Colab

最新の20件

私的AI研究会 > Colab

Google Colaboratory で機械学習 †

　機械学習や Deep Learning の実行環境に「Google Colaboratory」を使ってみる
　YOLO V7 を使ってカスタムデータによる学習を実行する
　環境設定／カスタムデータによる追加学習／ローカルマシンでの推論実行まで、1時間以内に終了する作業を目標とする

Google Colaboratory 動作確認 †

　無償で手軽に利用可能な クラウド上の Python 環境「Google Colaboratory」を確認する

Google Drive に作業領域を確保する †

1. Googleドライブを起動する
   Googleアカウントにログインしていなければログインする
   
   
2. マイドライブ直下に新しく「try」フォルダを作成する
   マイドライブ表示エリアでマウス右クリック、「新しいフォルダ」を選択する
   
   
   
   
   

Google Colaboratory を使ってみる †

1. Google の「Colaboratory へようこそ」 ページを開く
   
   
   
   
2. Googleアカウントにログインしていなければログインする
   
   
3. 「ファイル」メニューまたはダイアログのボタンから「ノートブックを新規作成」を選択
   
   
   
   
   
4. 左上のタイトル（Untitled*）を「yolov7_custom」に変更する
   
   
5. セルにコードを入力し、左側の三角ボタンをクリックするか「Ctrl」+「Enter」を押して実行する
   
   

   print("hello colaboratory!")

   入力欄（セル）の下に結果が表示され簡単に Python の実行環境が手に入る
   
   ※「ノートブック」でのコマンド実行手順
   1. セルが表示されていない場合は「＋コード」メニューを押しセルを表示する
      
   2. セルにコマンドを入力し、三角ボタンをクリックするか「Ctrl」+「Enter」を押す
      
   3. 先頭に「!」をつけることで、PythonではなくOSのコマンドラインに命令を送ることができる
      

Google Colaboratory の実行環境を設定(確認)する †

1. メニューから「ランタイム」→「ランタイムのタイプを変更」を選ぶ
   
   
   
   
2. 表示されたダイアログから「GPU」を選ぶ
   GPU を設定する
   
   
3. スペックを確認する
   
   
   1. 「lshw」をインストールする
      

      !apt install lshw

   2. ハードウェア情報を収集するために「lshw」を実行する
      

      !lshw

      ・結果表示

         :
          *-memory
               description: System memory
               physical id: 0
               size: 12GiB
          *-cpu
               product: Intel(R) Xeon(R) CPU @ 2.20GHz
               vendor: Intel Corp.
               physical id: 1
               bus info: cpu@0
               width: 64 bits
         :
             *-display
                  description: 3D controller
                  product: TU104GL [Tesla T4]
                  vendor: NVIDIA Corporation
                  physical id: 4
                  bus info: pci@0000:00:04.0
                  version: a1
                  width: 64 bits
                  clock: 33MHz
         :

      ※「CPU Intel(R) Xeon(R) 2.20GHz」「メモリ 12GB」「GPU Tesla T4」の実行環境が確認できる
      

Google Colaboratory に Googleドライブをマウントする †

1. Colaboratory ページの左「ファイル」①を押す
   
   
2. 左上の「ドライブをマウント」②を押す
   
   
3. ダイアログが表示された場合「Googleドライブに接続」を押す
   
   
   
   
4. マウントが完了すると「drive」フォルダ③が表示される
   前項で作成した「try」フォルダが表示されることを確認する
   
   参考 → GoogleColabでGoogleドライブをマウント
   

Google Colaboratory に「YOLO V7」を実装 †

　クラウド上の Python 仮想環境「Google Colaboratory」に「YOLO V7」を実行できる環境を構築する

物体検出 AI「YOLO V7」をインストール †

1. カレントディレクトリを「MyDrive/try7」へ移動する
   

   cd /content/drive/MyDrive/try

   ・結果表示

   /content/drive/MyDrive/try

   ※ カレントディレクトリを確認するコマンド
   

   !pwd

   
   
2. YOLO v7 オフィシャルサイト から下記コマンドでプロジェクトをクローン
   

   !git clone https://github.com/WongKinYiu/yolov7

   ▼　- log -　GoogleColab Tesla T4
   ▲　- log -　GoogleColab Tesla T4

   Cloning into 'yolov7'...
   remote: Enumerating objects: 1154, done.
   remote: Counting objects: 100% (15/15), done.
   remote: Compressing objects: 100% (8/8), done.
   remote: Total 1154 (delta 8), reused 13 (delta 7), pack-reused 1139
   Receiving objects: 100% (1154/1154), 70.42 MiB | 17.19 MiB/s, done.
   Resolving deltas: 100% (494/494), done.
   Updating files: 100% (104/104), done.

