Yolov3/Yolov3_baseline.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "SIMS_Yolov3_baseline.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p3HE4j_2wQru",
        "colab_type": "text"
      },
      "source": [
        "**Copy images and annotation from drive**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OkX9da7Bv6qI",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "43f64dba-b195-46f2-b748-599f759078c9"
      },
      "source": [
        "%cp '/content/drive/My Drive/SIMS_Dataset/images.zip' /content/\n",
        "%cp '/content/drive/My Drive/SIMS_Dataset/training.txt' /content/\n",
        "%cp '/content/drive/My Drive/SIMS_Dataset/test.txt' /content/\n",
        "%cp '/content/drive/My Drive/SIMS_Dataset/validation.txt' /content/\n",
        "!unzip -qq images.zip -d /content/\n",
        "\n",
        "print(\"Done Copying and Unzipping\")"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Done Copying and Unzipping\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "773gpD8uyDui",
        "colab_type": "text"
      },
      "source": [
        "**Preprocessing** Did once and copied processed categories into drive for later usage"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n-QXVSNIy-E3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#making directories\n",
        "\n",
        "!mkdir yolo_train/\n",
        "!mkdir yolo_train/images\n",
        "!mkdir yolo_train/labels\n",
        "\n",
        "!mkdir yolo_test/\n",
        "!mkdir yolo_test/images\n",
        "!mkdir yolo_test/labels\n",
        "\n",
        "!mkdir yolo_val/\n",
        "!mkdir yolo_val/images\n",
        "!mkdir yolo_val/labels"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oFs5UhxKwkLq",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 69
        },
        "outputId": "a0082be4-9768-47e2-ed26-aed1b58b8c9d"
      },
      "source": [
        "#separate images based on train,val and test and put corresponding files into \n",
        "#respective folder having images and labels as subfolders in it\n",
        "import shutil\n",
        "import os\n",
        "\n",
        "def copy(filepath,dest_dir):\n",
        "    with open(filepath) as fp:\n",
        "        for line in fp:\n",
        "            os.makedirs(dest_dir, exist_ok=True)  # succeeds even if directory exists.\n",
        "            shutil.copy(line.replace('\\n', ''), dest_dir+\"images\") #image\n",
        "            shutil.copy(line.replace('.jpg\\n', '.txt'), dest_dir+\"labels\") ##label\n",
        "    print(\"Done: \"+ filepath)\n",
        "            \n",
        "copy('./training.txt','yolo_train/')\n",
        "copy('./validation.txt','yolo_val/')\n",
        "copy('./test.txt','yolo_test/')"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Done: ./training.txt\n",
            "Done: ./validation.txt\n",
            "Done: ./test.txt\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BAjHUNBkEyau",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#make Dataset folder and place above created folders here\n",
        "!mkdir Dataset\n",
        "!mv yolo_train/ Dataset/\n",
        "!mv yolo_test/ Dataset/\n",
        "!mv yolo_val/ Dataset/"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Oomm6rmFPgl9",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 69
        },
        "outputId": "a46def48-97cf-4cdd-a5f7-4f43a66a463a"
      },
      "source": [
        "#update paths of train,test,val\n",
        "\n",
        "### replace ./images/0000.jpg to \"/content/yolov3/Dataset/yolo_train/images/0000.jpg\" \n",
        "\n",
        "def copy(filepath,newfile,binder):\n",
        "    with open(filepath) as fp:\n",
        "      print(filepath)\n",
        "      for line in fp:\n",
        "        in_file = line[2:]\n",
        "        new = binder+in_file\n",
        "        with open(newfile, \"a\") as f:\n",
        "          f.write(new)\n",
        "\n",
        "copy('training.txt','Dataset/yolo_train/yolo_train.txt',\n",
        "     \"/content/yolov3/Dataset/yolo_train/\")\n",
        "\n",
        "copy('test.txt','Dataset/yolo_test/yolo_test.txt',\n",
        "     \"/content/yolov3/Dataset/yolo_test/\")\n",
        "\n",
        "copy('validation.txt','Dataset/yolo_val/yolo_val.txt',\n",
        "     \"/content/yolov3/Dataset/yolo_val/\")"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "training.txt\n",
            "test.txt\n",
            "validation.txt\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ARl_oz0CIaHS",
        "colab_type": "text"
      },
      "source": [
        "Copy data and save into drive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XFmmeGG41F9V",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Copied dataset and relevant files to drive\n",
        "!zip -r SIMS_Dataset_yolo.zip /content/yolov3/Dataset/\n",
        "!cp /content/yolov3/SIMS_Dataset_yolo.zip '/content/drive/My Drive/SIMS_Dataset'"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "N-w1i-koGQ1s",
        "colab_type": "text"
      },
      "source": [
        "Preprocessing ended here"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cZZSt5sf26io",
        "colab_type": "text"
      },
      "source": [
        "**Clone repository and install requirements**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_p7tHeo622UK",
        "colab_type": "code",
        "outputId": "f4ed0b61-8b81-4d29-99fe-fe06cca3b19c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "!git clone https://github.com/ultralytics/yolov3\n",
        "%cd yolov3/\n",
        "!pip install -U -r requirements.txt"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'yolov3'...\n",
