Offline tracking example.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "import imageio\n",
    "import json\n",
    "import pandas as pd\n",
    "from stytra.experiments.fish_pipelines import pipeline_dict"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "metadata": false,
     "name": "#%% md\n"
    }
   },
   "source": [
    "To use this notebook, tracking parameters for the video have to be extracted. \n",
    "    \n",
    "To do so, you can use the offline version of stytra:\n",
    " \n",
    "1) with stytra installed, run `python -m stytra.offline.track_video` from the anaconda prompt or equivalent\n",
    "\n",
    "2) Adjust the tracking settings (refer to the [documentation](http://www.portugueslab.com/stytra/userguide/3_tracking.html) for hints)\n",
    "\n",
    "3) Select an output folder in the upper right corner of the window.\n",
    "\n",
    "4) Start and stop the \"protocol\", this saves the tracking parameters in the folder specified in the previous step\n",
    "\n",
    "5) Specify the video file and the file ending with _metadata.json generated by the previous step below\n",
    "\n",
    "6) Run through all the cells and use the resulting dataframe for further analysis."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {}
   },
   "source": [
    "Insert the paths here"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "parameter_path = \"\"\n",
    "video_path = \"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "def create_pipeline(json_file):\n",
    "    pipeline_data = json.load(open(mdpath, \"r\"));\n",
    "    pipeline = pipeline_dict(pipeline_data[\"pipeline_type\"])\n",
    "    pipeline.setup()        \n",
    "    paramtree.deserialize(pipeline_data[\"pipeline_params\"])\n",
    "    return pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {},
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "processing_pipeline = create_pipeline(mdpath)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {}
   },
   "source": [
    "# Process all frames"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "from tqdm import tqdm_notebook # install tqdm with pip for a nice progress bar"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "video = imageio.get_reader(video_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {},
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "video.get_length()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {},
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "data = []\n",
    "for fr in tqdm_notebook(video): # if tqdm is not installed, tqdm_notebook has to be be left out\n",
    "    data.append(processing_pipeline.run(fr[:,:,0]).data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "df = pd.DataFrame.from_records(data, columns=data[0]._fields) # creates a DataFrame from a list of named tuples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "%matplotlib widget\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "plt.plot(df.tail_sum)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}