   
   ・成功するとGoogleドライブ「/MyDrive/try」に「yolov7」というディレクトリが作成される
   
   
3. カレントディレクトリを「yolov7」へ移動する
   

   cd yolov7

   
   
   
4. 学習済みモデルをダウンロード
   ・YOLOv7 には学習済みモデルがいくつも用意されているが今回は「yolov7-tiny.pt」を使用する
   

   !wget https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-tiny.pt

   ▼　- log -　GoogleColab Tesla T4
   ▲　- log -　GoogleColab Tesla T4

   --2023-09-11 00:32:33--  https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-tiny.pt
   Resolving github.com (github.com)... 140.82.113.3
   Connecting to github.com (github.com)|140.82.113.3|:443... connected.
   HTTP request sent, awaiting response... 302 Found
   Location: https://objects.githubusercontent.com/github-production-release-asset-2e65be/511187726/ba7d01ee-125a-4134-8864-fa1abcbf94d5?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIWNJYAX4CSVEH53A%2F20230911%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20230911T003233Z&X-Amz-Expires=300&X-Amz-Signature=e69beb9fd21b30dd684c024a58df89cf2d7402d3d85998f3cee7397cd2680e1b&X-Amz-SignedHeaders=host&actor_id=0&key_id=0&repo_id=511187726&response-content-disposition=attachment%3B%20filename%3Dyolov7-tiny.pt&response-content-type=application%2Foctet-stream [following]
   --2023-09-11 00:32:33--  https://objects.githubusercontent.com/github-production-release-asset-2e65be/511187726/ba7d01ee-125a-4134-8864-fa1abcbf94d5?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIWNJYAX4CSVEH53A%2F20230911%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20230911T003233Z&X-Amz-Expires=300&X-Amz-Signature=e69beb9fd21b30dd684c024a58df89cf2d7402d3d85998f3cee7397cd2680e1b&X-Amz-SignedHeaders=host&actor_id=0&key_id=0&repo_id=511187726&response-content-disposition=attachment%3B%20filename%3Dyolov7-tiny.pt&response-content-type=application%2Foctet-stream
   Resolving objects.githubusercontent.com (objects.githubusercontent.com)... 185.199.109.133, 185.199.108.133, 185.199.110.133, ...
   Connecting to objects.githubusercontent.com (objects.githubusercontent.com)|185.199.109.133|:443... connected.
   HTTP request sent, awaiting response... 200 OK
   Length: 12639769 (12M) [application/octet-stream]
   Saving to: ‘yolov7-tiny.pt’
   
   yolov7-tiny.pt      100%[===================>]  12.05M  34.6MB/s    in 0.3s    
   
   2023-09-11 00:32:33 (34.6 MB/s) - ‘yolov7-tiny.pt’ saved [12639769/12639769]

サンプル画像の物体検出 †

1. サンプル画像の物体検出を実行する
   

   !python detect.py --source inference/images/ --weights yolov7-tiny.pt --conf 0.25 --img-size 1280 --device 0

   ▼　- log -　GoogleColab Tesla T4
   ▲　- log -　GoogleColab Tesla T4

   Namespace(weights=['yolov7-tiny.pt'], source='inference/images/', img_size=1280, conf_thres=0.25, iou_thres=0.45, device='0', view_img=False, save_txt=False, save_conf=False, nosave=False, classes=None, agnostic_nms=False, augment=False, update=False, project='runs/detect', name='exp', exist_ok=False, no_trace=False)
   YOLOR 🚀 v0.1-126-g84932d7 torch 2.0.1+cu118 CUDA:0 (Tesla T4, 15101.8125MB)
   
   Fusing layers... 
   Model Summary: 200 layers, 6219709 parameters, 229245 gradients
    Convert model to Traced-model... 
    traced_script_module saved! 
    model is traced! 
   
   /usr/local/lib/python3.10/dist-packages/torch/functional.py:504: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at ../aten/src/ATen/native/TensorShape.cpp:3483.)
     return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
   4 persons, 1 bus, Done. (13.4ms) Inference, (31.8ms) NMS
    The image with the result is saved in: runs/detect/exp/bus.jpg
   5 horses, Done. (12.3ms) Inference, (1.8ms) NMS
    The image with the result is saved in: runs/detect/exp/horses.jpg
   3 persons, Done. (15.3ms) Inference, (1.9ms) NMS
    The image with the result is saved in: runs/detect/exp/image1.jpg
   2 persons, Done. (12.9ms) Inference, (1.2ms) NMS
    The image with the result is saved in: runs/detect/exp/image2.jpg
   1 dog, 1 horse, Done. (12.6ms) Inference, (1.3ms) NMS
    The image with the result is saved in: runs/detect/exp/image3.jpg
   2 persons, 2 ties, Done. (10.9ms) Inference, (1.5ms) NMS
    The image with the result is saved in: runs/detect/exp/zidane.jpg
   Done. (1.881s)

   
   
   
   
   ・サンプル画像は「yolov7/inference/images」を「–source」オプションで指定
   ・推論結果の保存先は「yolov7/runs/detect/exp*」（ * は自動的に振られる番号）
   ・結果の画像は「Google Colab」のファイル、または「Google Drive」で参照（ダウンロード）できる
   
   ・出力画像（6枚）
   
   
   
   
   
   
   

ランタイム接続を切断する †

1. メニューバー「ランタイム」→「ランタイムを接続解除して削除」を選択する
   
   
2. ダイアログ選択で「はい」を押す
   
   ※ GPUを選択している場合、実行しない場合は接続解除するほうが望ましい（時間制限がかかるので）
   

カスタムデータで学習する †

　カスタムデータによる学習４「じゃんけんの判定（その３）」 で作成したデータセット「janken4_dataset」を使用して追加学習をする
　元になる学習モデルは「yolov7-tiny.pt」

事前準備（データセットのアップロード） †

1. Googleドライブを起動する
   
   
   
