OpenVINO Tips

最新の20件

私的AI研究会 > OpenVINO Tips

OpenVINO™ Toolkit Tips †

OpenVINO™ Toolkit「2021.2」 Tips †

学習済みモデルを準備する †

　OpenVINO™ ツールキットのバージョンに合った学習済みモデルをダウンロードする。
インストール・バージョンは 2021.2

• 学習済みモデルの場所
  → https://download.01.org/opencv/2021/openvinotoolkit/2021.2/open_model_zoo/models_bin/3/

• 学習済みモデルのドキュメント
  Model Zoo -学習済み機械学習モデルのアーカイブ-
  Overview of OpenVINO™ Toolkit Intel's Pre-Trained Models
  Overview of OpenVINO™ Toolkit Public Models
  Overview of OpenVINO™ Toolkit Public Models (GitHub)
  

モデルの入出力情報を知る †

• モデル・チェックプログラム

  vi model_check.py
  
  # -*- coding: utf-8 -*-
  ##--------------------------------------------
  ## OpenVINO™ 
  ##  Model Check Program
  ##               2021.02.10 Masahiro Izutsu
  ##--------------------------------------------
  ##  python3 model_check.py <model pass>
  ##--------------------------------------------
  
  import sys
  import cv2
  import numpy as np
  
  def main():
  
      # モジュール読み込み
      from openvino.inference_engine import IECore
  
      # Inference Engineコアオブジェクトの生成
      ie = IECore()
  
      # IRモデルファイルの読み込み
      net = ie.read_network(model=model+'.xml', weights=model+'.bin')
  
      # 入出力blobの名前の取得、入力blobのシェイプの取得
      input_key = list(net.input_info.keys())[0]         # 入力データ・キー名
      input_blob_name  = net.input_info[input_key].name
      output_blob_name = next(iter(net.outputs))
      batch,channel,height,width = net.input_info[input_blob_name].input_data.shape
  
      exec_net = ie.load_network(network=net, device_name='MYRIAD', num_requests=1)
  
      print("input blob: name='{}', N={}, C={}, H={}, W={}".format(input_blob_name, batch, channel, height, width))
  
  if __name__ == '__main__':
  
      # 学習済みモデル名
      args = sys.argv
      if 2 <= len(args):
          model = args[1]
      else:
          model = './public/mobilenet-ssd/FP16/mobilenet-ssd'
      print("model = '{}'".format(model))
      main()

• コマンドラインからモデルファイルのパス＋モデル名を指定して実行する（実行例）

  pi@raspberrypi:~/workspace $ python3 model_check.py
  model = './public/mobilenet-ssd/FP16/mobilenet-ssd'
  input blob: name='data', N=1, C=3, H=300, W=300
  
  pi@raspberrypi:~/workspace $ python3 model_check.py FP16/landmarks-regression-retail-0009
  model = 'FP16/landmarks-regression-retail-0009'
  input blob: name='0', N=1, C=3, H=48, W=48
  
  pi@raspberrypi:~/workspace $ python3 model_check.py ./FP16/face-detection-adas-0001
  model = './FP16/face-detection-adas-0001'
  input blob: name='data', N=1, C=3, H=384, W=672
  
  pi@raspberrypi:~/workspace $ python3 model_check.py public/squeezenet1.1/FP16/squeezenet1.1
  model = 'public/squeezenet1.1/FP16/squeezenet1.1'
  input blob: name='data', N=1, C=3, H=227, W=227

※ ドキュメントの入力 Name は違っている場合が多いので確認しておく。

OpenVINO™ ツールキット(2021.2) に含まれる OpenCV のバージョン †

Raspberry Pi OS (Rasbian) buster 2020/1202版 †

　システムイメージ → <<raspi_ncappzoo_2021_0225_001c.img>>

• OpenCV 詳細 調べる

  $ python3
  Python 3.7.3 (default, Jul 25 2020, 13:03:44) 
  [GCC 8.3.0] on linux
  Type "help", "copyright", "credits" or "license" for more information.
  >>> import cv2
  >>> print(cv2.getBuildInformation())

