NCAppRun3

Neural Compute Application 実行スクリプト３ †

　OpenVINO™ ツールキット付属のデモや作成したアプリケーションの実行を簡単にするために、シェルスクリプトを使った起動ファイルを作成する。
　オリジナルプログラムの入力パラメータを引き継ぎ各プログラムのディレクトリで簡単に実行できるようにする

Neural Compute Application 実行スクリプト３
参考資料

※ 最終更新:2021/08/11　

↑

共通仕様 †

↑

主な目的 †

プログラムの実行ディレクトリ・実行コマンドを共通にする。
プログラムの実行に必要な各種パラメータはオリジナルの指定を引き継ぐ。

↑

スクリプトの場所 †

~/run_app3

↑

スクリプトの実行 †

基本指定

$ ./XXXXXXXX.sh [ パラメータ１ ] [ パラメータ２ ]...

入力パラメータ表示
```
$ ./XXXXXXXX.sh -h
```
- パラメータを省略した場合は、あらかじめ設定されているデフォールト値が使われる。
- パラメータはオリジナルプログラムにそのまま引き継がれる。

↑

OpenVINO™ Toolkit の使い方 †

↑

DBFace on Intel® OpenVINO™ †

▼　「dbface-infer.sh」

#!/bin/sh

SHELL_NAME='dbface-infer.sh'
PROG_NAME='dbface-infer2.py'
RUN_DIR="$HOME/workspace/apps3/dbface-on-openvino-master"

echo "[$SHELL_NAME] \n'$PROG_NAME' Run !!"
cd $RUN_DIR

echo " DIR:     $RUN_DIR"
echo " COMMAND: python3 $PROG_NAME $@"

python3 $PROG_NAME $@

実行コマンド
```
$ ./dbface-infer.sh
```

実行コマンド（パラメータ表示）

$ ./dbface-infer.sh -h
[dbface-infer.sh] 
'dbface-infer2.py' Run !!
 DIR:     /home/mizutu/workspace/apps3/dbface-on-openvino-master
 COMMAND: python3 dbface-infer2.py -h

--- DBFace single-stage face detector 2 ---
4.5.3-openvino
OpenVINO inference_engine: 2021.4.0-3839-cd81789d294-releases/2021/4

usage: dbface-infer2.py [-h] [-i IMAGE_FILE] [-m MODEL] [-d DEVICE] [-t TITLE]
                        [-s SPEED] [-o IMAGE_OUT]

optional arguments:
  -h, --help            show this help message and exit
  -i IMAGE_FILE, --image IMAGE_FILE
                        Absolute path to image file. Default value is
                        ./selfie.jpg
  -m MODEL, --model MODEL
                        Detector Path to an .xml file with a trained
                        model.Default value is ./ir-models/dbface-4vga-
                        fp16/dbface.xml
  -d DEVICE, --device DEVICE
                        Optional. Specify a target device to infer on. CPU,
                        GPU, FPGA, HDDL or MYRIAD is acceptable. The demo will
                        look for a suitable plugin for the device specified.
                        Default value is CPU
  -t TITLE, --title TITLE
                        Program title flag.(y/n) Default value is 'y'
  -s SPEED, --speed SPEED
                        Speed display flag.(y/n) Default calue is 'y'
  -o IMAGE_OUT, --out IMAGE_OUT
                        Processed image file path. Default value is 'non'

↑

Object (Face) Detection Demo with Adaptive Partitioning to Improve Detection Rate †

▼　「objdet-adaptive-partitioning.sh」

#!/bin/sh

SHELL_NAME='objdet-adaptive-partitioning.sh'
PROG_NAME='objdet-adaptive-partitioning2.py'
RUN_DIR="$HOME/workspace/apps3/object-detection-with-adaptive-partitioning-master"

echo "[$SHELL_NAME] \n'$PROG_NAME' Run !!"
cd $RUN_DIR

echo " DIR:     $RUN_DIR"
echo " COMMAND: python3 $PROG_NAME $@"

python3 $PROG_NAME $@

実行コマンド
```
$ ./objdet-adaptive-partitioning.sh
```

実行コマンド（パラメータ表示）

$ ./objdet-adaptive-partitioning.sh -h
[objdet-adaptive-partitioning.sh] 
'objdet-adaptive-partitioning2.py' Run !!
 DIR:     /home/mizutu/workspace/apps3/object-detection-with-adaptive-partitioning-master
 COMMAND: python3 objdet-adaptive-partitioning2.py -h