2. データ・セット「yolov7-main/data/janken4_dataset」をフォルダごとGoogleドライブの「MyDrive/try/yolov7/data」にアップする
   

   yolov7　
     ┗ data
        ┗ janken4_dataset

   
   
3. 学習するのに必要な情報を記述したyamlファイル「yolov7-main/janken4_dataset.yaml」を「MyDrive/try/yolov7」の直下にアップする
   

   yolov7　
     ┠ data
     ┃  ┗ janken4_dataset
     ┗ janken4_dataset.yaml

   
   
4. アップロードには数分かかるので、右下の「アップロード完了」の表示を確認してから次に進む
   

Google Colaboratory で学習を実行する †

1. Colaboratory を起動し、ノートブック「yolov7_custom.ipynb」を選択する
   
   
2. 左サイドバーの「ファイル」→「ドライブをマウント」を選択することで Googleドライブをマウントする
   
   
3. カレントディレクトリを「MyDrive/try/yolov7」へ移動する
   

   cd /content/drive/MyDrive/try/yolov7

   /content/drive/MyDrive/try/yolov7

4. 下記のコマンドで「janken4_dataset」を学習する (yolov7-tiny)
   学習時間を少なくして効果を上げるため、学習モデル = yolov7-tiny・バッチサイズ = 16・エポック数 = 50 で学習する
   

   !python train.py --workers 8 --batch-size 16 --data janken4_dataset.yaml --cfg cfg/training/yolov7-tiny.yaml --weights 'yolov7-tiny.pt' --name yolov7-tiny_jk4 --hyp data/hyp.scratch.tiny.yaml --epochs 50 --device 0

   ▼　- log -　GoogleColab Tesla T4
   ▲　- log -　GoogleColab Tesla T4

   2023-09-11 04:53:56.828622: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
   To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
   2023-09-11 04:53:57.729827: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
   YOLOR 🚀 v0.1-126-g84932d7 torch 2.0.1+cu118 CUDA:0 (Tesla T4, 15101.8125MB)
   
   Namespace(weights='yolov7-tiny.pt', cfg='cfg/training/yolov7-tiny.yaml', data='janken4_dataset.yaml', hyp='data/hyp.scratch.tiny.yaml', epochs=50, batch_size=16, img_size=[640, 640], rect=False, resume=False, nosave=False, notest=False, noautoanchor=False, evolve=False, bucket='', cache_images=False, image_weights=False, device='0', multi_scale=False, single_cls=False, adam=False, sync_bn=False, local_rank=-1, workers=8, project='runs/train', entity=None, name='yolov7-tiny_jk4', exist_ok=False, quad=False, linear_lr=False, label_smoothing=0.0, upload_dataset=False, bbox_interval=-1, save_period=-1, artifact_alias='latest', freeze=[0], v5_metric=False, world_size=1, global_rank=-1, save_dir='runs/train/yolov7-tiny_jk4', total_batch_size=16)
   tensorboard: Start with 'tensorboard --logdir runs/train', view at http://localhost:6006/
   hyperparameters: lr0=0.01, lrf=0.01, momentum=0.937, weight_decay=0.0005, warmup_epochs=3.0, warmup_momentum=0.8, warmup_bias_lr=0.1, box=0.05, cls=0.5, cls_pw=1.0, obj=1.0, obj_pw=1.0, iou_t=0.2, anchor_t=4.0, fl_gamma=0.0, hsv_h=0.015, hsv_s=0.7, hsv_v=0.4, degrees=0.0, translate=0.1, scale=0.5, shear=0.0, perspective=0.0, flipud=0.0, fliplr=0.5, mosaic=1.0, mixup=0.05, copy_paste=0.0, paste_in=0.05, loss_ota=1
   wandb: Install Weights & Biases for YOLOR logging with 'pip install wandb' (recommended)
   Overriding model.yaml nc=80 with nc=3
   
