README.rst

pl-objectdetection

Table of Contents

• Abstract
• Contributer
• Synopsis
• Description
• Agruments
• Run
  • Using docker run

Abstract

This is the original runnable object dection python3 scripts and its running environment container on ppc64le. To get the ppc64le chrisapp version, click here:

https://github.com/FNNDSC/pl-objectdetection_moc_ppc64

check the parent scripts & amd64 container powered by nvidia:

https://github.com/NVIDIA/object-detection-tensorrt-example

Docker container published on dockerhub:

https://hub.docker.com/repository/docker/fnndsc/pl-objectdetection_moc_ppc64

For amd64 version scripts check here:

https://github.com/FNNDSC/object-detection-tensorrt-example

amd64 version Chris app:

https://github.com/FNNDSC/pl-object-detection

Contributer

@PupilTong modified the original object detection for benchmarking usage & built runnable ppc64le running environment. @h4x0rMadness made it chris app & benchmarking outputs

An app to ...

Synopsis

python objectdetection.py                                         \\
        [-h] [--help]                                               \\
        [--json]                                                    \\
        [--man]                                                     \\
        [--meta]                                                    \\
        [--savejson <DIR>]                                          \\
        [-v <level>] [--verbosity <level>]                          \\
        [--version]                                                 \\
        [--file <filename>]                                         \\
        <inputDir>                                                  \\
        <outputDir>

Description

objectdetection.py is a ChRIS-based application that...

Agruments

[-v <level>] [--verbosity <level>]
Verbosity level for app. Not used currently.

[--version]
If specified, print version number.

[--man]
If specified, print (this) man page.

[--meta]
If specified, print plugin meta data.

Run

Using docker run

To run using docker, be sure to assign an "input" directory to /incoming and an output directory to /outgoing. Make sure that the $(pwd)/out directory is world writable!

Now, prefix all calls with

docker run --security-opt label=type:nvidia_container_t
            -v $(pwd)/in:/incoming -v $(pwd)/out:/outgoing
            docker.io/fnndsc/pl-objectdetection_x86
            objectdetection.py -f filename.webm
            /incoming /outgoing

Thus, getting inline help is:

mkdir in out && chmod 777 out
docker run --security-opt label=type:nvidia_container_t
            -v $(pwd)/in:/incoming -v $(pwd)/out:/outgoing
            docker.io/fnndsc/pl-objectdetection_x86
            objectdetection.py -f filename.webm
            /incoming /outgoing