Towards-Realtime-MOT-Vehicle-Tracking.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Towards_Realtime_MOT_Vehicle_Tracking.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/hardik0/AI-basketball-analysis-on-google-colab/blob/master/Towards-Realtime-MOT-Vehicle-Tracking.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UaFgmAfi0FPS"
      },
      "source": [
        "# Towards Realtime MOT - Vehicle Tracking\n",
        "Joint Detection and Embedding (JDE) model. JDE is a fast and high-performance multiple-object tracker that learns the object detection task and appearance embedding task simutaneously in a shared neural network. Techical details are described in [ECCV 2020 paper](https://arxiv.org/pdf/1909.12605v1.pdf). \n",
        "\n",
        "**Original code (Pedestrian):** [Towards-Realtime-MOT](https://github.com/Zhongdao/Towards-Realtime-MOT) by [@ZhongdaoWang](https://github.com/Zhongdao)\n",
        "\n",
        "**[Modified version] Vehicle tracking code (UA-DETRAC):** [Towards-Realtime-MOT:detrac](https://github.com/hardik0/Towards-Realtime-MOT/tree/detrac)\n",
        "\n",
        "### Citation\n",
        "```\n",
        "@article{wang2019towards,\n",
        "  title={Towards Real-Time Multi-Object Tracking},\n",
        "  author={Wang, Zhongdao and Zheng, Liang and Liu, Yixuan and Wang, Shengjin},\n",
        "  journal={arXiv preprint arXiv:1909.12605},\n",
        "  year={2019}\n",
        "}\n",
        "```"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k6BEOjSAJOyD"
      },
      "source": [
        "! nvidia-smi\n",
        "! /usr/local/cuda/bin/nvcc --version"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oBPmoyNt3cBd"
      },
      "source": [
        "To access Google Drive Folder and Files"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mR-YYiITJU1Y",
        "outputId": "8f2cd64b-8a01-4854-edb6-bb1819f8566e",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/gdrive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wd9BMrhy3iXL"
      },
      "source": [
        "## Installation\n",
        "clone this repository"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ncF66MRnKSye",
        "outputId": "c3787782-423f-4828-beba-b46ffef28008",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "#! git clone https://github.com/hardik0/Towards-Realtime-MOT\n",
        "! git clone -b detrac https://github.com/hardik0/Towards-Realtime-MOT\n",
        "%cd Towards-Realtime-MOT"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'Towards-Realtime-MOT'...\n",
            "remote: Enumerating objects: 407, done.\u001b[K\n",
            "remote: Total 407 (delta 0), reused 0 (delta 0), pack-reused 407\u001b[K\n",
            "Receiving objects: 100% (407/407), 21.42 MiB | 50.41 MiB/s, done.\n",
            "Resolving deltas: 100% (246/246), done.\n",
            "/content/Towards-Realtime-MOT\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KX3v5qi93val"
      },
      "source": [
        "check torch version"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1xxzjiQQOWjY",
        "outputId": "b19c73a7-221a-4001-ef30-696b249deeaa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "import torch\n",
        "torch.__version__"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'1.7.0+cu101'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "K-MNsE9j3zQg"
      },
      "source": [
        "Downgrade the PyTorch version to 1.2"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zgMsawkwNquN"
      },
      "source": [
        "# uncomment below line ONLY if you run into any dependency issues while following\n",
        "#! pip install torch==1.2.0 torchvision==0.4.0"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q4_HhHQcXOlt"
      },
      "source": [
        "import torch\n",
        "torch.__version__"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SoI_dzCS38NW"
      },
      "source": [
        "Install the Dependencies"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YEuhvGJ0KZIU"
      },
      "source": [
        "! pip install motmetrics cython_bbox ffmpeg lap pillow"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hOrr4qHe4DsY"
      },
      "source": [