                    from  n    params  module                                  arguments                     
     0                -1  1       928  models.common.Conv                      [3, 32, 3, 2, None, 1, LeakyReLU(negative_slope=0.1)]
     1                -1  1     18560  models.common.Conv                      [32, 64, 3, 2, None, 1, LeakyReLU(negative_slope=0.1)]
     2                -1  1      2112  models.common.Conv                      [64, 32, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
     3                -2  1      2112  models.common.Conv                      [64, 32, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
     4                -1  1      9280  models.common.Conv                      [32, 32, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
     5                -1  1      9280  models.common.Conv                      [32, 32, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
     6  [-1, -2, -3, -4]  1         0  models.common.Concat                    [1]                           
     7                -1  1      8320  models.common.Conv                      [128, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
     8                -1  1         0  models.common.MP                        []                            
     9                -1  1      4224  models.common.Conv                      [64, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    10                -2  1      4224  models.common.Conv                      [64, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    11                -1  1     36992  models.common.Conv                      [64, 64, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    12                -1  1     36992  models.common.Conv                      [64, 64, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    13  [-1, -2, -3, -4]  1         0  models.common.Concat                    [1]                           
    14                -1  1     33024  models.common.Conv                      [256, 128, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    15                -1  1         0  models.common.MP                        []                            
    16                -1  1     16640  models.common.Conv                      [128, 128, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    17                -2  1     16640  models.common.Conv                      [128, 128, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    18                -1  1    147712  models.common.Conv                      [128, 128, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    19                -1  1    147712  models.common.Conv                      [128, 128, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    20  [-1, -2, -3, -4]  1         0  models.common.Concat                    [1]                           
    21                -1  1    131584  models.common.Conv                      [512, 256, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    22                -1  1         0  models.common.MP                        []                            
    23                -1  1     66048  models.common.Conv                      [256, 256, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    24                -2  1     66048  models.common.Conv                      [256, 256, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    25                -1  1    590336  models.common.Conv                      [256, 256, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    26                -1  1    590336  models.common.Conv                      [256, 256, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    27  [-1, -2, -3, -4]  1         0  models.common.Concat                    [1]                           
    28                -1  1    525312  models.common.Conv                      [1024, 512, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    29                -1  1    131584  models.common.Conv                      [512, 256, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    30                -2  1    131584  models.common.Conv                      [512, 256, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    31                -1  1         0  models.common.SP                        [5]                           
    32                -2  1         0  models.common.SP                        [9]                           
    33                -3  1         0  models.common.SP                        [13]                          
    34  [-1, -2, -3, -4]  1         0  models.common.Concat                    [1]                           
    35                -1  1    262656  models.common.Conv                      [1024, 256, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    36          [-1, -7]  1         0  models.common.Concat                    [1]                           
    37                -1  1    131584  models.common.Conv                      [512, 256, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    38                -1  1     33024  models.common.Conv                      [256, 128, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    39                -1  1         0  torch.nn.modules.upsampling.Upsample    [None, 2, 'nearest']          
    40                21  1     33024  models.common.Conv                      [256, 128, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    41          [-1, -2]  1         0  models.common.Concat                    [1]                           
    42                -1  1     16512  models.common.Conv                      [256, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    43                -2  1     16512  models.common.Conv                      [256, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    44                -1  1     36992  models.common.Conv                      [64, 64, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    45                -1  1     36992  models.common.Conv                      [64, 64, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    46  [-1, -2, -3, -4]  1         0  models.common.Concat                    [1]                           
    47                -1  1     33024  models.common.Conv                      [256, 128, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    48                -1  1      8320  models.common.Conv                      [128, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    49                -1  1         0  torch.nn.modules.upsampling.Upsample    [None, 2, 'nearest']          
    50                14  1      8320  models.common.Conv                      [128, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    51          [-1, -2]  1         0  models.common.Concat                    [1]                           
    52                -1  1      4160  models.common.Conv                      [128, 32, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    53                -2  1      4160  models.common.Conv                      [128, 32, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    54                -1  1      9280  models.common.Conv                      [32, 32, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    55                -1  1      9280  models.common.Conv                      [32, 32, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    56  [-1, -2, -3, -4]  1         0  models.common.Concat                    [1]                           
    57                -1  1      8320  models.common.Conv                      [128, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    58                -1  1     73984  models.common.Conv                      [64, 128, 3, 2, None, 1, LeakyReLU(negative_slope=0.1)]
    59          [-1, 47]  1         0  models.common.Concat                    [1]                           
    60                -1  1     16512  models.common.Conv                      [256, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    61                -2  1     16512  models.common.Conv                      [256, 64, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    62                -1  1     36992  models.common.Conv                      [64, 64, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    63                -1  1     36992  models.common.Conv                      [64, 64, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    64  [-1, -2, -3, -4]  1         0  models.common.Concat                    [1]                           
    65                -1  1     33024  models.common.Conv                      [256, 128, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    66                -1  1    295424  models.common.Conv                      [128, 256, 3, 2, None, 1, LeakyReLU(negative_slope=0.1)]
    67          [-1, 37]  1         0  models.common.Concat                    [1]                           
    68                -1  1     65792  models.common.Conv                      [512, 128, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    69                -2  1     65792  models.common.Conv                      [512, 128, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    70                -1  1    147712  models.common.Conv                      [128, 128, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    71                -1  1    147712  models.common.Conv                      [128, 128, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    72  [-1, -2, -3, -4]  1         0  models.common.Concat                    [1]                           
    73                -1  1    131584  models.common.Conv                      [512, 256, 1, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    74                57  1     73984  models.common.Conv                      [64, 128, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    75                65  1    295424  models.common.Conv                      [128, 256, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    76                73  1   1180672  models.common.Conv                      [256, 512, 3, 1, None, 1, LeakyReLU(negative_slope=0.1)]
    77      [74, 75, 76]  1     22544  models.yolo.IDetect                     [3, [[10, 13, 16, 30, 33, 23], [30, 61, 62, 45, 59, 119], [116, 90, 156, 198, 373, 326]], [128, 256, 512]]
   Model Summary: 263 layers, 6020400 parameters, 6020400 gradients
   