            "remote: Enumerating objects: 9120, done.\u001b[K\n",
            "remote: Total 9120 (delta 0), reused 0 (delta 0), pack-reused 9120\u001b[K\n",
            "Receiving objects: 100% (9120/9120), 7.06 MiB | 9.49 MiB/s, done.\n",
            "Resolving deltas: 100% (6285/6285), done.\n",
            "/content/yolov3\n",
            "Requirement already up-to-date: numpy in /usr/local/lib/python3.6/dist-packages (from -r requirements.txt (line 2)) (1.18.4)\n",
            "Collecting opencv-python>=4.1\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/72/c2/e9cf54ae5b1102020ef895866a67cb2e1aef72f16dd1fde5b5fb1495ad9c/opencv_python-4.2.0.34-cp36-cp36m-manylinux1_x86_64.whl (28.2MB)\n",
            "\u001b[K     |████████████████████████████████| 28.2MB 115kB/s \n",
            "\u001b[?25hRequirement already up-to-date: torch>=1.5 in /usr/local/lib/python3.6/dist-packages (from -r requirements.txt (line 4)) (1.5.0+cu101)\n",
            "Requirement already up-to-date: matplotlib in /usr/local/lib/python3.6/dist-packages (from -r requirements.txt (line 5)) (3.2.1)\n",
            "Requirement already up-to-date: pycocotools in /usr/local/lib/python3.6/dist-packages (from -r requirements.txt (line 6)) (2.0.0)\n",
            "Collecting tqdm\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/c9/40/058b12e8ba10e35f89c9b1fdfc2d4c7f8c05947df2d5eb3c7b258019fda0/tqdm-4.46.0-py2.py3-none-any.whl (63kB)\n",
            "\u001b[K     |████████████████████████████████| 71kB 11.8MB/s \n",
            "\u001b[?25hCollecting pillow\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/e0/50/8e78e6f62ffa50d6ca95c281d5a2819bef66d023ac1b723e253de5bda9c5/Pillow-7.1.2-cp36-cp36m-manylinux1_x86_64.whl (2.1MB)\n",
            "\u001b[K     |████████████████████████████████| 2.1MB 52.7MB/s \n",
            "\u001b[?25hRequirement already up-to-date: tensorboard>=1.14 in /usr/local/lib/python3.6/dist-packages (from -r requirements.txt (line 9)) (2.2.1)\n",
            "Requirement already satisfied, skipping upgrade: future in /usr/local/lib/python3.6/dist-packages (from torch>=1.5->-r requirements.txt (line 4)) (0.16.0)\n",
            "Requirement already satisfied, skipping upgrade: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->-r requirements.txt (line 5)) (0.10.0)\n",
            "Requirement already satisfied, skipping upgrade: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->-r requirements.txt (line 5)) (2.4.7)\n",
            "Requirement already satisfied, skipping upgrade: python-dateutil>=2.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->-r requirements.txt (line 5)) (2.8.1)\n",
            "Requirement already satisfied, skipping upgrade: kiwisolver>=1.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->-r requirements.txt (line 5)) (1.2.0)\n",
            "Requirement already satisfied, skipping upgrade: setuptools>=41.0.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (46.1.3)\n",
            "Requirement already satisfied, skipping upgrade: markdown>=2.6.8 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (3.2.1)\n",
            "Requirement already satisfied, skipping upgrade: google-auth-oauthlib<0.5,>=0.4.1 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (0.4.1)\n",
            "Requirement already satisfied, skipping upgrade: absl-py>=0.4 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (0.9.0)\n",
            "Requirement already satisfied, skipping upgrade: grpcio>=1.24.3 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (1.28.1)\n",
            "Requirement already satisfied, skipping upgrade: werkzeug>=0.11.15 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (1.0.1)\n",
            "Requirement already satisfied, skipping upgrade: wheel>=0.26; python_version >= \"3\" in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (0.34.2)\n",
            "Requirement already satisfied, skipping upgrade: requests<3,>=2.21.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (2.23.0)\n",
            "Requirement already satisfied, skipping upgrade: google-auth<2,>=1.6.3 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (1.7.2)\n",
            "Requirement already satisfied, skipping upgrade: six>=1.10.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (1.12.0)\n",
            "Requirement already satisfied, skipping upgrade: tensorboard-plugin-wit>=1.6.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (1.6.0.post3)\n",
            "Requirement already satisfied, skipping upgrade: protobuf>=3.6.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard>=1.14->-r requirements.txt (line 9)) (3.10.0)\n",
            "Requirement already satisfied, skipping upgrade: requests-oauthlib>=0.7.0 in /usr/local/lib/python3.6/dist-packages (from google-auth-oauthlib<0.5,>=0.4.1->tensorboard>=1.14->-r requirements.txt (line 9)) (1.3.0)\n",
            "Requirement already satisfied, skipping upgrade: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests<3,>=2.21.0->tensorboard>=1.14->-r requirements.txt (line 9)) (2020.4.5.1)\n",
            "Requirement already satisfied, skipping upgrade: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests<3,>=2.21.0->tensorboard>=1.14->-r requirements.txt (line 9)) (1.24.3)\n",
            "Requirement already satisfied, skipping upgrade: chardet<4,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests<3,>=2.21.0->tensorboard>=1.14->-r requirements.txt (line 9)) (3.0.4)\n",
            "Requirement already satisfied, skipping upgrade: idna<3,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests<3,>=2.21.0->tensorboard>=1.14->-r requirements.txt (line 9)) (2.9)\n",
