_view-model=yolov8x-imgsz=2048.pt',\n",
    "    conf=0.25,\n",
    "    iou=0.6,\n",
    "    imgsz=960,\n",
    "    device='mps',\n",
    "    classes=0,\n",
    "    augment=True,\n",
    "    max_det=35\n",
    ")\n",
    "\n",
    "motion_model = slk.motion_model.load(\n",
    "    model_name='kalmanfilter',\n",
    "    dt=1/30,\n",
    "    process_noise=500,\n",
    "    measurement_noise=10,\n",
    "    confidence_scaler=1\n",
    ")\n",
    "\n",
    "matching_fn = slk.matching.SimpleMatchingFunction(\n",
    "    metric=slk.metrics.EuclideanCMM(use_pred_box=True),\n",
    "    gate=0.5\n",
    ")\n",
    "\n",
    "tracker = SORTTracker(\n",
    "    detection_model=det_model,\n",
    "    motion_model=motion_model,\n",
    "    matching_fn=matching_fn,\n",
    "    max_staleness=2,\n",
    "    min_length=2\n",
    ")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Tracking normally with SORT"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Tracking Progress: 100%|██████████| 10/10 [00:45<00:00,  4.60s/it, Active: 28, Dead: 0]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "HOTA score: 0.48\n"
     ]
    }
   ],
   "source": [
    "from sportslabkit.metrics import hota_score\n",
    "\n",
    "bbdf = tracker.track(frames)\n",
    "score = hota_score(bboxes_track=bbdf, bboxes_gt=bbdf_gt)[\"HOTA\"]\n",
    "\n",
    "print(f'HOTA score: {score:.2f}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Writing video: 100it [00:10,  9.80it/s]\n"
     ]
    }
   ],
   "source": [
    "bbdf.visualize_frames(cam.video_path, root/'notebooks/02_user_guide/assets/sort_not_tuned.mp4')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Tuning\n",
    "\n",
    "Each tunable component will have an associated class variable named `hparam_search_space`. This is a nested dictionary that contains key-value pairs in the format `param: {\"type\": \"categorical\", \"values\": list of possible values}` for a categorical search space, or `param: {\"type\": 'logfloat', \"low\": min value, \"high\": max value}` for a continuous search space.\n",
    "\n",
    "For instance, a `KalmanFilterMotionModel` has the following hyperparameter search space:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'dt': {'type': 'categorical',\n",
       "  'values': [10,\n",
       "   2,\n",
       "   1,\n",
       "   0.03333333333333333,\n",
       "   0.016666666666666666,\n",
       "   0.008333333333333333]},\n",
       " 'process_noise': {'type': 'logfloat', 'low': 1e-06, 'high': 100.0},\n",
       " 'measurement_noise': {'type': 'logfloat', 'low': 0.001, 'high': 100.0},\n",
       " 'confidence_scaler': {'type': 'logfloat', 'low': 0.001, 'high': 100}}"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "slk.motion_model.KalmanFilter.hparam_search_space"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We use [Optuna](https://optuna.org) to \n",
    "\n",
    "Each tracker object has a `tracker.create_hparam_dict()` method that generates a dictionary containing hyperparameter search spaces for each component. You can pass a dictionary into the `hparam_search_space` argument to overwrite the default search space. \n",
    "\n",
    "The `hparam_search_space` dictionary should be formatted as follows:\n",
    "\n",
    "```\n",
    "{\n",
    "    'component name (motion_model/detection_model/image_model etc.)`: {\n",
    "        'param name (imgsz/conf/t_lost etc.)': \n",
    "        {\"type\": 'logfloat', \"low\": min value, \"high\": max value} -> for continuous params\n",
    "        or \n",
    "        {\"type\": 'categorical', \"low\": min value, \"high\": max value} -> for categorical params\n",
    "    }\n",
    "}\n",
    "```\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[36mIPython:0004  💬| self: \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \tmetric_gate: {'type': 'float', 'low': 0.01, 'high': 1} \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \tt_lost: {'type': 'int', 'low': 1, 'high': 1000.0} \u001b[0m\n",