   Transferred 330/344 items from yolov7-tiny.pt
   Scaled weight_decay = 0.0005
   Optimizer groups: 58 .bias, 58 conv.weight, 61 other
   train: Scanning 'data/janken4_dataset/train/labels.cache' images and labels... 480 found, 0 missing, 0 empty, 0 corrupted: 100% 480/480 [00:00<?, ?it/s]
   val: Scanning 'data/janken4_dataset/valid/labels.cache' images and labels... 120 found, 0 missing, 0 empty, 0 corrupted: 100% 120/120 [00:00<?, ?it/s]
   
   autoanchor: Analyzing anchors... anchors/target = 3.13, Best Possible Recall (BPR) = 1.0000
   Image sizes 640 train, 640 test
   Using 2 dataloader workers
   Logging results to runs/train/yolov7-tiny_jk4
   Starting training for 50 epochs...
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         0/49     2.82G   0.05163   0.03024   0.03289    0.1148        40       640: 100% 30/30 [00:48<00:00,  1.61s/it]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95:   0% 0/4 [00:00<?, ?it/s]/usr/local/lib/python3.10/dist-packages/torch/functional.py:504: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at ../aten/src/ATen/native/TensorShape.cpp:3483.)
     return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:04<00:00,  1.17s/it]
                    all         120         120      0.0294      0.0833      0.0121      0.0019
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         1/49     2.76G   0.04142   0.01522    0.0269   0.08355        39       640: 100% 30/30 [00:24<00:00,  1.21it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.12it/s]
                    all         120         120      0.0508       0.267      0.0419      0.0107
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         2/49     3.03G   0.04004   0.01346   0.02455   0.07805        43       640: 100% 30/30 [00:25<00:00,  1.20it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.44it/s]
                    all         120         120       0.248       0.475       0.269      0.0753
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         3/49     3.03G   0.03464   0.01214   0.02277   0.06955        40       640: 100% 30/30 [00:25<00:00,  1.20it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.86it/s]
                    all         120         120       0.281         0.6       0.344       0.136
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         4/49     3.03G   0.03583   0.01056   0.02009   0.06647        32       640: 100% 30/30 [00:24<00:00,  1.21it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.00it/s]
                    all         120         120       0.289       0.658       0.409       0.173
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         5/49     3.03G   0.03825   0.01048   0.02177    0.0705        44       640: 100% 30/30 [00:25<00:00,  1.16it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.28it/s]
                    all         120         120        0.22       0.308       0.177      0.0539
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         6/49     3.04G    0.0388   0.01073   0.02577    0.0753        38       640: 100% 30/30 [00:25<00:00,  1.17it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.96it/s]
                    all         120         120       0.303       0.625       0.391       0.144
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         7/49     3.04G   0.04666  0.009941   0.02295   0.07955        45       640: 100% 30/30 [00:25<00:00,  1.20it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.07it/s]
                    all         120         120       0.415        0.35       0.366       0.139
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         8/49     3.04G   0.03045  0.009827   0.01737   0.05764        43       640: 100% 30/30 [00:23<00:00,  1.26it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.44it/s]
                    all         120         120       0.317         0.4       0.291       0.125
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
         9/49     3.04G   0.03207  0.009744   0.01779   0.05961        40       640: 100% 30/30 [00:22<00:00,  1.31it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.40it/s]
                    all         120         120       0.393       0.514       0.464       0.241
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        10/49     3.04G   0.04525  0.009894   0.02294   0.07808        43       640: 100% 30/30 [00:23<00:00,  1.26it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.45it/s]
                    all         120         120       0.389       0.442       0.371       0.153
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        11/49     3.04G   0.03351   0.01057    0.0189   0.06298        35       640: 100% 30/30 [00:23<00:00,  1.27it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.62it/s]
                    all         120         120       0.325         0.3       0.334       0.126
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        12/49     3.04G   0.03801  0.009947   0.01937   0.06733        33       640: 100% 30/30 [00:23<00:00,  1.29it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.17it/s]
                    all         120         120       0.587       0.585       0.616       0.271
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        13/49     3.04G    0.0348   0.01007   0.01838   0.06325        41       640: 100% 30/30 [00:24<00:00,  1.22it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.79it/s]
                    all         120         120      0.0701       0.183      0.0393     0.00638
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        14/49     3.04G   0.03803   0.01172    0.0212   0.07095        39       640: 100% 30/30 [00:24<00:00,  1.21it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.12it/s]
                    all         120         120       0.423       0.408       0.314       0.111
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        15/49     3.04G   0.03323    0.0099   0.01628   0.05941        31       640: 100% 30/30 [00:23<00:00,  1.26it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.07it/s]
                    all         120         120       0.553       0.808       0.746        0.39
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        16/49     3.04G    0.0414  0.009267   0.01829   0.06896        34       640: 100% 30/30 [00:24<00:00,  1.25it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.13it/s]
                    all         120         120       0.438         0.5       0.452       0.223
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        17/49     3.04G    0.0422  0.009625   0.01798   0.06981        40       640: 100% 30/30 [00:25<00:00,  1.19it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.19it/s]
                    all         120         120       0.627         0.8       0.796       0.376
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        18/49     3.04G   0.03946   0.01036   0.01854   0.06837        43       640: 100% 30/30 [00:25<00:00,  1.17it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.12it/s]
                    all         120         120       0.416       0.515       0.435       0.168
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        19/49     3.04G   0.03422  0.009949   0.01627   0.06044        45       640: 100% 30/30 [00:24<00:00,  1.20it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.99it/s]
                    all         120         120       0.673       0.697       0.762       0.424
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        20/49     3.04G   0.03836    0.0101   0.01839   0.06684        32       640: 100% 30/30 [00:24<00:00,  1.20it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.90it/s]
                    all         120         120       0.511       0.725       0.647       0.299
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        21/49     3.04G    0.0348  0.009871   0.01717   0.06184        45       640: 100% 30/30 [00:25<00:00,  1.17it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.29it/s]
                    all         120         120       0.501        0.57        0.53       0.164
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        22/49     3.04G   0.03335  0.009804   0.01746   0.06061        43       640: 100% 30/30 [00:25<00:00,  1.17it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.91it/s]
                    all         120         120       0.552       0.692        0.69       0.368