            "Requirement already satisfied, skipping upgrade: pyasn1-modules>=0.2.1 in /usr/local/lib/python3.6/dist-packages (from google-auth<2,>=1.6.3->tensorboard>=1.14->-r requirements.txt (line 9)) (0.2.8)\n",
            "Requirement already satisfied, skipping upgrade: rsa<4.1,>=3.1.4 in /usr/local/lib/python3.6/dist-packages (from google-auth<2,>=1.6.3->tensorboard>=1.14->-r requirements.txt (line 9)) (4.0)\n",
            "Requirement already satisfied, skipping upgrade: cachetools<3.2,>=2.0.0 in /usr/local/lib/python3.6/dist-packages (from google-auth<2,>=1.6.3->tensorboard>=1.14->-r requirements.txt (line 9)) (3.1.1)\n",
            "Requirement already satisfied, skipping upgrade: oauthlib>=3.0.0 in /usr/local/lib/python3.6/dist-packages (from requests-oauthlib>=0.7.0->google-auth-oauthlib<0.5,>=0.4.1->tensorboard>=1.14->-r requirements.txt (line 9)) (3.1.0)\n",
            "Requirement already satisfied, skipping upgrade: pyasn1<0.5.0,>=0.4.6 in /usr/local/lib/python3.6/dist-packages (from pyasn1-modules>=0.2.1->google-auth<2,>=1.6.3->tensorboard>=1.14->-r requirements.txt (line 9)) (0.4.8)\n",
            "\u001b[31mERROR: albumentations 0.1.12 has requirement imgaug<0.2.7,>=0.2.5, but you'll have imgaug 0.2.9 which is incompatible.\u001b[0m\n",
            "Installing collected packages: opencv-python, tqdm, pillow\n",
            "  Found existing installation: opencv-python 4.1.2.30\n",
            "    Uninstalling opencv-python-4.1.2.30:\n",
            "      Successfully uninstalled opencv-python-4.1.2.30\n",
            "  Found existing installation: tqdm 4.41.1\n",
            "    Uninstalling tqdm-4.41.1:\n",
            "      Successfully uninstalled tqdm-4.41.1\n",
            "  Found existing installation: Pillow 7.0.0\n",
            "    Uninstalling Pillow-7.0.0:\n",
            "      Successfully uninstalled Pillow-7.0.0\n",
            "Successfully installed opencv-python-4.2.0.34 pillow-7.1.2 tqdm-4.46.0\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.colab-display-data+json": {
              "pip_warning": {
                "packages": [
                  "PIL",
                  "tqdm"
                ]
              }
            }
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EqOwJUq3-eV_",
        "colab_type": "text"
      },
      "source": [
        "**created sims_classes.names and sims.data files**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zaN1KBtzJnDU",
        "colab_type": "code",
        "outputId": "f3eb7d0d-77b4-4ed0-c690-0b76003a0353",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "%cd yolov3"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/yolov3\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pppOEQTOIzT1",
        "colab_type": "text"
      },
      "source": [
        "**Load Preprocessed Data from Drive**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9BxPfot1Zkxc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#using copied/preprocessed data now\n",
        "##!mkdir /content/yolov3/Dataset\n",
        "%cp '/content/drive/My Drive/SIMS_Dataset/SIMS_yolo.zip' /content/yolov3/\n",
        "!unzip SIMS_yolo.zip -d /content/yolov3/"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7OUnd4rDBQw0",
        "colab_type": "code",
        "outputId": "5fc7ad1f-0ade-4877-c1b1-5305e92e6681",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        }
      },
      "source": [
        "#download pretrained weights\n",
        "!. weights/download_yolov3_weights.sh"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current\n",
            "                                 Dload  Upload   Total   Spent    Left  Speed\n",
            "100   408    0   408    0     0    681      0 --:--:-- --:--:-- --:--:--   681\n",
            "  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0\n",
            "  0     0    0     0    0     0      0      0 --:--:--  0:00:01 --:--:--     0\n",
            "100  236M    0  236M    0     0  53.4M      0 --:--:--  0:00:04 --:--:-- 88.4M\n",
            "Downloading https://drive.google.com/uc?export=download&id=1SHNFyoe5Ni8DajDNEqgB2oVKBb_NoEad as weights/yolov3.pt... Done (5.1s)\n",
            "  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current\n",
            "                                 Dload  Upload   Total   Spent    Left  Speed\n",
            "100   408    0   408    0     0   1291      0 --:--:-- --:--:-- --:--:--  1291\n",
            "  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0\n",
            "  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0\n",
            "100  240M    0  240M    0     0  35.9M      0 --:--:--  0:00:06 --:--:-- 50.8M\n",
            "Downloading https://drive.google.com/uc?export=download&id=1f6Ovy3BSq2wYq4UfvFUpxJFNDFfrIDcR as weights/yolov3-spp.pt... Done (7.1s)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "omNsjqJ4iE7o",
        "colab_type": "code",
        "outputId": "0e6d07f8-40fb-4598-f5ae-e4dfb01d55ef",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "#start training\n",
        "!python3 train.py --batch 8 --epochs 60 --img-size 512 --data Dataset/sims.txt --cache-images --rec --cfg yolov3.cfg --name from_yolov3 --weights weights/yolov3.pt"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Apex recommended for faster mixed precision training: https://github.com/NVIDIA/apex\n",
            "Namespace(adam=False, batch_size=8, bucket='', cache_images=True, cfg='yolov3.cfg', data='Dataset/sims.txt', device='', epochs=60, evolve=False, img_size=[512], multi_scale=False, name='from_yolov3', nosave=False, notest=False, rect=True, resume=False, single_cls=False, weights='weights/yolov3.pt')\n",
            "Using CUDA device0 _CudaDeviceProperties(name='Tesla P4', total_memory=7611MB)\n",