• 表示結果

  General configuration for OpenCV 4.5.1-openvino 
  =====================================
    Version control:               705e7b207c4c95231fb18c7ddfad353795aa6531
  
    Platform:
      Timestamp:                   2020-12-01T12:29:06Z
      Host:                        Linux 4.15.0-29-generic x86_64
      Target:                      Linux 1 armv7l
      CMake:                       3.7.2
      CMake generator:             Ninja
      CMake build tool:            /usr/bin/ninja
      Configuration:               Release
  
    CPU/HW features:
      Baseline:                    NEON
        required:                  NEON
        disabled:                  VFPV3
  
    C/C++:
      Built as dynamic libs?:      YES
      C++ standard:                11
      C++ Compiler:                /usr/bin/arm-linux-gnueabihf-g++  (ver 6.3.0)
      C++ flags (Release):         -mthumb  -fdata-sections -Wa,--noexecstack -fsigned-char -Wno-psabi   -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wuninitialized -Winit-self -Wno-psabi -Wsuggest-override -Wno-delete-non-virtual-dtor -Wno-comment -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections  -mfpu=neon -fvisibility=hidden -fvisibility-inlines-hidden -fstack-protector-strong -fPIC -O2 -DNDEBUG  -DNDEBUG -D_FORTIFY_SOURCE=2
      C++ flags (Debug):           -mthumb  -fdata-sections -Wa,--noexecstack -fsigned-char -Wno-psabi   -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wuninitialized -Winit-self -Wno-psabi -Wsuggest-override -Wno-delete-non-virtual-dtor -Wno-comment -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections  -mfpu=neon -fvisibility=hidden -fvisibility-inlines-hidden -fstack-protector-strong -fPIC -g  -O0 -DDEBUG -D_DEBUG
      C Compiler:                  /usr/bin/arm-linux-gnueabihf-gcc
     C flags (Release):           -mthumb  -fdata-sections -Wa,--noexecstack -fsigned-char -Wno-psabi   -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wuninitialized -Winit-self -Wno-psabi -Wno-comment -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections  -mfpu=neon -fvisibility=hidden -fstack-protector-strong -fPIC -O2 -DNDEBUG  -DNDEBUG -D_FORTIFY_SOURCE=2
      C flags (Debug):             -mthumb  -fdata-sections -Wa,--noexecstack -fsigned-char -Wno-psabi   -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wuninitialized -Winit-self -Wno-psabi -Wno-comment -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections  -mfpu=neon -fvisibility=hidden -fstack-protector-strong -fPIC -g  -O0 -DDEBUG -D_DEBUG
      Linker flags (Release):      -Wl,--fix-cortex-a8 -Wl,--no-undefined -Wl,--gc-sections -Wl,-z,noexecstack -Wl,-z,relro -Wl,-z,now   -Wl,--gc-sections -Wl,--as-needed  -z noexecstack -z relro -z now  
      Linker flags (Debug):        -Wl,--fix-cortex-a8 -Wl,--no-undefined -Wl,--gc-sections -Wl,-z,noexecstack -Wl,-z,relro -Wl,-z,now   -Wl,--gc-sections -Wl,--as-needed  -z noexecstack -z relro -z now  
      ccache:                      YES
      Precompiled headers:         NO
      Extra dependencies:          dl m pthread rt
      3rdparty dependencies:
  
    OpenCV modules:
      To be built:                 calib3d core dnn features2d flann gapi highgui imgcodecs imgproc ml objdetect photo python3 stitching ts video videoio
      Disabled:                    world
      Disabled by dependency:      -
      Unavailable:                 java js python2
      Applications:                tests perf_tests apps
      Documentation:               NO
      Non-free algorithms:         NO
  