        "restart the runtime if required"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fGwI9Ugi09tp",
        "outputId": "f2c11f9c-e57e-40ac-ed65-db641c2caa60",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# change dir after restart\n",
        "%cd Towards-Realtime-MOT/"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/Towards-Realtime-MOT\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RNJThXYm5Fo6"
      },
      "source": [
        "## Dataset preparation"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ijbw1gMrSHxw"
      },
      "source": [
        "### Download DETRAC dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y6KZeLt9ViDe"
      },
      "source": [
        "%%shell\n",
        "\n",
        "wget -O DETRAC-train-data.zip http://detrac-db.rit.albany.edu/Data/DETRAC-train-data.zip\n",
        "#wget -O DETRAC-test-data.zip http://detrac-db.rit.albany.edu/Data/DETRAC-test-data.zip\n",
        "wget -O DETRAC-Train-Annotations-XML.zip http://detrac-db.rit.albany.edu/Data/DETRAC-Train-Annotations-XML.zip\n",
        "mkdir data/MOT\n",
        "unzip -q DETRAC-train-data.zip -d data/\n",
        "#unzip -q DETRAC-test-data.zip -d data/\n",
        "unzip -q DETRAC-Train-Annotations-XML.zip -d data/\n",
        "#rm -r *.zip"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "s1dmAhI35aHl"
      },
      "source": [
        "### Data Format\n",
        "All the datasets have the following structure:\n",
        "\n",
        "```\n",
        "Caltech\n",
        "   |——————images\n",
        "   |        └——————00001.jpg\n",
        "   |        |—————— ...\n",
        "   |        └——————0000N.jpg\n",
        "   └——————labels_with_ids\n",
        "            └——————00001.txt\n",
        "            |—————— ...\n",
        "            └——————0000N.txt\n",
        "```\n",
        "\n",
        "Every image has a corresponding annotation text. Given an image path, the annotation text path can be generated by replacing the string `images` with `labels_with_ids` and replacing `.jpg` with `.txt`.\n",
        "\n",
        "In the annotation text, each line is describing a bounding box and has the following format:\n",
        "\n",
        "`[class] [identity] [x_center] [y_center] [width] [height]`\n",
        "\n",
        "The field `[class]` should be `0`. Only single-class multi-object tracking is supported in this version.\n",
        "\n",
        "The field `[identity]` is an integer from `0` to `num_identities - 1`, or `-1` if this box has no identity annotation.\n",
        "\n",
        "***Note** that the values of `[x_center] [y_center] [width] [height]` are normalized by the width/height of the image, so they are floating point numbers ranging from 0 to 1."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bUeqoz2PWB4a"
      },
      "source": [
        "Script to convert DETRAC tracking dataset to MOT format"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-_TlOWAEjFhy"
      },
      "source": [
        "# will take about 25-30 minutes \n",
        "! python gen_labels_detrac.py > gen_labels_detrac.log 2>&1\n",
        "#! python gen_labels_detrac.py"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DMIfSrRCjLfS"
      },
      "source": [
        "! tail gen_labels_detrac.log"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lCmH8NNkHfKb",
        "outputId": "cd435ef1-8c35-473c-9285-afde517ce970",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        }
      },
      "source": [
        "! head data/detrac.train"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "DETRAC/images/train/MVI_20011/img1/img00001.jpg\n",
            "DETRAC/images/train/MVI_20011/img1/img00002.jpg\n",
            "DETRAC/images/train/MVI_20011/img1/img00003.jpg\n",
            "DETRAC/images/train/MVI_20011/img1/img00004.jpg\n",
            "DETRAC/images/train/MVI_20011/img1/img00005.jpg\n",
            "DETRAC/images/train/MVI_20011/img1/img00006.jpg\n",
            "DETRAC/images/train/MVI_20011/img1/img00007.jpg\n",
            "DETRAC/images/train/MVI_20011/img1/img00008.jpg\n",
            "DETRAC/images/train/MVI_20011/img1/img00009.jpg\n",
            "DETRAC/images/train/MVI_20011/img1/img00010.jpg\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6mNpf8qPjOyY",