            "\n",
            "Start Tensorboard with \"tensorboard --logdir=runs\", view at http://localhost:6006/\n",
            "2020-05-06 04:43:12.184919: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1\n",
            "Model Summary: 222 layers, 6.15991e+07 parameters, 6.15991e+07 gradients\n",
            "Optimizer groups: 75 .bias, 75 Conv2d.weight, 72 other\n",
            "/usr/local/lib/python3.6/dist-packages/torch/optim/lr_scheduler.py:123: UserWarning: Detected call of `lr_scheduler.step()` before `optimizer.step()`. In PyTorch 1.1.0 and later, you should call them in the opposite order: `optimizer.step()` before `lr_scheduler.step()`.  Failure to do this will result in PyTorch skipping the first value of the learning rate schedule. See more details at https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate\n",
            "  \"https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate\", UserWarning)\n",
            "Caching labels (3232 found, 0 missing, 0 empty, 0 duplicate, for 3233 images): 100% 3233/3233 [00:00<00:00, 9132.51it/s]\n",
            "Caching images (1.9GB): 100% 3233/3233 [00:36<00:00, 87.86it/s]\n",
            "Caching labels (1740 found, 0 missing, 0 empty, 0 duplicate, for 1741 images): 100% 1741/1741 [00:00<00:00, 7946.25it/s]\n",
            "Caching images (1.0GB): 100% 1741/1741 [00:19<00:00, 90.91it/s]\n",
            "Image sizes 512 - 512 train, 512 test\n",
            "Using 2 dataloader workers\n",
            "Starting training for 60 epochs...\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "      0/59     4.88G       5.2       4.3      4.57      14.1         0       512: 100% 405/405 [05:10<00:00,  1.31it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:20<00:00,  2.71it/s]\n",
            "                 all  1.74e+03  2.12e+04    0.0694     0.101    0.0587    0.0741\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "      1/59     4.89G      4.04       2.2      3.86      10.1         0       512: 100% 405/405 [05:11<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:19<00:00,  2.76it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.155     0.238     0.165     0.156\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "      2/59     4.89G      3.38      1.69      3.43       8.5         0       512: 100% 405/405 [05:11<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:20<00:00,  2.72it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.149     0.397     0.195     0.195\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "  0% 0/405 [00:00<?, ?it/s]\n",
            "Model Bias Summary:    layer        regression        objectness    classification\n",
            "                          82      -0.15+/-0.19      -5.14+/-0.89      -3.10+/-0.37 \n",
            "                          94      -0.08+/-0.23      -4.91+/-0.25      -3.11+/-0.34 \n",
            "                         106       0.16+/-0.27      -5.56+/-0.66      -3.15+/-0.49 \n",
            "      3/59     4.89G      3.08      1.44      2.91      7.43         0       512: 100% 405/405 [05:11<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:16<00:00,  2.86it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.199     0.441     0.266     0.252\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "      4/59     4.89G      2.78      1.28      2.52      6.59         0       512: 100% 405/405 [05:11<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:13<00:00,  2.96it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.223     0.475     0.325      0.28\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "      5/59     4.89G      2.65      1.17      2.22      6.04         0       512: 100% 405/405 [05:11<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:12<00:00,  3.02it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.228     0.549     0.357     0.312\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "      6/59     4.89G      2.52      1.08      1.96      5.56         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:12<00:00,  3.01it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.222     0.538     0.356     0.306\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "      7/59     4.89G      2.07      1.03      1.76      4.86         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:09<00:00,  3.13it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.244     0.567     0.404     0.327\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "      8/59     4.89G      1.85     0.979      1.64      4.47         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:09<00:00,  3.16it/s]\n",
            "                 all  1.74e+03  2.12e+04      0.25     0.576     0.411     0.335\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "      9/59     4.89G      1.75     0.948      1.57      4.27         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:08<00:00,  3.18it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.254     0.577     0.419     0.339\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     10/59     4.89G       1.7     0.931       1.5      4.13         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:07<00:00,  3.22it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.258     0.581     0.421     0.343\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     11/59     4.89G      1.66     0.906      1.43         4         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:07<00:00,  3.24it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.267      0.58     0.428     0.352\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     12/59     4.89G      