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        23/49     3.04G   0.04454  0.009487   0.01796   0.07199        40       640: 100% 30/30 [00:25<00:00,  1.18it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.32it/s]
                    all         120         120       0.696       0.907       0.879       0.421
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        24/49     3.04G   0.03541  0.009314   0.01514   0.05986        45       640: 100% 30/30 [00:25<00:00,  1.19it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.30it/s]
                    all         120         120       0.749       0.892       0.877       0.522
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        25/49     3.04G    0.0346  0.008935   0.01424   0.05777        38       640: 100% 30/30 [00:24<00:00,  1.23it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.21it/s]
                    all         120         120       0.737       0.865       0.873       0.473
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        26/49     3.04G   0.03659  0.009053   0.01299   0.05863        41       640: 100% 30/30 [00:25<00:00,  1.18it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.25it/s]
                    all         120         120       0.879        0.85       0.938       0.536
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        27/49     3.04G   0.03182  0.009148   0.01238   0.05335        40       640: 100% 30/30 [00:25<00:00,  1.18it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.43it/s]
                    all         120         120       0.839       0.924       0.946       0.522
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        28/49     3.04G   0.03359  0.008742    0.0132   0.05553        39       640: 100% 30/30 [00:26<00:00,  1.14it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.75it/s]
                    all         120         120         0.9        0.93       0.968       0.607
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        29/49     3.04G   0.02708  0.008494   0.01253   0.04811        44       640: 100% 30/30 [00:24<00:00,  1.24it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.36it/s]
                    all         120         120       0.942       0.939       0.977       0.586
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        30/49     3.04G   0.02979  0.008439   0.01063   0.04886        49       640: 100% 30/30 [00:25<00:00,  1.16it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.30it/s]
                    all         120         120       0.774       0.882       0.879       0.428
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        31/49     3.04G   0.03019  0.008725   0.01339    0.0523        34       640: 100% 30/30 [00:26<00:00,  1.15it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.16it/s]
                    all         120         120       0.793       0.887       0.914        0.54
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        32/49     3.04G   0.03222  0.009168   0.01255   0.05395        33       640: 100% 30/30 [00:25<00:00,  1.17it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.26it/s]
                    all         120         120       0.872       0.908       0.934       0.594
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        33/49     3.04G    0.0282  0.008521    0.0134   0.05012        38       640: 100% 30/30 [00:24<00:00,  1.24it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.64it/s]
                    all         120         120       0.916       0.946       0.974       0.619
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        34/49     3.04G   0.02161  0.008843   0.01127   0.04173        40       640: 100% 30/30 [00:24<00:00,  1.25it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.58it/s]
                    all         120         120       0.926       0.967       0.979       0.653
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        35/49     3.04G   0.02493   0.00863   0.01233   0.04589        40       640: 100% 30/30 [00:25<00:00,  1.18it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.76it/s]
                    all         120         120       0.958       0.952        0.98       0.614
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        36/49     3.04G   0.02508  0.008566   0.01639   0.05004        26       640: 100% 30/30 [00:24<00:00,  1.24it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.42it/s]
                    all         120         120       0.951       0.975       0.986       0.617
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        37/49     3.04G   0.02741  0.008481   0.01651    0.0524        31       640: 100% 30/30 [00:23<00:00,  1.27it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.43it/s]
                    all         120         120       0.981       0.956       0.985        0.67
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        38/49     3.04G   0.02756  0.008425   0.01379   0.04977        35       640: 100% 30/30 [00:23<00:00,  1.26it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.87it/s]
                    all         120         120       0.979       0.967       0.984        0.65
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        39/49     3.04G   0.02563  0.009084   0.01572   0.05044        40       640: 100% 30/30 [00:23<00:00,  1.26it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.65it/s]
                    all         120         120        0.95       0.975       0.984       0.639
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        40/49     3.04G   0.02511  0.008971   0.01684   0.05092        42       640: 100% 30/30 [00:24<00:00,  1.24it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.43it/s]
                    all         120         120       0.966       0.975       0.985       0.658
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        41/49     3.04G   0.02154  0.009098   0.01421   0.04485        37       640: 100% 30/30 [00:25<00:00,  1.18it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  1.66it/s]
                    all         120         120       0.988        0.95       0.985        0.68
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        42/49     3.04G   0.02811  0.008275   0.01537   0.05175        34       640: 100% 30/30 [00:23<00:00,  1.26it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.26it/s]
                    all         120         120       0.989       0.966       0.986       0.674
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        43/49     3.04G   0.02775  0.008882    0.0157   0.05233        36       640: 100% 30/30 [00:24<00:00,  1.24it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.61it/s]
                    all         120         120       0.981       0.958       0.988       0.656
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        44/49     3.04G   0.02123  0.008521   0.01461   0.04436        40       640: 100% 30/30 [00:25<00:00,  1.18it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.08it/s]
                    all         120         120       0.992       0.961       0.989        0.68
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        45/49     3.04G   0.02376  0.008377    0.0143   0.04644        37       640: 100% 30/30 [00:25<00:00,  1.20it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:02<00:00,  2.00it/s]
                    all         120         120       0.965       0.975       0.988       0.701
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        46/49     3.04G    0.0289  0.008536   0.01608   0.05352        38       640: 100% 30/30 [00:25<00:00,  1.18it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.09it/s]
                    all         120         120        0.96       0.959       0.987       0.699
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        47/49     3.04G   0.02046   0.00847   0.01375   0.04268        40       640: 100% 30/30 [00:25<00:00,  1.17it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.24it/s]
                    all         120         120       0.986       0.957       0.987       0.708
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        48/49     3.04G   0.02173  0.008446   0.01309   0.04327        41       640: 100% 30/30 [00:24<00:00,  1.22it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:01<00:00,  2.04it/s]
                    all         120         120       0.971       0.975       0.989        0.71
   