        "outputId": "ec3250a1-76fe-4524-8979-146ddc40d832",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "source": [
        "! du -shc data/MOT/DETRAC/*"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "9.2G\tdata/MOT/DETRAC/images\n",
            "324M\tdata/MOT/DETRAC/labels_with_ids\n",
            "9.5G\ttotal\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BLC4OZCV6zAg"
      },
      "source": [
        "Remove unnecessary files(optional)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "T8re5GhLjZDs"
      },
      "source": [
        "#%rm -r DETRAC-train-data.zip\n",
        "%rm -r data/Insight-MVT_Annotation_Train\n",
        "%rm -r data/DETRAC-Train-Annotations-XML"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "N9L1OslX7LeY"
      },
      "source": [
        "zip processed dataset for further use"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rsfC5pfalXn_"
      },
      "source": [
        "%%shell\n",
        "\n",
        "cd data/MOT\n",
        "zip -qr DETRAC.zip DETRAC/"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-YGg5lIo7cCR"
      },
      "source": [
        "copy zip file to gdrive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_7s4nwo1l4jN"
      },
      "source": [
        "! cp -r data/MOT/DETRAC.zip '../gdrive/My Drive/Colab Notebooks/MOT/'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C53xzgIG9oxj"
      },
      "source": [
        "! ls '../gdrive/My Drive/Colab Notebooks/MOT/'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "isBhw8bz7w0d"
      },
      "source": [
        "copy dataset from google drive | using google_utils script "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y7v6nivdqalx"
      },
      "source": [
        "! cp -r '../gdrive/My Drive/Colab Notebooks/MOT/DETRAC.zip' data/"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "s4aLlBM18ig7"
      },
      "source": [
        "Share files from Google Drive check [this](https://support.google.com/drive/answer/2494822?co=GENIE.Platform%3DDesktop&hl=en) page & copy File ID"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QG0KrXNoG68e"
      },
      "source": [
        "! wget -O utils/google_utils.py https://raw.githubusercontent.com/hardik0/Multi-Object-Tracking-Google-Colab/main/utils/google_utils.py\n",
        "from utils.google_utils import *\n",
        "#gdrive_download(id='{google_drive_File_ID}', name='path_to/DETRAC.zip')\n",
        "gdrive_download(id='', name='data/DETRAC.zip')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Y9e47jmi9kHv"
      },
      "source": [
        "unzip DETRAC.zip file to `data/MOT`"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yjmUnHfRqBG_"
      },
      "source": [
        "%%shell\n",
        "\n",
        "mkdir data/MOT\n",
        "cd data/MOT\n",
        "unzip -q ../DETRAC.zip\n",
        "rm -r ../DETRAC.zip"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GhlTQ98jwsoW"
      },
      "source": [
        "# uncomment below lines to train model with different resolutions\n",
        "#! rm -r  cfg/yolov3_detrac.cfg    # default yolov3_1088x608\n",
        "#! cp cfg/yolov3_1088x608.cfg cfg/yolov3_detrac.cfg"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "a2atBAF0-eDg"
      },
      "source": [
        "`cfg/detrac.json` file is already created if you change/modify something then create again (optional)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "80qfNJjYBjob",
        "outputId": "83ee3feb-a3ca-4984-f2b8-b1f5a4a8f8fd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "%rm -r cfg/detrac.json\n",
        "%%writefile cfg/detrac.json\n",
        "{\n",
        "    \"root\":\"/content/Towards-Realtime-MOT/data/MOT\",\n",
        "    \"train\":\n",
        "    {\n",
        "        \"detrac\":\"./data/detrac.train\"\n",
        "    },\n",
        "    \"test_emb\":\n",
        "    {\n",
        "        \"detrac\":\"./data/detrac.train\"\n",
        "    },\n",
        "    \"test\":\n",
        "    {\n",
        "        \"detrac\":\"./data/detrac.train\"\n",
        "    }\n",
        "}"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Writing cfg/detrac.json\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LYDSQpTi_XYP"