--- Object Detection with adaptive partitioning 2 ---
4.5.3-openvino
OpenVINO inference_engine: 2021.4.0-3839-cd81789d294-releases/2021/4

usage: objdet-adaptive-partitioning2.py [-h] [-i IMAGE_FILE] [-m MODEL]
                                        [-d DEVICE] [-o IMAGE_OUT]

optional arguments:
  -h, --help            show this help message and exit
  -i IMAGE_FILE, --image IMAGE_FILE
                        Absolute path to image file. Default value is
                        ./image.jpg
  -m MODEL, --model MODEL
                        Detector Path to an .xml file with a trained model.
                        Default value is /home/mizutu/model/intel/FP16/face-
                        detection-0200.xml
  -d DEVICE, --device DEVICE
                        Optional. Specify a target device to infer on. CPU,
                        GPU, FPGA, HDDL or MYRIAD is acceptable. The demo will
                        look for a suitable plugin for the device specified.
                        Default value is CPU
  -o IMAGE_OUT, --out IMAGE_OUT
                        Processed image file path. Default value is 'non'

▼　「naive-object.sh」

#!/bin/sh

SHELL_NAME='naive-object.sh'
PROG_NAME='naive-objdet2.py'
RUN_DIR="$HOME/workspace/apps3/object-detection-with-adaptive-partitioning-master"

echo "[$SHELL_NAME] \n'$PROG_NAME' Run !!"
cd $RUN_DIR

echo " DIR:     $RUN_DIR"
echo " COMMAND: python3 $PROG_NAME $@"

python3 $PROG_NAME $@

実行コマンド
```
$ ./naive-object.sh
```

実行コマンド（パラメータ表示）

$ ./naive-object.sh -h
[naive-object.sh] 
'naive-objdet2.py' Run !!
 DIR:     /home/mizutu/workspace/apps3/object-detection-with-adaptive-partitioning-master
 COMMAND: python3 naive-objdet2.py -h

--- Native Object Detection 2 ---
4.5.3-openvino
OpenVINO inference_engine: 2021.4.0-3839-cd81789d294-releases/2021/4

usage: naive-objdet2.py [-h] [-i IMAGE_FILE] [-m MODEL] [-d DEVICE]
                        [-o IMAGE_OUT]

optional arguments:
  -h, --help            show this help message and exit
  -i IMAGE_FILE, --image IMAGE_FILE
                        Absolute path to image file. Default value is
                        ./image.jpg
  -m MODEL, --model MODEL
                        Detector Path to an .xml file with a trained model.
                        Default value is /home/mizutu/model/intel/FP16/face-
                        detection-0200.xml
  -d DEVICE, --device DEVICE
                        Optional. Specify a target device to infer on. CPU,
                        GPU, FPGA, HDDL or MYRIAD is acceptable. The demo will
                        look for a suitable plugin for the device specified.
                        Default value is CPU
  -o IMAGE_OUT, --out IMAGE_OUT
                        Processed image file path. Default value is 'non'

↑

Object Tracking with Line Crossing and Area Intrusion Detection †

▼　「object-detection-and-line-cross.sh」

#!/bin/sh

SHELL_NAME='object-detection-and-line-cross.sh'
PROG_NAME='object-detection-and-line-cross2.py'
RUN_DIR="$HOME/workspace/apps3/object-tracking-line-crossing-area-intrusion-master"

echo "[$SHELL_NAME] \n'$PROG_NAME' Run !!"
cd $RUN_DIR

echo " DIR:     $RUN_DIR"
echo " COMMAND: python3 $PROG_NAME $@"

python3 $PROG_NAME $@

実行コマンド
```
$ ./object-detection-and-line-cross.sh
```

実行コマンド（パラメータ表示）

$ ./object-detection-and-line-cross.sh -h
[object-detection-and-line-cross.sh] 
'object-detection-and-line-cross2.py' Run !!
 DIR:     /home/mizutu/workspace/apps3/object-tracking-line-crossing-area-intrusion-master
 COMMAND: python3 object-detection-and-line-cross2.py -h