1.62      0.89      1.36      3.87         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:07<00:00,  3.24it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.273     0.583     0.437     0.359\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     13/59     4.89G      1.61     0.871      1.27      3.75         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.26it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.276     0.586     0.445     0.362\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     14/59     4.89G      1.58     0.854       1.2      3.63         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:07<00:00,  3.23it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.284     0.591     0.452      0.37\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     15/59     4.89G       1.6     0.839      1.12      3.56         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:07<00:00,  3.23it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.284     0.588     0.453      0.37\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     16/59     4.89G      1.61     0.812      1.05      3.47         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:07<00:00,  3.25it/s]\n",
            "                 all  1.74e+03  2.12e+04       0.3     0.592     0.467     0.385\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     17/59     4.89G      1.62     0.796     0.988      3.41         0       512: 100% 405/405 [05:12<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.27it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.304     0.599      0.47     0.388\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     18/59     4.89G      1.65     0.788     0.933      3.38         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.30it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.312     0.577     0.464     0.395\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     19/59     4.89G      1.66     0.768     0.871       3.3         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.31it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.316     0.581     0.473     0.397\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     20/59     4.89G      1.71     0.753     0.826      3.29         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.30it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.308     0.587      0.47     0.391\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     21/59     4.89G      1.74     0.737     0.783      3.26         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.29it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.365     0.576     0.485      0.41\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     22/59     4.89G      1.75     0.721      0.74      3.21         0       512: 100% 405/405 [05:12<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.30it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.427       0.6     0.489     0.422\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     23/59     4.89G      1.76     0.709     0.692      3.16         0       512: 100% 405/405 [05:12<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.28it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.362     0.586     0.496     0.427\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     24/59     4.89G      1.77     0.691     0.658      3.12         0       512: 100% 405/405 [05:12<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.27it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.336     0.577     0.473     0.411\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     25/59     4.89G      1.84     0.692     0.654      3.19         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.27it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.403     0.609     0.484     0.429\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     26/59     4.89G      1.86     0.682     0.634      3.17         0       512: 100% 405/405 [05:12<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.29it/s]\n",
            "                 all  1.74e+03  2.12e+04      0.43     0.581     0.495     0.448\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     27/59     4.89G      1.86     0.681      0.64      3.18         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.30it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.426     0.614     0.501     0.447\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     28/59     4.89G       1.9     0.674     0.602      3.17         0       512: 100% 405/405 [05:12<00:00,  1.30it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.27it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.429      0.62     0.503     0.453\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     29/59     4.89G      1.84     0.655      0.57      3.06         0       512: 100% 405/405 [05:12<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.30it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.447      0.62     0.514     0.468\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     30/59     4.89G      1.89     0.645      0.56      3.09         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.32it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.499     0.613     0.508     0.463\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     31/59     4.89G       1.9     0.642     0.541      3.09         0       512: 100% 405/405 [05:12<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.31it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.413     0.623     0.487      0.45\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     32/59     4.89G      1.89     