      },
      "source": [
        "make directory on gdrive to save training model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sQ6kUTLYwsoE"
      },
      "source": [
        "#%mkdir '../gdrive/My Drive/Colab Notebooks/MOT'\n",
        "%mkdir '../gdrive/My Drive/Colab Notebooks/MOT/Towards_Realtime_MOT'\n",
        "%ls '../gdrive/My Drive/Colab Notebooks/MOT/Towards_Realtime_MOT/'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qRhZMNrRO5bv"
      },
      "source": [
        "Download Darknet-53 ImageNet pretrained model: [DarkNet Official](https://pjreddie.com/media/files/darknet53.conv.74) | or download from gdrive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l05qOKbwxgbl"
      },
      "source": [
        "%mkdir weights\n",
        "! cp -r '../gdrive/My Drive/Colab Notebooks/MOT/darknet53.conv.74' weights/"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9RGNnfiMPhfy"
      },
      "source": [
        "## Start the Training"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EKMpC8Z_wspB"
      },
      "source": [
        "# training \n",
        "\n",
        "#! CUDA_VISIBLE_DEVICES=0 python train.py \n",
        "! CUDA_VISIBLE_DEVICES=0 python -m torch.distributed.launch train.py \\\n",
        "                                --cfg cfg/yolov3_detrac.cfg --data-cfg cfg/detrac.json \\\n",
        "                                --batch-size 6 --print-interval 1000 --save-model-after 2 \\\n",
        "                                --weights-to '../gdrive/My Drive/Colab Notebooks/MOT/Towards_Realtime_MOT/' \n",
        "                               # --resume --weights-from '../gdrive/My Drive/Colab Notebooks/MOT/Towards_Realtime_MOT/run12_10_09_15/'  "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "P3XkoFq3PzII"
      },
      "source": [
        "%ls '../gdrive/My Drive/Colab Notebooks/MOT/Towards_Realtime_MOT/'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YJRPjUkGPliZ"
      },
      "source": [
        " ### Resume training"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xDpa-7Dev447",
        "outputId": "71c02dd8-3f22-4880-b56a-5c76b8634513",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 641
        }
      },
      "source": [
        "# resume training \n",
        "\n",
        "#! CUDA_VISIBLE_DEVICES=0 python train.py \n",
        "! CUDA_VISIBLE_DEVICES=0 python -m torch.distributed.launch train.py \\\n",
        "                                --cfg cfg/yolov3.cfg --data-cfg cfg/detrac.json \\\n",
        "                                --batch-size 6 --print-interval 1000 --save-model-after 2 \\\n",
        "                                --weights-to '../gdrive/My Drive/Colab Notebooks/MOT/Towards_Realtime_MOT/' \\\n",
        "                                --resume --weights-from '../gdrive/My Drive/Colab Notebooks/MOT/Towards_Realtime_MOT/run12_10_09_15/' "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "================================================================================\n",
            "dataset summary\n",
            "OrderedDict([('detrac', 5984.0)])\n",
            "total # identities: 5985\n",
            "start index\n",
            "OrderedDict([('detrac', 0)])\n",
            "================================================================================\n",
            "2020-10-14 06:22:35 [INFO]:    Epoch       Batch       box      conf        id     total  nTargets      time\n",
            "2020-10-14 06:22:44 [INFO]:     2/29     0/13679     0.634    0.0464      11.7     -4.17        66      9.45\n",
            "2020-10-14 06:44:06 [INFO]:     2/29  1000/13679     0.927     0.027      17.9     -5.13      42.5     0.656\n",
            "2020-10-14 07:05:37 [INFO]:     2/29  2000/13679      0.65    0.0262      17.9     -4.52      42.4     0.654\n",
            "2020-10-14 07:27:10 [INFO]:     2/29  3000/13679     0.649     0.026      17.9     -4.41      42.4     0.656\n",
            "2020-10-14 07:48:47 [INFO]:     2/29  4000/13679     0.563    0.0254      17.9     -1.62      42.4     0.664\n",
            "2020-10-14 08:10:26 [INFO]:     2/29  5000/13679     0.489    0.0249      17.8        -2      42.3     0.664\n",