      "\u001b[36mIPython:0004  💬| detection_model: \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \tmax_det: {'type': 'int', 'values': None, 'low': 20, 'high': 50} \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \timgsz: {'type': 'int', 'values': None, 'low': 1280, 'high': 3840} \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \tconf: {'type': 'float', 'values': None, 'low': 0.1, 'high': 1.0} \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \tiou: {'type': 'float', 'values': None, 'low': 0.1, 'high': 1.0} \u001b[0m\n",
      "\u001b[36mIPython:0004  💬| motion_model: \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \tdt: {'type': 'categorical', 'values': [10, 2, 1, 0.03333333333333333, 0.016666666666666666, 0.008333333333333333], 'low': None, 'high': None} \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \tprocess_noise: {'type': 'logfloat', 'values': None, 'low': 1e-06, 'high': 100.0} \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \tmeasurement_noise: {'type': 'logfloat', 'values': None, 'low': 0.001, 'high': 100.0} \u001b[0m\n",
      "\u001b[36mIPython:0006  💬| \tconfidence_scaler: {'type': 'logfloat', 'values': None, 'low': 0.001, 'high': 100} \u001b[0m\n"
     ]
    }
   ],
   "source": [
    "from sportslabkit.logger import logger\n",
    "hparams = tracker.create_hparam_dict()\n",
    "for attribute, param_space in hparams.items():\n",
    "    logger.info(f\"{attribute}:\")\n",
    "    for param_name, param_values in param_space.items():\n",
    "        logger.info(f\"\\t{param_name}: {param_values}\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define our own search space:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "hparam_search_space = {\n",
    "    'self': {\n",
    "        'max_staleness': {'type': 'int', 'low': 10, 'high': 100},\n",
    "        'min_length': {'type': 'int', 'low': 10, 'high': 100},\n",
    "    },\n",
    "    'motion_model':{\n",
    "        'process_noise': {'type': 'float', 'low': 100,'high': 1000},\n",
    "        'measurement_noise': {'type': 'float','low': 1,'high': 100},\n",
    "        'confidence_scaler': {'type': 'logfloat','low': 1e-3, 'high': 1e3}\n",
    "    },\n",
    "    'detection_model':{\n",
    "        \"max_det\": {\"type\": \"int\", \"low\": 20, \"high\": 50},\n",
    "        \"conf\": {\"type\": \"float\", \"low\": 0.1, \"high\": .6},\n",
    "        \"iou\": {\"type\": \"float\", \"low\": 0.3, \"high\": .5},\n",
    "    },\n",
    "    'matching_fn':{\n",
    "        'gate': {'type': 'logfloat', 'low': 1e-4, 'high': 1}\n",
    "    }\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[36mtune_hparams:0253  💬| Hyperparameter search space: \u001b[0m\n",
      "\u001b[36mtune_hparams:0255  💬| self: \u001b[0m\n",
      "\u001b[36mtune_hparams:0257  💬| \tmax_staleness: {'type': 'int', 'low': 10, 'high': 100} \u001b[0m\n",
      "\u001b[36mtune_hparams:0257  💬| \tmin_length: {'type': 'int', 'low': 10, 'high': 100} \u001b[0m\n",
      "\u001b[36mtune_hparams:0255  💬| motion_model: \u001b[0m\n",
      "\u001b[36mtune_hparams:0257  💬| \tprocess_noise: {'type': 'float', 'low': 100, 'high': 1000} \u001b[0m\n",
      "\u001b[36mtune_hparams:0257  💬| \tmeasurement_noise: {'type': 'float', 'low': 1, 'high': 100} \u001b[0m\n",
      "\u001b[36mtune_hparams:0257  💬| \tconfidence_scaler: {'type': 'logfloat', 'low': 0.001, 'high': 1000.0} \u001b[0m\n",
      "\u001b[36mtune_hparams:0255  💬| detection_model: \u001b[0m\n",
      "\u001b[36mtune_hparams:0257  💬| \tmax_det: {'type': 'int', 'low': 20, 'high': 50} \u001b[0m\n",
      "\u001b[36mtune_hparams:0257  💬| \tconf: {'type': 'float', 'low': 0.1, 'high': 0.6} \u001b[0m\n",
      "\u001b[36mtune_hparams:0257  💬| \tiou: {'type': 'float', 'low': 0.3, 'high': 0.5} \u001b[0m\n",
      "\u001b[36mtune_hparams:0255  💬| matching_fn: \u001b[0m\n",
      "\u001b[36mtune_hparams:0257  💬| \tgate: {'type': 'logfloat', 'low': 0.0001, 'high': 1} \u001b[0m\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[I 2023-08-21 14:49:14,330] A new study created in memory with name: no-name-baffd161-43db-4814-8d65-dbed99038e62\n",
      "Tracking Progress:   0%|          | 0/100 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Tracking Progress: 100%|██████████| 100/100 [00:44<00:00,  2.24it/s, Active: 1704, Dead: 0]\n",