        Epoch   gpu_mem       box       obj       cls     total    labels  img_size
        49/49     3.04G   0.02241  0.008406    0.0145   0.04532        30       640: 100% 30/30 [00:25<00:00,  1.19it/s]
                  Class      Images      Labels           P           R      mAP@.5  mAP@.5:.95: 100% 4/4 [00:04<00:00,  1.01s/it]
                    all         120         120       0.977       0.975       0.989       0.709
                    goo         120          40       0.962           1       0.993       0.687
                  choki         120          40       0.969        0.95        0.98       0.659
                    par         120          40           1       0.975       0.992       0.782
   50 epochs completed in 0.403 hours.
   
   Optimizer stripped from runs/train/yolov7-tiny_jk4/weights/last.pt, 12.3MB
   Optimizer stripped from runs/train/yolov7-tiny_jk4/weights/best.pt, 12.3MB

   
   
   
5. およそ30分で正常に終了すると、学習結果は「yolov7/runs/train/yolov7-tiny_jk4/」に保存される
   
   
6. onnx パッケージをインストールする
   

   !pip install onnx

   ▼　- log -　GoogleColab Tesla T4
   ▲　- log -　GoogleColab Tesla T4

   Collecting onnx
     Downloading onnx-1.14.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (14.6 MB)
        ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 14.6/14.6 MB 37.4 MB/s eta 0:00:00
   Requirement already satisfied: numpy in /usr/local/lib/python3.10/dist-packages (from onnx) (1.23.5)
   Requirement already satisfied: protobuf>=3.20.2 in /usr/local/lib/python3.10/dist-packages (from onnx) (3.20.3)
   Requirement already satisfied: typing-extensions>=3.6.2.1 in /usr/local/lib/python3.10/dist-packages (from onnx) (4.5.0)
   Installing collected packages: onnx
   Successfully installed onnx-1.14.1

   
   
7. 学習結果モデル「runs/train/yolov7-tiny_jk4/weights/best.pt」を onnx 形式に変換する
   

   !python export.py --weights runs/train/yolov7-tiny_jk4/weights/best.pt

   ▼　- log -　GoogleColab Tesla T4
   ▲　- log -　GoogleColab Tesla T4

   Import onnx_graphsurgeon failure: No module named 'onnx_graphsurgeon'
   Namespace(weights='runs/train/yolov7-tiny_jk4/weights/best.pt', img_size=[640, 640], batch_size=1, dynamic=False, dynamic_batch=False, grid=False, end2end=False, max_wh=None, topk_all=100, iou_thres=0.45, conf_thres=0.25, device='cpu', simplify=False, include_nms=False, fp16=False, int8=False)
   YOLOR 🚀 v0.1-126-g84932d7 torch 2.0.1+cu118 CPU
   
   Fusing layers... 
   IDetect.fuse
   Model Summary: 208 layers, 6013008 parameters, 0 gradients
   
   Starting TorchScript export with torch 2.0.1+cu118...
   TorchScript export success, saved as runs/train/yolov7-tiny_jk4/weights/best.torchscript.pt
   CoreML export failure: No module named 'coremltools'
   
   Starting TorchScript-Lite export with torch 2.0.1+cu118...
   TorchScript-Lite export success, saved as runs/train/yolov7-tiny_jk4/weights/best.torchscript.ptl
   
   Starting ONNX export with onnx 1.14.1...
   /content/drive/MyDrive/try/yolov7/models/yolo.py:582: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
     if augment:
   /content/drive/MyDrive/try/yolov7/models/yolo.py:614: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
     if profile:
   /content/drive/MyDrive/try/yolov7/models/yolo.py:629: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
     if profile:
   ============= Diagnostic Run torch.onnx.export version 2.0.1+cu118 =============
   verbose: False, log level: Level.ERROR
   ======================= 0 NONE 0 NOTE 0 WARNING 0 ERROR ========================
   
   ONNX export success, saved as runs/train/yolov7-tiny_jk4/weights/best.onnx
   
   Export complete (8.19s). Visualize with https://github.com/lutzroeder/netron.