0.632      0.52      3.04         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:04<00:00,  3.36it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.522     0.616     0.529     0.495\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     33/59     4.89G      1.91     0.633     0.501      3.04         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.32it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.449     0.619     0.531     0.473\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     34/59     4.89G      1.95     0.625     0.471      3.05         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.32it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.516     0.628     0.535     0.476\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     35/59     4.89G      1.94     0.617     0.461      3.02         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.30it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.429     0.619      0.54     0.479\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     36/59     4.89G      1.93     0.613     0.442      2.99         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.28it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.413     0.681     0.524     0.484\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     37/59     4.89G      1.93      0.61     0.443      2.98         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.31it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.417     0.635     0.531     0.485\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     38/59     4.89G      2.01     0.617     0.448      3.08         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:06<00:00,  3.28it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.435     0.697      0.53     0.505\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     39/59     4.89G      1.97     0.615     0.454      3.03         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.33it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.395      0.68     0.516     0.479\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     40/59     4.89G      1.87     0.599      0.41      2.88         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.34it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.439     0.661     0.547     0.501\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     41/59     4.89G      1.95     0.611     0.416      2.98         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.35it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.401     0.663     0.523     0.477\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     42/59     4.89G      1.93     0.602     0.397      2.92         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.35it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.412     0.678     0.516      0.48\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     43/59     4.89G      1.94     0.598     0.403      2.94         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:04<00:00,  3.37it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.453     0.673     0.563     0.522\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     44/59     4.89G       1.9     0.589     0.376      2.87         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.35it/s]\n",
            "                 all  1.74e+03  2.12e+04      0.47     0.635     0.545     0.505\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     45/59     4.89G      1.95     0.589     0.388      2.93         0       512: 100% 405/405 [05:13<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:04<00:00,  3.38it/s]\n",
            "                 all  1.74e+03  2.12e+04      0.53     0.683     0.571     0.534\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     46/59     4.89G      1.93     0.582     0.379      2.89         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:04<00:00,  3.37it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.516     0.638     0.554     0.512\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     47/59     4.89G      1.89     0.578      0.36      2.83         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.34it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.514     0.638     0.569      0.51\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     48/59     4.89G      1.86     0.567     0.353      2.78         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:04<00:00,  3.36it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.449     0.657     0.537     0.511\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     49/59     4.89G      1.87      0.57     0.365       2.8         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.33it/s]\n",
            "                 all  1.74e+03  2.12e+04      0.45     0.649     0.546     0.503\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     50/59     4.89G       1.9     0.577      0.39      2.87         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.34it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.514     0.638     0.551     0.481\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     51/59     4.89G      1.85     0.578      0.39      2.82         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.35it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.459     0.708     0.592     0.544\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     52/59     4.89G      1.82     0.575     0.367      2.77         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.33it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.484     0.682     0.581     0.536\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     53/59     4.89G      1.86     0.578     0.372      2.81         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.33it/s]\n",