      "[W 2023-08-21 14:49:59,072] Trial 0 failed with parameters: {'max_staleness': 81, 'min_length': 26, 'process_noise': 107.45501951679715, 'measurement_noise': 14.47835048351415, 'confidence_scaler': 20.61163322052972, 'max_det': 46, 'conf': 0.48191798620983384, 'iou': 0.46148564631937755, 'gate': 0.00021187769045602145} because of the following error: The value nan is not acceptable.\n",
      "[W 2023-08-21 14:49:59,073] Trial 0 failed with value nan.\n",
      "Tracking Progress: 100%|██████████| 100/100 [00:33<00:00,  2.96it/s, Active: 22, Dead: 0]\n",
      "[I 2023-08-21 14:50:33,332] Trial 1 finished with value: 0.672605557674046 and parameters: {'max_staleness': 79, 'min_length': 33, 'process_noise': 114.86857770421655, 'measurement_noise': 97.41671292185173, 'confidence_scaler': 14.990523117527346, 'max_det': 29, 'conf': 0.5964694336761054, 'iou': 0.30773303095524673, 'gate': 0.5593959171041554}. Best is trial 1 with value: 0.672605557674046.\n",
      "Tracking Progress: 100%|██████████| 100/100 [00:34<00:00,  2.92it/s, Active: 76, Dead: 6]\n",
      "[I 2023-08-21 14:51:08,026] Trial 2 finished with value: 0.49012696883173557 and parameters: {'max_staleness': 42, 'min_length': 22, 'process_noise': 189.46016805819926, 'measurement_noise': 31.2345719536671, 'confidence_scaler': 0.1533187596159073, 'max_det': 41, 'conf': 0.4713590856174691, 'iou': 0.42087058246153164, 'gate': 0.017923491188670263}. Best is trial 1 with value: 0.672605557674046.\n",
      "Tracking Progress: 100%|██████████| 100/100 [00:34<00:00,  2.92it/s, Active: 65, Dead: 0]\n",
      "[I 2023-08-21 14:51:42,666] Trial 3 finished with value: 0.5550200949914603 and parameters: {'max_staleness': 46, 'min_length': 59, 'process_noise': 583.7538967090054, 'measurement_noise': 68.3898337204985, 'confidence_scaler': 0.5655957468833475, 'max_det': 40, 'conf': 0.4902223916312901, 'iou': 0.39179235266883816, 'gate': 0.016646672110167793}. Best is trial 1 with value: 0.672605557674046.\n",
      "Tracking Progress: 100%|██████████| 100/100 [00:34<00:00,  2.94it/s, Active: 24, Dead: 0]\n",
      "[I 2023-08-21 14:52:17,208] Trial 4 finished with value: 0.7272012391795679 and parameters: {'max_staleness': 70, 'min_length': 80, 'process_noise': 771.3671783689742, 'measurement_noise': 4.131281955898294, 'confidence_scaler': 0.10322804443222602, 'max_det': 37, 'conf': 0.41699959275723997, 'iou': 0.43222379993994675, 'gate': 0.026828469275201567}. Best is trial 4 with value: 0.7272012391795679.\n",
      "Tracking Progress: 100%|██████████| 100/100 [00:34<00:00,  2.90it/s, Active: 87, Dead: 0]\n",
      "[I 2023-08-21 14:52:52,077] Trial 5 pruned. \n",
      "Tracking Progress: 100%|██████████| 100/100 [00:34<00:00,  2.92it/s, Active: 25, Dead: 0]\n",
      "[I 2023-08-21 14:53:26,824] Trial 6 pruned. \n",
      "Tracking Progress: 100%|██████████| 100/100 [00:34<00:00,  2.86it/s, Active: 137, Dead: 0]\n",
      "[I 2023-08-21 14:54:02,176] Trial 7 pruned. \n",
      "Tracking Progress: 100%|██████████| 100/100 [00:34<00:00,  2.89it/s, Active: 22, Dead: 0]\n",
      "[I 2023-08-21 14:54:37,196] Trial 8 finished with value: 0.722826489435095 and parameters: {'max_staleness': 31, 'min_length': 77, 'process_noise': 552.7464524051331, 'measurement_noise': 25.307681861610796, 'confidence_scaler': 1.6190280098477405, 'max_det': 46, 'conf': 0.4917345548523706, 'iou': 0.4781606695387891, 'gate': 0.022745772672614288}. Best is trial 4 with value: 0.7272012391795679.\n",
      "Tracking Progress: 100%|██████████| 100/100 [00:34<00:00,  2.94it/s, Active: 27, Dead: 0]\n",
      "[I 2023-08-21 14:55:11,702] Trial 9 pruned. \n"
     ]
    }
   ],
   "source": [
    "import optuna\n",
    "\n",
    "sampler = optuna.samplers.CmaEsSampler()\n",
    "pruner = optuna.pruners.HyperbandPruner()\n",
    "best_params, best_hota, study = tracker.tune_hparams(\n",
    "    frames_list=[frames],\n",
    "    bbdf_gt_list=[bbdf_gt],\n",
    "    n_trials=10,\n",
    "    reuse_detections=False,\n",
    "    hparam_search_space=hparam_search_space,\n",