   
   
8. メニューバー「ランタイム」→「ランタイムを接続解除して削除」を選択し、ランタイム接続を切断する
   
   
9. 学習結果 モデルは「yolov7-tiny_jk4/weights」評価指標は「yolov7-tiny_jk4」フォルダ に保存される
   

   F1 curve	P curve	PR curve	R curve

   
   
   
   
   
   
   
   
   
   
10. GoogleColab を終了し、Googleドライブから学習結果をローカル・マシンにダウンロードする
    Google Drive 上の「yolov7-tiny_jk4」フォルダをダウンロード（圧縮ファイルを展開）
    「MyDrive/try/yolov7/runs/train/yolov7-tiny_jk4」→「yolov7-main/runs/train/yolov7-tiny_jk4」
    

学習済みモデルで推論を実行する †

　物体認識の深層学習タスク：YOLO V7 で作成・使用した「anaconda」上の仮想環境（py38a）を利用する
　プロジェクト・フォルダは「/anaconda_win/work/yolov7」
　使用するカスタムデータによる学習済みモデルは「/anaconda_win/work/yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx」

ローカルマシンで推論実行 †

• OpenVINO™ での実行
  ・実行ディレクトリへ切り替える
  

  (py38a) PS > cd /anaconda_win/work/yolov7

• 実行コマンドと結果
  

  入力ソース	コマンド	実行結果
  カメラ画像	python object_detect_yolo7.py -m ../yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx -l janken.names_jp -i cam -d GPU	-
  janken3.jpg	python object_detect_yolo7.py -m ../yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx -l janken.names_jp -i ../../Images/janken3.jpg -d GPU
  janken_test2.mp4	python object_detect_yolo7.py -m ../yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx -l janken.names_jp -i ../../Videos/janken_test2.mp4

  　※1 インテル製 GPU を使用しないときは「-d GPU」オプションを除く
  
  
  ▼　- log -　Intel® Core™ i7-6700 / Intel® HD Graphics 530
  ▲　- log -　Intel® Core™ i7-6700 / Intel® HD Graphics 530

  (py38a) PS > cd /anaconda_win/work/yolov7
  (py38a) PS > python object_detect_yolo7.py -m ../yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx -l janken.names_jp -i cam -d GPU
  Starting..
   - Program title  : Object detection YOLO V7
   - OpenCV version : 4.5.5
   - OpenVINO engine: 2022.1.0-7019-cdb9bec7210-releases/2022/1
   - Input image    : cam
   - Model          : ../yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx
   - Device         : GPU
   - Confidence thr : 0.25
   - IOU threshold  : 0.45
   - Label          : janken.names_jp
   - Log level      : 3
   - Title flag     : y
   - Speed flag     : y
   - Processed out  : non
   - Preprocessing  : False
   - Batch size     : 1
   - number of inf  : 1
   - With grid      : False
  [ WARN:0@14.381] global D:\a\opencv-python\opencv-python\opencv\modules\videoio\src\cap_msmf.cpp (539) `anonymous-namespace'::SourceReaderCB::~SourceReaderCB terminating async callback
  FPS average:       9.30
  Finished.
  
  (py38a) PS > python object_detect_yolo7.py -m ../yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx -l janken.names_jp -i ../../Images/janken3.jpg -d GPU -o janken3_yolov7-tiny_jk4.jpg
  Starting..
   - Program title  : Object detection YOLO V7
   - OpenCV version : 4.5.5
   - OpenVINO engine: 2022.1.0-7019-cdb9bec7210-releases/2022/1
   - Input image    : ../../Images/janken3.jpg
   - Model          : ../yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx
   - Device         : GPU
   - Confidence thr : 0.25
   - IOU threshold  : 0.45
   - Label          : janken.names_jp
   - Log level      : 3
   - Title flag     : y
   - Speed flag     : y
   - Processed out  : janken3_yolov7-tiny_jk4.jpg
   - Preprocessing  : False
   - Batch size     : 1
   - number of inf  : 1
   - With grid      : False
  FPS average:       9.90
  Finished.
  
  (py38a) PS > python object_detect_yolo7.py -m ../yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx -l janken.names_jp -i ../../Videos/janken_test2.mp4 -o janken_test2_yolov7-tiny_jk4.mp4
  Starting..
   - Program title  : Object detection YOLO V7
   - OpenCV version : 4.5.5
   - OpenVINO engine: 2022.1.0-7019-cdb9bec7210-releases/2022/1
   - Input image    : ../../Videos/janken_test2.mp4
   - Model          : ../yolov7-main/runs/train/yolov7-tiny_jk4/weights/best.onnx
   - Device         : CPU
   - Confidence thr : 0.25
   - IOU threshold  : 0.45
   - Label          : janken.names_jp
   - Log level      : 3
   - Title flag     : y
   - Speed flag     : y
   - Processed out  : janken_test2_yolov7-tiny_jk4.mp4
   - Preprocessing  : False
   - Batch size     : 1
   - number of inf  : 1
   - With grid      : False
  FPS average:       5.10
  Finished.

   

更新履歴 †

• 2023/02/22 初版
  
• 2023/09/10 全面改訂 YOLOv7 カスタムデータによる学習
  
   

参考資料 †

• Google Colaboratory
  
  • Colaboratory へようこそ
    
  • Google Colaboratoryの無料GPU環境を使ってみた
    

• YOLO V7
  
  • Official YOLOv7
    
  • YOLOv7_OpenVINO
    

• サイト内参考ページ
  
  • カスタムデータによる学習２「じゃんけんの判定１」
    
  • 学習パラメータ考察
    
  • カスタムデータによる学習３「じゃんけんの判定２」
    
  • カスタムデータによる学習４「じゃんけんの判定３」
    
  • 元になる学習モデルの違いによる考察
    
  • 「じゃんけん」カスタムデータによる学習モデルまとめ