            "                 all  1.74e+03  2.12e+04       0.5     0.679     0.601     0.539\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     54/59     4.89G      1.82     0.566     0.341      2.72         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.32it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.496     0.651     0.574     0.498\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     55/59     4.89G      1.86     0.567     0.347      2.77         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.35it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.413     0.673     0.529     0.478\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     56/59     4.89G      1.75     0.545     0.306       2.6         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.35it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.544     0.691     0.608     0.565\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     57/59     4.89G      1.83     0.547     0.305      2.68         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:04<00:00,  3.36it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.503     0.671     0.586     0.547\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     58/59     4.89G      1.83      0.55     0.301      2.68         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.34it/s]\n",
            "                 all  1.74e+03  2.12e+04      0.51     0.717     0.596     0.555\n",
            "\n",
            "     Epoch   gpu_mem      GIoU       obj       cls     total   targets  img_size\n",
            "     59/59     4.89G      1.76     0.533     0.281      2.58         0       512: 100% 405/405 [05:14<00:00,  1.29it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 218/218 [01:05<00:00,  3.31it/s]\n",
            "                 all  1.74e+03  2.12e+04     0.522     0.645     0.581      0.52\n",
            "60 epochs completed in 6.394 hours.\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3NgOyNCXXNgD",
        "colab_type": "code",
        "outputId": "697c7ea8-8882-42e2-ca67-5223dcae1e24",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 414
        }
      },
      "source": [
        "from utils import utils;\n",
        "utils.plot_results()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x432 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8cDEJRYL3vPM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Load the TensorBoard notebook extension\n",
        "%load_ext tensorboard\n",
        "%tensorboard --logdir=runs"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8TyL8C4aOpUH",
        "colab_type": "code",
        "outputId": "6dc9b7c7-eeb6-4f0b-9f3a-20f49eadb618",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 227
        }
      },
      "source": [
        "!python3 detect.py --names Dataset/sims_classes.names --cfg yolov3.cfg --weights weights/best_from_yolov3.pt"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Namespace(agnostic_nms=False, augment=False, cfg='yolov3.cfg', classes=None, conf_thres=0.3, device='', fourcc='mp4v', half=False, img_size=512, iou_thres=0.6, names='Dataset/sims_classes.names', output='output', save_txt=False, source='data/samples', view_img=False, weights='weights/best_from_yolov3.pt')\n",
            "Using CUDA device0 _CudaDeviceProperties(name='Tesla P4', total_memory=7611MB)\n",
            "\n",
            "Model Summary: 222 layers, 6.15991e+07 parameters, 6.15991e+07 gradients\n",
            "image 1/5 data/samples/0011.jpg: 384x512 2 airliners, Done. (0.031s)\n",
            "image 2/5 data/samples/0037.jpg: 384x512 4 trucks, 1 longvehicles, Done. (0.032s)\n",
            "image 3/5 data/samples/0061.jpg: 384x512 8 cars, Done. (0.031s)\n",
            "image 4/5 data/samples/0136.jpg: 384x512 14 chartereds, Done. (0.031s)\n",
            "image 5/5 data/samples/0177.jpg: 384x512 1 cars, 2 boats, Done. (0.031s)\n",
            "Results saved to /content/yolov3/output\n",
            "Done. (0.759s)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XiH3Lc5mjVg4",
        "colab_type": "code",
        "outputId": "546c356e-5b22-4114-ebc1-fa3c7508798c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 625
        }
      },
      "source": [
        "#test on last model\n",
        "!python3 test.py --data Dataset/sims_test.txt --cfg yolov3.cfg --batch-size 8 --weights weights/last_from_yolov3.pt --save-json --img-size 512"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Namespace(augment=False, batch_size=8, cfg='yolov3.cfg', conf_thres=0.001, data='Dataset/sims_test.txt', device='', img_size=512, iou_thres=0.6, save_json=True, single_cls=False, task='test', weights='weights/last_from_yolov3.pt')\n",
            "Using CUDA device0 _CudaDeviceProperties(name='Tesla P4', total_memory=7611MB)\n",
            "\n",
            "Model Summary: 222 layers, 6.15991e+07 parameters, 6.15991e+07 gradients\n",
            "Fusing layers...\n",
            "Model Summary: 150 layers, 6.15728e+07 parameters, 6.15728e+07 gradients\n",
            "Reading image shapes: 100% 748/748 [00:00<00:00, 2000.05it/s]\n",
            "Caching labels (747 found, 0 missing, 0 empty, 0 duplicate, for 748 images): 100% 748/748 [00:00<00:00, 2220.81it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 94/94 [00:25<00:00,  3.68it/s]\n",
            "                 all       748  7.98e+03     0.526     0.659     0.605     0.534\n",
            "                 car       748  3.68e+03     0.679     0.878       0.8     0.765\n",