            "2020-10-14 08:32:07 [INFO]:     2/29  6000/13679     0.648    0.0247      17.7     -2.03      42.4     0.669\n",
            "2020-10-14 08:53:47 [INFO]:     2/29  7000/13679      0.72    0.0243      17.6     -1.99      42.4      0.68\n",
            "2020-10-14 09:15:29 [INFO]:     2/29  8000/13679     0.636    0.0239      17.6     -2.68      42.4      0.67\n",
            "2020-10-14 09:37:11 [INFO]:     2/29  9000/13679     0.573    0.0241      17.6     -3.13      42.3     0.689\n",
            "2020-10-14 09:58:50 [INFO]:     2/29 10000/13679     0.519    0.0239      17.5     -3.58      42.4     0.667\n",
            "2020-10-14 10:20:32 [INFO]:     2/29 11000/13679     0.506    0.0238      17.5     -3.82      42.4     0.678\n",
            "2020-10-14 10:42:15 [INFO]:     2/29 12000/13679     0.526    0.0238      17.4    -0.533      42.4     0.676\n",
            "2020-10-14 11:03:59 [INFO]:     2/29 13000/13679      0.54    0.0234      17.4     0.282      42.4     0.664\n",
            "2020-10-14 11:18:48 [INFO]:    Epoch       Batch       box      conf        id     total  nTargets      time\n",
            "2020-10-14 11:18:54 [INFO]:     3/29     0/13679     0.173    0.0246      10.1     -6.26        70      9.75\n",
            "2020-10-14 11:40:38 [INFO]:     3/29  1000/13679     0.227    0.0214      16.7      13.9      42.4     0.669\n",
            "2020-10-14 12:02:21 [INFO]:     3/29  2000/13679     0.169    0.0211      16.5       5.9      42.5     0.679\n",
            "2020-10-14 12:24:06 [INFO]:     3/29  3000/13679     0.385    0.0234      16.8      3.46      42.5     0.677\n",
            "2020-10-14 12:45:50 [INFO]:     3/29  4000/13679     0.347    0.0243      16.9      1.72      42.6     0.688\n",
            "2020-10-14 13:07:34 [INFO]:     3/29  5000/13679     0.373    0.0241      16.9      2.28      42.6     0.666\n",
            "2020-10-14 13:29:16 [INFO]:     3/29  6000/13679     0.325    0.0246      16.9      6.58      42.6     0.676\n",
            "2020-10-14 13:50:58 [INFO]:     3/29  7000/13679     0.293    0.0248      16.9      6.04      42.4     0.683\n",
            "2020-10-14 14:12:41 [INFO]:     3/29  8000/13679     0.262    0.0245      16.8      4.39      42.5      0.68\n",
            "2020-10-14 14:34:22 [INFO]:     3/29  9000/13679     0.265    0.0242      16.7      4.05      42.6      0.67\n",
            "2020-10-14 14:56:02 [INFO]:     3/29 10000/13679     0.254    0.0237      16.7      3.09      42.5     0.657\n",
            "2020-10-14 15:17:41 [INFO]:     3/29 11000/13679     0.233    0.0232      16.6      2.03      42.5     0.685\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OzToKqJvX7Ae",
        "outputId": "1ac8a9b9-f923-4eb4-ce8f-f5a7ed5ea7e7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "%ls '../gdrive/My Drive/Colab Notebooks/MOT/Towards_Realtime_MOT/run14_10_06_21/'"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\u001b[0m\u001b[01;34mcfg\u001b[0m/  latest.pt  weights_epoch_2.pt\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p4Kpw4s1RzTH"
      },
      "source": [
        "copy weight file from gdrive to `model/` directory"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0CysoMJpYKh2"
      },
      "source": [
        "%mkdir models\n",
        "%cp -r '../gdrive/My Drive/Colab Notebooks/MOT/Towards_Realtime_MOT/run14_10_06_21/latest.pt' models/"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-wlB3J9NRTEJ"
      },
      "source": [
        "## Inference"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fZmaRk51SZkJ"
      },
      "source": [
        "### Download video from Youtube | Copy video from Google Drive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r4qe_E8lXpqR",
        "outputId": "c4c6ef72-e775-41e5-9889-122ca95cbca6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 426
        }
      },
      "source": [
        "!pip install youtube-dl\n",
        "from IPython.display import YouTubeVideo\n",
        "\n",
        "YOUTUBE_ID = 'jjlBnrzSGjc'\n",
        "\n",
        "YouTubeVideo(YOUTUBE_ID)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting youtube-dl\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/70/b4/c83b577bfb1be7c88dff10eecd70650eca7729b25135345108ea676e6f3b/youtube_dl-2020.9.20-py2.py3-none-any.whl (1.8MB)\n",