    "    verbose=True,\n",
    "    sampler=sampler,\n",
    "    pruner=pruner,\n",
    "    return_study=True,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best HOTA: 0.7272012391795679\n",
      "max_staleness: 70\n",
      "min_length: 80\n",
      "process_noise: 771.3671783689742\n",
      "measurement_noise: 4.131281955898294\n",
      "confidence_scaler: 0.10322804443222602\n",
      "max_det: 37\n",
      "conf: 0.41699959275723997\n",
      "iou: 0.43222379993994675\n",
      "gate: 0.026828469275201567\n"
     ]
    }
   ],
   "source": [
    "print(f\"Best HOTA: {best_hota}\")\n",
    "for key in best_params:\n",
    "    print(f\"{key}: {best_params[key]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/40437371.1.gpu/ipykernel_3186863/1036994904.py:1: ExperimentalWarning: plot_param_importances is experimental (supported from v2.2.0). The interface can change in the future.\n",
      "  optuna.visualization.matplotlib.plot_param_importances(study)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Axes: title={'center': 'Hyperparameter Importances'}, xlabel='Importance for Objective Value', ylabel='Hyperparameter'>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "optuna.visualization.matplotlib.plot_param_importances(study)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/40437371.1.gpu/ipykernel_3186863/1455980257.py:1: ExperimentalWarning: plot_optimization_history is experimental (supported from v2.2.0). The interface can change in the future.\n",
      "  optuna.visualization.matplotlib.plot_optimization_history(study)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Axes: title={'center': 'Optimization History Plot'}, xlabel='Trial', ylabel='Objective Value'>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "optuna.visualization.matplotlib.plot_optimization_history(study)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'best_params' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[14], line 4\u001b[0m\n\u001b[1;32m      1\u001b[0m det_model \u001b[39m=\u001b[39m slk\u001b[39m.\u001b[39mdetection_model\u001b[39m.\u001b[39mload(\n\u001b[1;32m      2\u001b[0m     model_name\u001b[39m=\u001b[39m\u001b[39m'\u001b[39m\u001b[39myolov8\u001b[39m\u001b[39m'\u001b[39m,\n\u001b[1;32m      3\u001b[0m     model\u001b[39m=\u001b[39mroot\u001b[39m/\u001b[39m\u001b[39m'\u001b[39m\u001b[39mmodels/yolov8/soccer_top_view-model=yolov8x-imgsz=2048.pt\u001b[39m\u001b[39m'\u001b[39m,\n\u001b[0;32m----> 4\u001b[0m     conf\u001b[39m=\u001b[39mbest_params[\u001b[39m'\u001b[39m\u001b[39mconf\u001b[39m\u001b[39m'\u001b[39m],\n\u001b[1;32m      5\u001b[0m     iou\u001b[39m=\u001b[39mbest_params[\u001b[39m'\u001b[39m\u001b[39miou\u001b[39m\u001b[39m'\u001b[39m],\n\u001b[1;32m      6\u001b[0m     imgsz\u001b[39m=\u001b[39m\u001b[39m2048\u001b[39m,\n\u001b[1;32m      7\u001b[0m     device\u001b[39m=\u001b[39m\u001b[39m0\u001b[39m,\n\u001b[1;32m      8\u001b[0m     classes\u001b[39m=\u001b[39m\u001b[39m0\u001b[39m,\n\u001b[1;32m      9\u001b[0m     augment\u001b[39m=\u001b[39m\u001b[39mTrue\u001b[39;00m,\n\u001b[1;32m     10\u001b[0m     max_det\u001b[39m=\u001b[39mbest_params[\u001b[39m'\u001b[39m\u001b[39mmax_det\u001b[39m\u001b[39m'\u001b[39m],\n\u001b[1;32m     11\u001b[0m )\n\u001b[1;32m     13\u001b[0m motion_model \u001b[39m=\u001b[39m slk\u001b[39m.\u001b[39mmotion_model\u001b[39m.\u001b[39mload(\n\u001b[1;32m     14\u001b[0m     model_name\u001b[39m=\u001b[39m\u001b[39m'\u001b[39m\u001b[39mkalmanfilter\u001b[39m\u001b[39m'\u001b[39m,\n\u001b[1;32m     15\u001b[0m     dt\u001b[39m=\u001b[39m\u001b[39m1\u001b[39m\u001b[39m/\u001b[39m\u001b[39m30\u001b[39m,\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m     18\u001b[0m     confidence_scaler\u001b[39m=\u001b[39mbest_params[\u001b[39m'\u001b[39m\u001b[39mconfidence_scaler\u001b[39m\u001b[39m'\u001b[39m],\n\u001b[1;32m     19\u001b[0m )\n\u001b[1;32m     21\u001b[0m matching_fn \u001b[39m=\u001b[39m slk\u001b[39m.\u001b[39mmatching\u001b[39m.\u001b[39mSimpleMatchingFunction(\n\u001b[1;32m     22\u001b[0m     metric\u001b[39m=\u001b[39mslk\u001b[39m.\u001b[39mmetrics\u001b[39m.\u001b[39mEuclideanCMM(use_pred_box\u001b[39m=\u001b[39m\u001b[39mTrue\u001b[39;00m),\n\u001b[1;32m     23\u001b[0m     gate\u001b[39m=\u001b[39mbest_params[\u001b[39m'\u001b[39m\u001b[39mgate\u001b[39m\u001b[39m'\u001b[39m],\n\u001b[1;32m     24\u001b[0m )\n",