    GUI: 
      GTK+:                        YES (ver 3.22.11)
        GThread :                  YES (ver 2.50.3)
        GtkGlExt:                  NO
  
    Media I/O: 
      ZLib:                        build (ver 1.2.11)
      JPEG:                        build-libjpeg-turbo (ver 2.0.6-62)
      PNG:                         build (ver 1.6.37)
      HDR:                         YES
      SUNRASTER:                   YES
      PXM:                         YES
      PFM:                         YES
  
    Video I/O:
      FFMPEG:                      YES
        avcodec:                   YES (57.64.101)
        avformat:                  YES (57.56.101)
        avutil:                    YES (55.34.101)
        swscale:                   YES (4.2.100)
        avresample:                NO
      GStreamer:                   YES (1.10.4)
      v4l/v4l2:                    YES (linux/videodev2.h)
  
    Parallel framework:            pthreads
  
    Trace:                         YES (with Intel ITT)
  
    Other third-party libraries:
      Inference Engine:            YES (2021020000 / 2.1.0)
          * libs:                  /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/debian9arm/ie/inference_engine/lib/armv7l/libinference_engine.so
          * includes:              /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/debian9arm/ie/inference_engine/include
     nGraph:                      YES (0.0.0+d2e3e1f)
          * libs:                  /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/debian9arm/ie/ngraph/lib/libngraph.so
          * includes:              /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/debian9arm/ie/ngraph/include
      Custom HAL:                  YES (carotene (ver 0.0.1))
      Protobuf:                    build (3.5.1)
  
    Python 3:
      Interpreter:                 /usr/bin/python3 (ver 3.5.3)
      Libraries:                   
      numpy:                       /usr/lib/python3.5/dist-packages/numpy/core/include (ver undefined - cannot be probed because of the cross-compilation)
      install path:                /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/debian9arm/build_release/install/python/python3
  
    Python (for build):            /usr/bin/python2.7
  
    Install to:                    /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/debian9arm/build_release/install
  -----------------------------------------------------------------

Ubuntu 20.04 LTS版 †

• OpenCV 詳細 調べる

  $ python3
  Python 3.8.5 (default, Jul 28 2020, 12:59:40) 
  [GCC 9.3.0] on linux
  Type "help", "copyright", "credits" or "license" for more information.
  >>> import cv2
  >>> print(cv2.getBuildInformation())

• 表示結果

  General configuration for OpenCV 4.5.1-openvino =====================================
    Version control:               705e7b207c4c95231fb18c7ddfad353795aa6531
  
    Platform:
      Timestamp:                   2020-12-01T12:30:45Z
      Host:                        Linux 4.15.0-29-generic x86_64
      CMake:                       3.14.5
      CMake generator:             Ninja
      CMake build tool:            /opt/miniconda/envs/build/bin/ninja
      Configuration:               Release
  
    CPU/HW features:
      Baseline:                    SSE SSE2 SSE3 SSSE3 SSE4_1 POPCNT SSE4_2
        requested:                 SSE4_2
      Dispatched code generation:  FP16 AVX AVX2 AVX512_SKX
        requested:                 SSE4_1 SSE4_2 AVX FP16 AVX2 AVX512_SKX
        FP16 (1 files):            + FP16 AVX
        AVX (5 files):             + AVX
        AVX2 (31 files):           + FP16 FMA3 AVX AVX2
        AVX512_SKX (7 files):      + FP16 FMA3 AVX AVX2 AVX_512F AVX512_COMMON AVX512_SKX
  