            "\u001b[K     |████████████████████████████████| 1.8MB 6.5MB/s \n",
            "\u001b[?25hInstalling collected packages: youtube-dl\n",
            "Successfully installed youtube-dl-2020.9.20\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "\n",
              "        <iframe\n",
              "            width=\"400\"\n",
              "            height=\"300\"\n",
              "            src=\"https://www.youtube.com/embed/jjlBnrzSGjc\"\n",
              "            frameborder=\"0\"\n",
              "            allowfullscreen\n",
              "        ></iframe>\n",
              "        "
            ],
            "text/plain": [
              "<IPython.lib.display.YouTubeVideo at 0x7fe68e848518>"
            ],
            "image/jpeg": "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\n"
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gY0cUVGZY5M9"
      },
      "source": [
        "#!rm -rf youtube.mp4\n",
        "# download the youtube with the given ID\n",
        "! youtube-dl -f 22 --output \"youtube.%(ext)s\" https://www.youtube.com/watch?v=$YOUTUBE_ID\n",
        "#!youtube-dl -f 'bestvideo[ext=mp4]' --output \"youtube.%(ext)s\" https://www.youtube.com/watch?v=$YOUTUBE_ID\n",
        "\n",
        "# cut the first 30 seconds\n",
        "! ffmpeg -y -loglevel info -i youtube.mp4 -t 30 car.mp4"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bI3pHhxvS6fd"
      },
      "source": [
        "### Run the demo"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VnSj-VHbKw1z"
      },
      "source": [
        "# line 163 - nID=5985 (if you get any nID error modify 163 line from file browser)\n",
        "! python demo.py --input-video car.mp4 --weights models/latest.pt --cfg cfg/yolov3_detrac.cfg --output-format video --output-root outputs/detrac_car_tracking"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g_gbNjQPiXHu",
        "outputId": "862da2d1-d225-469d-fc8a-90206359fe69",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "%ls outputs/detrac_car_tracking/"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\u001b[0m\u001b[01;34mframe\u001b[0m/  result.avi  result.mp4  results.txt\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Hz7-JlCtTYLv"
      },
      "source": [
        "rename result.mp4 file (optional)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WnQfCxrZmCJc"
      },
      "source": [
        "%mv outputs/detrac_car_tracking/result.mp4 outputs/detrac_car_tracking/detrac_car.mp4 "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lJz8wq7GTcko"
      },
      "source": [
        "Reduce video size (optional)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YtlrDasLTdZp"
      },
      "source": [
        "! ffmpeg -y -loglevel panic -i outputs/detrac_car_tracking/detrac_car.mp4 outputs/detrac_car_tracking/detrac_car_tracking_jde_1088X608_result.mp4"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eCXCSJ3tT0MB"
      },
      "source": [
        "Copy video to google drive | download video from file browser (optional)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2k6g_dpVl6sS"
      },
      "source": [
        "%cp -r outputs/detrac_car_tracking/detrac_car_tracking_jde_1088X608_result.mp4 '../gdrive/My Drive/Colab Notebooks/MOT/'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eiu6mlDrT8eR"
      },
      "source": [
        "Playing videos on google colab"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x7pAf9RRkIEU"
      },
      "source": [
        "# define helper function to display videos\n",
        "import io \n",
        "from IPython.display import HTML\n",
        "from base64 import b64encode\n",
        "def show_video(file_name, width=640):\n",
        "  # show resulting deepsort video\n",
        "  mp4 = open(file_name,'rb').read()\n",
        "  data_url = \"data:video/mp4;base64,\" + b64encode(mp4).decode()\n",
        "  return HTML(\"\"\"\n",
        "  <video width=\"{0}\" controls>\n",
        "        <source src=\"{1}\" type=\"video/mp4\">\n",
        "  </video>\n",
        "  \"\"\".format(width, data_url))"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}