--- Object Tracking with Line Crossing 2 ---
4.5.3-openvino
OpenVINO inference_engine: 2021.4.0-3839-cd81789d294-releases/2021/4

usage: object-detection-and-line-cross2.py [-h] [-i IMAGE_FILE] [-d DEVICE]
                                           [-t TITLE] [-s SPEED]
                                           [-o IMAGE_OUT]

optional arguments:
  -h, --help            show this help message and exit
  -i IMAGE_FILE, --image IMAGE_FILE
                        Absolute path to image file
  -d DEVICE, --device DEVICE
                        Optional. Specify a target device to infer on. CPU,
                        GPU, FPGA, HDDL or MYRIAD is acceptable. The demo will
                        look for a suitable plugin for the device specified.
                        Default value is CPU
  -t TITLE, --title TITLE
                        Program title flag.(y/n) Default value is 'y'
  -s SPEED, --speed SPEED
                        Speed display flag.(y/n) Default calue is 'y'
  -o IMAGE_OUT, --out IMAGE_OUT
                        Processed image file path. Default value is 'non'

↑

Gaze Estimation Demo with Sparking Laser Beam ;-) †

▼　「gaze-estimation.sh」

#!/bin/sh

SHELL_NAME='gaze-estimation.sh'
PROG_NAME='gaze-estimation2.py'
RUN_DIR="$HOME/workspace/apps3/gaze-estimation-with-laser-sparking-master"

echo "[$SHELL_NAME] \n'$PROG_NAME' Run !!"
cd $RUN_DIR

echo " DIR:     $RUN_DIR"
echo " COMMAND: python3 $PROG_NAME $@"

python3 $PROG_NAME $@

実行コマンド
```
$ ./gaze-estimation.sh
```

実行コマンド（パラメータ表示）

$ ./gaze-estimation.sh -h
[gaze-estimation.sh] 
'gaze-estimation2.py' Run !!
 DIR:     /home/mizutu/workspace/apps3/gaze-estimation-with-laser-sparking-master
 COMMAND: python3 gaze-estimation2.py -h

--- Gaze Estimation Demo with Sparking Laser Beam 2 ---
4.5.3-openvino
OpenVINO inference_engine: 2021.4.0-3839-cd81789d294-releases/2021/4

usage: gaze-estimation2.py [-h] [-i IMAGE_FILE] [-d DEVICE] [-t TITLE]
                           [-s SPEED] [-o IMAGE_OUT]

optional arguments:
  -h, --help            show this help message and exit
  -i IMAGE_FILE, --image IMAGE_FILE
                        Absolute path to image file. Default value is cam
  -d DEVICE, --device DEVICE
                        Optional. Specify a target device to infer on. CPU,
                        GPU, FPGA, HDDL or MYRIAD is acceptable. The demo will
                        look for a suitable plugin for the device specified.
                        Default value is CPU
  -t TITLE, --title TITLE
                        Program title flag.(y/n) Default value is 'y'
  -s SPEED, --speed SPEED
                        Speed display flag.(y/n) Default calue is 'y'
  -o IMAGE_OUT, --out IMAGE_OUT
                        Processed image file path. Default value is 'non'

↑

Interactive Image Inpainting Demo †

▼　「image-inpainting.sh」

#!/bin/sh

SHELL_NAME='image-inpainting.sh'
PROG_NAME='image-inpainting2.py'
RUN_DIR="$HOME/workspace/apps3/interactive-image-inpainting-master"

echo "[$SHELL_NAME] \n'$PROG_NAME' Run !!"
cd $RUN_DIR

echo " DIR:     $RUN_DIR"
echo " COMMAND: python3 $PROG_NAME $@"

python3 $PROG_NAME $@

実行コマンド
```
$ ./image-inpainting.sh
```

実行コマンド（パラメータ表示）

$ ./image-inpainting.sh -h
[image-inpainting.sh] 
'image-inpainting2.py' Run !!
 DIR:     /home/mizutu/workspace/apps3/interactive-image-inpainting-master
 COMMAND: python3 image-inpainting2.py -h

--- image inpainting 2 ---
4.5.3-openvino
OpenVINO inference_engine: 2021.4.0-3839-cd81789d294-releases/2021/4

usage: image-inpainting2.py [-h] [-i IMAGE_FILE] [-d DEVICE] [-o IMAGE_OUT]

optional arguments:
  -h, --help            show this help message and exit
  -i IMAGE_FILE, --image IMAGE_FILE
                        Absolute path to image file. Default value is
                        ./image.jpg
  -d DEVICE, --device DEVICE
                        Optional. Specify a target device to infer on. CPU,
                        GPU, FPGA, HDDL or MYRIAD is acceptable. The demo will
                        look for a suitable plugin for the device specified.
                        Default value is CPU
  -o IMAGE_OUT, --out IMAGE_OUT
                        Processed image file path. Default value is 'non'