    C/C++:
      Built as dynamic libs?:      YES
      C++ standard:                11
      C++ Compiler:                /usr/bin/c++  (ver 9.3.0)
      C++ flags (Release):         -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wuninitialized -Winit-self -Wsuggest-override -Wno-delete-non-virtual-dtor -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -Wno-long-long -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections  -msse -msse2 -msse3 -mssse3 -msse4.1 -mpopcnt -msse4.2 -fvisibility=hidden -fvisibility-inlines-hidden -fstack-protector-strong -fPIC -O2 -DNDEBUG  -DNDEBUG -D_FORTIFY_SOURCE=2
      C++ flags (Debug):           -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wuninitialized -Winit-self -Wsuggest-override -Wno-delete-non-virtual-dtor -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -Wno-long-long -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections  -msse -msse2 -msse3 -mssse3 -msse4.1 -mpopcnt -msse4.2 -fvisibility=hidden -fvisibility-inlines-hidden -fstack-protector-strong -fPIC -g  -O0 -DDEBUG -D_DEBUG
      C Compiler:                  /usr/bin/cc
      C flags (Release):           -fsigned-char -W -Wall -Werror=return-type -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wuninitialized -Winit-self -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -Wno-long-long -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections  -msse -msse2 -msse3 -mssse3 -msse4.1 -mpopcnt -msse4.2 -fvisibility=hidden -fstack-protector-strong -fPIC -O2 -DNDEBUG  -DNDEBUG -D_FORTIFY_SOURCE=2
      C flags (Debug):             -fsigned-char -W -Wall -Werror=return-type -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wuninitialized -Winit-self -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -Wno-long-long -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections  -msse -msse2 -msse3 -mssse3 -msse4.1 -mpopcnt -msse4.2 -fvisibility=hidden -fstack-protector-strong -fPIC -g  -O0 -DDEBUG -D_DEBUG
      Linker flags (Release):      -Wl,--exclude-libs,libippicv.a -Wl,--exclude-libs,libippiw.a   -Wl,--gc-sections -Wl,--as-needed  -z noexecstack -z relro -z now  
      Linker flags (Debug):        -Wl,--exclude-libs,libippicv.a -Wl,--exclude-libs,libippiw.a   -Wl,--gc-sections -Wl,--as-needed  -z noexecstack -z relro -z now  
      ccache:                      YES
      Precompiled headers:         NO
      Extra dependencies:          dl m pthread rt
      3rdparty dependencies:
  
    OpenCV modules:
      To be built:                 calib3d core dnn features2d flann gapi highgui imgcodecs imgproc ml objdetect photo python3 stitching ts video videoio
      Disabled:                    world
      Disabled by dependency:      -
      Unavailable:                 java js python2
      Applications:                tests perf_tests apps
      Documentation:               NO
      Non-free algorithms:         NO
  
    GUI: 
      GTK+:                        YES (ver 3.24.20)
        GThread :                  YES (ver 2.64.3)
        GtkGlExt:                  NO
  
    Media I/O: 
      ZLib:                        build (ver 1.2.11)
      JPEG:                        build-libjpeg-turbo (ver 2.0.6-62)
      PNG:                         build (ver 1.6.37)
      HDR:                         YES
      SUNRASTER:                   YES
      PXM:                         YES
      PFM:                         YES
  
    Video I/O:
      FFMPEG:                      YES
        avcodec:                   YES (58.54.100)
        avformat:                  YES (58.29.100)
        avutil:                    YES (56.31.100)
        swscale:                   YES (5.5.100)
        avresample:                YES (4.0.0)
      GStreamer:                   YES (1.16.2)
      v4l/v4l2:                    YES (linux/videodev2.h)
      Intel Media SDK:             YES (/mnt/nfs/msdk/lin-18.4.1/lib64/libmfx.so)
  
    Parallel framework:            TBB (ver 2020.1 interface 11101)
  
    Trace:                         YES (with Intel ITT)
  
    Other third-party libraries:
      Intel IPP:                   2020.0.0 Gold [2020.0.0]
             at:                   /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/ubuntu20/build_release/3rdparty/ippicv/ippicv_lnx/icv
      Intel IPP IW:                sources (2020.0.0)
                at:                /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/ubuntu20/build_release/3rdparty/ippicv/ippicv_lnx/iw
      Inference Engine:            YES (2021020000 / 2.1.0)
          * libs:                  /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/ubuntu20/ie/inference_engine/lib/intel64/libinference_engine.so
          * includes:              /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/ubuntu20/ie/inference_engine/include
      nGraph:                      YES (0.0.0+d2e3e1f)
          * libs:                  /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/ubuntu20/ie/ngraph/lib/libngraph.so
          * includes:              /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/ubuntu20/ie/ngraph/include
      Custom HAL:                  NO
      Protobuf:                    build (3.5.1)
  