            "               truck       748       446     0.503     0.644     0.559     0.565\n",
            "                 van       748       874     0.586     0.755     0.685      0.66\n",
            "         longvehicle       748       222     0.394     0.847      0.53     0.537\n",
            "                 bus       748       366      0.59     0.904     0.825     0.714\n",
            "            airliner       748       161      0.62     0.981     0.924      0.76\n",
            "           propeller       748        25     0.518     0.774     0.701     0.621\n",
            "             trainer       748        49     0.532     0.918     0.736     0.674\n",
            "           chartered       748       103     0.506     0.932     0.773     0.656\n",
            "             fighter       748         8         1     0.125      0.69     0.222\n",
            "               other       748        85     0.055       0.4    0.0665    0.0967\n",
            "          stairtruck       748        83     0.318     0.434      0.29     0.367\n",
            "       pushbacktruck       748        55     0.258     0.152     0.136     0.191\n",
            "          helicopter       748         9     0.448     0.188     0.419     0.265\n",
            "                boat       748  1.82e+03     0.884     0.945     0.943     0.914\n",
            "Speed: 19.6/2.9/22.5 ms inference/NMS/total per 512x512 image at batch-size 8\n",
            "\n",
            "COCO mAP with pycocotools...\n",
            "Traceback (most recent call last):\n",
            "  File \"test.py\", line 257, in <module>\n",
            "    opt.augment)\n",
            "  File \"test.py\", line 211, in test\n",
            "    cocoGt = COCO(glob.glob('../coco/annotations/instances_val*.json')[0])  # initialize COCO ground truth api\n",
            "IndexError: list index out of range\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i8OW66a7vuqM",
        "colab_type": "code",
        "outputId": "d1dbd5a8-ca09-4abf-a00f-398455eadbee",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 607
        }
      },
      "source": [
        "#test on best model\n",
        "!python3 test.py --data Dataset/sims_test.txt --cfg yolov3.cfg --batch-size 8 --weights weights/best_from_yolov3.pt --save-json --img-size 512"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Namespace(augment=False, batch_size=8, cfg='yolov3.cfg', conf_thres=0.001, data='Dataset/sims_test.txt', device='', img_size=512, iou_thres=0.6, save_json=True, single_cls=False, task='test', weights='weights/best_from_yolov3.pt')\n",
            "Using CUDA device0 _CudaDeviceProperties(name='Tesla P4', total_memory=7611MB)\n",
            "\n",
            "Model Summary: 222 layers, 6.15991e+07 parameters, 6.15991e+07 gradients\n",
            "Fusing layers...\n",
            "Model Summary: 150 layers, 6.15728e+07 parameters, 6.15728e+07 gradients\n",
            "Caching labels (747 found, 0 missing, 0 empty, 0 duplicate, for 748 images): 100% 748/748 [00:00<00:00, 7115.16it/s]\n",
            "               Class    Images   Targets         P         R   mAP@0.5        F1: 100% 94/94 [00:25<00:00,  3.76it/s]\n",
            "                 all       748  7.98e+03     0.549     0.697     0.634     0.581\n",
            "                 car       748  3.68e+03     0.768     0.836     0.834     0.801\n",
            "               truck       748       446     0.497      0.78     0.674     0.607\n",
            "                 van       748       874     0.499     0.747     0.626     0.598\n",
            "         longvehicle       748       222     0.395     0.856       0.6     0.541\n",
            "                 bus       748       366     0.733     0.847     0.823     0.786\n",
            "            airliner       748       161     0.696     0.981     0.962     0.815\n",
            "           propeller       748        25     0.254      0.92     0.689     0.398\n",
            "             trainer       748        49     0.533     0.863     0.782     0.659\n",
            "           chartered       748       103     0.677     0.854     0.838     0.755\n",
            "             fighter       748         8      0.53     0.625     0.657     0.574\n",
            "               other       748        85    0.0956     0.259    0.0944      0.14\n",
            "          stairtruck       748        83     0.395      0.47     0.328     0.429\n",
            "       pushbacktruck       748        55     0.255     0.145      0.13     0.185\n",
            "          helicopter       748         9         1     0.352      0.54      0.52\n",
            "                boat       748  1.82e+03     0.898     0.918     0.936     0.908\n",
            "Speed: 19.6/2.7/22.3 ms inference/NMS/total per 512x512 image at batch-size 8\n",
            "\n",
            "COCO mAP with pycocotools...\n",
            "Traceback (most recent call last):\n",
            "  File \"test.py\", line 257, in <module>\n",
            "    opt.augment)\n",
            "  File \"test.py\", line 211, in test\n",
            "    cocoGt = COCO(glob.glob('../coco/annotations/instances_val*.json')[0])  # initialize COCO ground truth api\n",
            "IndexError: list index out of range\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HLFRoNvnSglq",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Zip yolov3 contents into drive\n",
        "!zip -r /content/yolov3_baseline_v1_epochs60.zip /content/yolov3/\n",
        "!cp /content/yolov3_baseline_v1_epochs60.zip '/content/drive/My Drive/SIMS_Dataset'"
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}