      "\u001b[0;31mNameError\u001b[0m: name 'best_params' is not defined"
     ]
    }
   ],
   "source": [
    "det_model = slk.detection_model.load(\n",
    "    model_name='yolov8',\n",
    "    model=root/'models/yolov8/soccer_top_view-model=yolov8x-imgsz=2048.pt',\n",
    "    conf=best_params['conf'],\n",
    "    iou=best_params['iou'],\n",
    "    imgsz=2048,\n",
    "    device=0,\n",
    "    classes=0,\n",
    "    augment=True,\n",
    "    max_det=best_params['max_det'],\n",
    ")\n",
    "\n",
    "motion_model = slk.motion_model.load(\n",
    "    model_name='kalmanfilter',\n",
    "    dt=1/30,\n",
    "    process_noise=best_params['process_noise'],\n",
    "    measurement_noise=best_params['measurement_noise'],\n",
    "    confidence_scaler=best_params['confidence_scaler'],\n",
    ")\n",
    "\n",
    "matching_fn = slk.matching.SimpleMatchingFunction(\n",
    "    metric=slk.metrics.EuclideanCMM(use_pred_box=True),\n",
    "    gate=best_params['gate'],\n",
    ")\n",
    "\n",
    "tracker = SORTTracker(\n",
    "    detection_model=det_model,\n",
    "    motion_model=motion_model,\n",
    "    matching_fn=matching_fn,\n",
    "    max_staleness=best_params['max_staleness'],\n",
    "    min_length=best_params['min_length'],\n",
    ")\n",
    "\n",
    "bbdf = tracker.track(frames)\n",
    "score = hota_score(bboxes_track=bbdf, bboxes_gt=bbdf_gt)[\"HOTA\"]\n",
    "\n",
    "print(f'HOTA score: {score:.2f}')\n",
    "bbdf.visualize_frames(cam.video_path, root/'notebooks/02_user_guide/assets/sort_tuned.mp4')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[36mtune_hparams:0260  💬| Hyperparameter search space: \u001b[0m\n",
      "\u001b[36mtune_hparams:0262  💬| self: \u001b[0m\n",
      "\u001b[36mtune_hparams:0264  💬| \tmax_staleness: {'type': 'int', 'low': 2, 'high': 3} \u001b[0m\n",
      "\u001b[36mtune_hparams:0264  💬| \tmin_length: {'type': 'int', 'low': 2, 'high': 3} \u001b[0m\n",
      "\u001b[36mtune_hparams:0262  💬| motion_model: \u001b[0m\n",
      "\u001b[36mtune_hparams:0264  💬| \tprocess_noise: {'type': 'float', 'low': 100, 'high': 1000} \u001b[0m\n",
      "\u001b[36mtune_hparams:0264  💬| \tmeasurement_noise: {'type': 'float', 'low': 1, 'high': 100} \u001b[0m\n",
      "\u001b[36mtune_hparams:0264  💬| \tconfidence_scaler: {'type': 'logfloat', 'low': 0.001, 'high': 1000.0} \u001b[0m\n",
      "\u001b[36mtune_hparams:0262  💬| matching_fn: \u001b[0m\n",
      "\u001b[36mtune_hparams:0264  💬| \tgate: {'type': 'logfloat', 'low': 0.0001, 'high': 1} \u001b[0m\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 81.75it/s, Active: 29, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 257.31it/s, Active: 31, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 256.87it/s, Active: 31, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 225.70it/s, Active: 37, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 197.78it/s, Active: 36, Dead: 9]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 303.87it/s, Active: 28, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 240.29it/s, Active: 28, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 261.55it/s, Active: 31, Dead: 3]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 230.63it/s, Active: 43, Dead: 27]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 210.27it/s, Active: 58, Dead: 13]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 222.77it/s, Active: 46, Dead: 12]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 275.76it/s, Active: 29, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 210.66it/s, Active: 41, Dead: 18]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 252.02it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 164.27it/s, Active: 33, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 274.27it/s, Active: 28, Dead: 1]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 