    OpenCL:                        YES (no extra features)
      Include path:                /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/ubuntu20/opencv/3rdparty/include/opencl/1.2
      Link libraries:              Dynamic load
  
    Python 3:
      Interpreter:                 /opt/miniconda/envs/py3_env/bin/python (ver 3.4.5)
      Libraries:                   /opt/miniconda/envs/py3_env/lib/libpython3.4m.so (ver 3.4.5)
      numpy:                       /opt/miniconda/envs/py3_env/lib/python3.4/site-packages/numpy/core/include (ver 1.11.3)
      install path:                /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/ubuntu20/build_release/install/python/python3
  
    Python (for build):            /usr/bin/python2.7
  
    Install to:                    /home/jenkins/workspace/OpenCV/OpenVINO/2021.2/build/ubuntu20/build_release/install
  -----------------------------------------------------------------

   

OpenVINO™ ツールキット「2021.2」に付属するデモの種類 †

• ubuntu20.04 OpenVINO™インストールフォルダ

  /opt/intel/openvino_2021/
               │
               ├── deployment_tools
                         │
                         ├── demo
                         │        ├── demo_benchmark_app.sh
                         │        ├── demo_security_barrier_camera.sh
                         │        ├── demo_squeezenet_download_convert_run.sh
                         │
                         ├── inference_engine                   
                         │        │
                         │        ├── samples
                         │        │        ├── c
                         │        │        ├── cpp
                         │        │        ├── python
                         │        │
                         │        ├── demos (../open_model_zoo/demos)
                         │        │        ├── buld_demos.sh
                         │        │        │
                         │        │        ├── python_demos
                         │
                         ├── open_model_zoo
                         │        │
                         │        ├── demos
                         │        │        ├── buld_demos.sh
                         │        │        │
                         │        │        ├── python_demos
                         │        │        │

  　
• デモのある場所の詳細 → OpenVINO™ に付属するデモソフト一覧
  

OpenVINO™ 動作確認用デモ (deployment_tools/demo) †

• モデルのダウンロードからアプリのビルドまで一括で実行するスクリプト付きのデモ
  
• OpenVINO™ インストール後に動作確認として使用可能
  

Inference Engine サンプル (deployment_tools/inference_engine/samples) †

• シンプルなコーディングになっており提供される API の使い方を理解するのに使用
  
• その分デモとしてもシンプル
  
• 「c」「c++」「python」それぞれのソースコードとmakefile
  

※ ~/openvino_demos_python ディレクトリに python/ フォルダを python_sample/ フォルダとしてコピーして使用することにする。

Inference Engine デモ (deployment_tools/open_model_zoo/demos) †

• Open Model Zoo(OMZ) で提供されるモデルの使い方の参考として活用
  
• デモとしても見栄えのするものが多いがその分コードは少し複雑
  
• deployment_tools/inference_engine/demos/ ディレクトリと同じ
  

※ フォルダ内の build_demos.sh を実行することで、~/omz_demos_build　(~/inference_engine_demos_build) ディレクトリに実行環境がコピーされる。~/omz_demos_build を使用することにする

更新履歴 †

• 2021/03/19 初版
   

参考資料 †

• オフィシャルサイト
  
  • Intel® OpenVINO™ Toolkit Storage
    
  • Open Model Zoo Demos (latest)
  • Inference Engine Samples

• OpenVINO™ / アクセラレーター
• OpenVINO™ に付属しているデモまとめ (2020.3 ベース)
• デモのサンプルビデオ