↑

Handwritten Japanese Deep Learning Based OCR with Touch Panel Demo †

▼　「handwritten-japanese-OCR-touch-panel-demo.sh」

#!/bin/sh

SHELL_NAME='handwritten-japanese-OCR-touch-panel-demo.sh'
PROG_NAME='handwritten-japanese-OCR-touch-panel-demo2.py'
RUN_DIR="$HOME/workspace/apps3/handwritten-japanese-ocr-master"

echo "[$SHELL_NAME] \n'$PROG_NAME' Run !!"
cd $RUN_DIR

echo " DIR:     $RUN_DIR"
echo " COMMAND: python3 $PROG_NAME $@"

python3 $PROG_NAME $@

実行コマンド

$ ./handwritten-japanese-OCR-touch-panel-demo.sh

実行コマンド（パラメータ表示）

$ ./handwritten-japanese-OCR-touch-panel-demo.sh -h
[handwritten-japanese-OCR-touch-panel-demo.sh] 
'handwritten-japanese-OCR-touch-panel-demo2.py' Run !!
 DIR:     /home/mizutu/workspace/apps3/handwritten-japanese-ocr-master
 COMMAND: python3 handwritten-japanese-OCR-touch-panel-demo2.py -h

--- Handwritten Japanese OCR demo 2 ---
4.5.3-openvino
OpenVINO inference_engine: 2021.4.0-3839-cd81789d294-releases/2021/4

Handwritten Japanese OCR Demo
ESC: Quit
Mouse L-Button: Draw
Mouse R-Button: Erase
Threshold = Text area detect threshold
usage: handwritten-japanese-OCR-touch-panel-demo2.py [-h] [-d DEVICE]
                                                     [-o IMAGE_OUT]

optional arguments:
  -h, --help            show this help message and exit
  -d DEVICE, --device DEVICE
                        Optional. Specify a target device to infer on. CPU,
                        GPU, FPGA, HDDL or MYRIAD is acceptable. The demo will
                        look for a suitable plugin for the device specified.
                        Default value is CPU
  -o IMAGE_OUT, --out IMAGE_OUT
                        Processed image file path. Default value is 'non'

↑

Intel(r) OpenVINO(tm) IR model utility tool †

▼　「ir-summary.sh」

#!/bin/sh

SHELL_NAME='ir-summary.sh'
PROG_NAME='ir-summary.py'
RUN_DIR="$HOME/workspace/apps3/openvino-ir-utility-master"

echo "[$SHELL_NAME] \n'$PROG_NAME' Run !!"
cd $RUN_DIR

echo " DIR:     $RUN_DIR"
echo " COMMAND: python3 $PROG_NAME $@"

python3 $PROG_NAME $@

実行コマンド
```
$ ./ir-summary.sh
```

実行コマンド（パラメータ表示）

$ ./ir-summary.sh -h
[ir-summary.sh] 
'ir-summary.py' Run !!
 DIR:     /home/mizutu/workspace/apps3/openvino-ir-utility-master
 COMMAND: python3 ir-summary.py -h
usage: ir-summary.py [-h] [-m MODEL] [-d DIR] [-v]

optional arguments:
  -h, --help            show this help message and exit
  -m MODEL, --model MODEL
                        input IR model path
  -d DIR, --dir DIR     input IR model directory
  -v, --verbose         output detailed information

↑

更新履歴 †

2021/08/11 初版

↑

最新の20件

Neural Compute Application 実行スクリプト３ †

共通仕様 †

主な目的 †

スクリプトの場所 †

スクリプトの実行 †

OpenVINO™ Toolkit の使い方 †

DBFace on Intel® OpenVINO™ †

Object (Face) Detection Demo with Adaptive Partitioning to Improve Detection Rate †

Object Tracking with Line Crossing and Area Intrusion Detection †

Gaze Estimation Demo with Sparking Laser Beam ;-) †

Interactive Image Inpainting Demo †

Handwritten Japanese Deep Learning Based OCR with Touch Panel Demo †

Intel(r) OpenVINO(tm) IR model utility tool †

更新履歴 †

参考資料 †