Skip to content

README.md

Latest commit

 

History

History
87 lines (68 loc) · 3.35 KB

README.md

File metadata and controls

87 lines (68 loc) · 3.35 KB

CLI TOOL USAGE INSTRUCTIONS

Prerequisites

After installing the prerequisites, you may install this tool by running pip install . or pip install -e . from this folder.

Please note: You must have Python installed on your system and it (along with its 'site-packages' directory) must be in your PATH. If this is not the case, the CLI tool will install, but you won't be able to run it. Please see the Python documentation for installation instructions and post-install modifications (Mac/Linux)

CLI TOOL HELP & COMMANDS

Usage: azdacli [OPTIONS] COMMAND [ARGS]...

Options:
  --help                          Show help message and exit.

Commands:
  build
  create
  push

CREATE CUSTOM DEEPSTREAM DOCKER IMAGE COMMAND

Usage: azdacli create [OPTIONS]

Options:
  --deepstream-image-type         available options - [base|devel|triton|iot|samples]
                                  [default: samples]
  --device-type [dGPU|Jetson]     available options - [default: dGPU]
  --tag TEXT                      [required]
  --backend                       available options - [pytorch|dali|onnxruntime|fil|python|tensorflow1|tensorflow2|tensorrt|square|openvino_2021_4|identity|repeat]
  --add TEXT                      Provide: SOURCE file path <space>
                                  DESTINATION folder path - example:
                                  /home/source/folder/file /opt/nvidia/deepstr
                                  eam/deepstream-6.1/destination/folder/
  --run TEXT                      Type complete command in double quotes, example - "apt install python3-pip"
  --build-image / --no-build-image
                                  [default: build-image]
  --build-dir TEXT
                                  Provide file path of 'docker-build' folder
                                  under 'ds-ai-pipeline' code base. This is
                                  where docker file will be created
                                  [required]
  --help                          Show this message and exit.

PUSH DOCKER IMAGE TO CONTAINER REGISTRY

Usage: azdacli push [OPTIONS]

Options:
  --image-name TEXT          [required]
  --container-registry TEXT  [required]
  --username TEXT            [required]
  --password TEXT            [required, comes with prompt]
  --help                     Show this message and exit.

BUILD GENERIC DOCKER IMAGE

Usage: azdacli build [OPTIONS]

Options:
  --tag TEXT  [default: azdacliimage], Dockerfile has to be in the working directory
  --help      Show this message and exit.

EXAMPLE

azdacli create \
--deepstream-image-type triton \
--device-type dGPU \
--tag demo1 \
--backend tensorrt \
--backend pytorch \
--run "git clone https://github.com/NVIDIA-AI-IOT/deepstream_python_apps" --run "apt install -y python3-gi python3-dev python3-gst-1.0 python-gi-dev git python-dev python3 python3-pip python3.8-dev cmake g++ build-essential libglib2.0-dev libglib2.0-dev-bin libgstreamer1.0-dev libtool m4 autoconf automake libgirepository1.0-dev libcairo2-dev" --run "pip3 install https://github.com/NVIDIA-AI-IOT/deepstream_python_apps/releases/download/v1.1.3/pyds-1.1.3-py3-none-linux_x86_64.whl" \
--build-image --build-dir /home/ds-ai-pipeline/docker-build