181.74it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 163.43it/s, Active: 73, Dead: 17]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 276.04it/s, Active: 28, Dead: 0]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 221.45it/s, Active: 28, Dead: 5]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 272.39it/s, Active: 29, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 243.89it/s, Active: 34, Dead: 16]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 246.21it/s, Active: 31, Dead: 8]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 179.63it/s, Active: 61, Dead: 27]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 180.91it/s, Active: 32, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 179.69it/s, Active: 29, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 232.65it/s, Active: 28, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 269.73it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 302.90it/s, Active: 28, Dead: 0]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 258.39it/s, Active: 29, Dead: 3]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 247.56it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 276.44it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 227.75it/s, Active: 28, Dead: 5]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 260.78it/s, Active: 34, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 288.87it/s, Active: 28, Dead: 0]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 281.14it/s, Active: 29, Dead: 7]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 263.57it/s, Active: 29, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 229.79it/s, Active: 28, Dead: 0]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 279.70it/s, Active: 29, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 260.50it/s, Active: 29, Dead: 7]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 251.29it/s, Active: 29, Dead: 3]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 198.29it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 270.07it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 312.77it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 137.09it/s, Active: 82, Dead: 8]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 278.10it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 246.07it/s, Active: 31, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 267.34it/s, Active: 28, Dead: 2]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 250.11it/s, Active: 28, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 272.11it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 225.43it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 257.83it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 246.97it/s, Active: 32, Dead: 13]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 278.52it/s, Active: 31, Dead: 3]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 277.31it/s, Active: 28, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 260.93it/s, Active: 28, Dead: 5]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 213.48it/s, Active: 33, Dead: 16]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 258.83it/s, Active: 31, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 126.17it/s, Active: 32, Dead: 10]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 153.52it/s, Active: 31, Dead: 11]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 225.06it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 251.83it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 245.59it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 256.02it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 200.99it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 245.57it/s, Active: 28, Dead: 5]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 280.45it/s, Active: 28, Dead: 0]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 247.29it/s, Active: 28, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 254.99it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 207.51it/s, Active: 49, Dead: 15]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 207.96it/s, Active: 34, Dead: 14]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 246.51it/s, Active: 28, Dead: 0]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 276.70it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 242.96it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 301.50it/s, Active: 28, Dead: 0]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 244.82it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 248.67it/s, Active: 32, Dead: 11]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 211.82it/s, Active: 31, Dead: 15]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 296.05it/s, Active: 28, Dead: 0]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 97.62it/s, Active: 28, Dead: 2]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 254.34it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 235.23it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 235.44it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 288.08it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 208.10it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 276.03it/s, Active: 28, Dead: 2]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 253.04it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 270.49it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 240.27it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 103.67it/s, Active: 28, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 262.68it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 218.52it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 179.44it/s, Active: 28, Dead: 6]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 251.74it/s, Active: 28, Dead: 0]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 258.38it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 244.56it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 275.63it/s, Active: 28, Dead: 1]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 257.28it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 276.14it/s, Active: 28, Dead: 4]\n",
      "Tracking Progress: 100%|██████████| 10/10 [00:00<00:00, 279.47it/s, Active: 28, Dead: 4]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best HOTA: 0.49278041397454936\n",
      "self: {'max_staleness': 2, 'min_length': 3}\n",
      "motion_model: {'process_noise': 457.8303187474995, 'measurement_noise': 46.663412151089474, 'confidence_scaler': 0.9561528737259082}\n",
      "matching_fn: {'gate': 0.010461641958828564}\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "import optuna\n",
    "\n",
    "hparam_search_space = {\n",
    "    'self': {\n",
    "        'max_staleness': {'type': 'int', 'low': 2, 'high': 3},\n",
    "        'min_length': {'type': 'int', 'low': 2, 'high': 3},\n",
    "    },\n",
    "    'motion_model':{\n",
    "        'process_noise': {'type': 'float', 'low': 100,'high': 1000},\n",
    "        'measurement_noise': {'type': 'float','low': 1,'high': 100},\n",
    "        'confidence_scaler': {'type': 'logfloat','low': 1e-3, 'high': 1e3}\n",
    "    },\n",
    "    'matching_fn':{\n",
    "        'gate': {'type': 'logfloat', 'low': 1e-4, 'high': 1}\n",
    "    }\n",
    "}\n",
    "\n",
    "sampler = optuna.samplers.CmaEsSampler()\n",
    "pruner = optuna.pruners.HyperbandPruner()\n",
    "best_params, best_hota, study = tracker.tune_hparams(\n",
    "    frames_list=[frames],\n",
    "    bbdf_gt_list=[bbdf_gt],\n",
    "    n_trials=10,\n",
    "    reuse_detections=True, # Changed to TRUE to reuse detections\n",
    "    hparam_search_space=hparam_search_space,\n",
    "    verbose=False, # Changed to False to quiet down the output\n",
    "    sampler=sampler,\n",
    "    pruner=pruner,\n",
    "    return_study=True,\n",
    ")\n",
    "print(f\"Best HOTA: {best_hota}\")\n",
    "for key in best_params:\n",
    "    print(f\"{key}: {best_params[key]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "tracker.apply_hyperparameters(tracker.trial_params[0])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.12"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}