Inference DS consists of two parts: the Inference DS Core and the User Interface. The Core is available as a Python package. The User Interface is a portable executable that can be used to connect to a Core running on the same system or remotely. Go to the section Inference DS Releases to download both for your particular platform or follow the installation guide below.

Installation Guide

Windows

You can directly use the Inference DS User Interface after downloading it (see releases below) to connect to a running Inference DS Core.

In order to run the Inference DS Core on Windows, Python is required. The latest release of the Core is available for Python 3.6, 3.7, and 3.8 on Windows. Python can be downloaded directly from python.org or via distributions such as anaconda.

After successfully installing Python, you can install the Core using the package manager pip. Download the correct version of the inds-core package for your platform and install it using one of the following commands:

CPU version:

# exchange X.X.X with the correct version number of inds-core and YY with
# the Python version, e.g. 38 for Python 3.8, or 37m for Python 3.7
python -m pip install inds_core-X.X.X-cpYY-cpYY-win_amd64.whl[cpu]
POWERSHELL

GPU version:

# exchange X.X.X with the correct version number of inds-core and YY with
# the Python version, e.g. 38 for Python 3.8, or 37m for Python 3.7
python -m pip install inds_core-X.X.X-cpYY-cpYY-win_amd64.whl[gpu]
POWERSHELL

Offline Installation

In case you need to do an offline installation, you can download all required packages using the command python -m pip wheel inds_core_[...].whl and copy them to the offline system.
Install the package from the downloaded wheels via:
python -m pip install inds_core_[...].whl --no-index -f path/to/wheels

Both, the CPU and the GPU version, are based on inferencing using the ONNX Runtime. In order to use the GPU version, you need to install Nvidia CUDA 10.2 and cuDNN 8.

Once you have installed the Core and its dependencies, you can run it via:

python -m inds.app
POWERSHELL

It can be shut down using CTRL+C which can take a few seconds until the graceful shutdown finishes.

Linux

You can directly use the Inference DS User Interface after downloading it (see releases below) to connect to a running Inference DS Core. The User Interface is provided as an App Image.

In order to run the Inference DS Core on Linux, Python is required. The latest release of the Core is available for Python 3.6, 3.7, and 3.8 on Linux. The installation process depends on your platform. For example, you can install Python 3.7 on Ubuntu 18.04 using the package manager apt:

sudo apt update

# install python 3.7 and pip
sudo apt install python3.7
sudo apt install python3-pip

# upgrade pip to the latest version
python3.7 -m pip install --upgrade pip
BASH

After successfully installing Python, you can install the Core using the package manager pip. Download the correct version of the inds-core package for your platform and install it using one of the following commands:

CPU version:

# exchange X.X.X with the correct version number of inds-core
python3.7 -m pip install inds_core-X.X.X-cp37m-cp37m-linux_x86_64.whl[cpu]
BASH

GPU version:

# exchange X.X.X with the correct version number of inds-core
python3.7 -m pip install inds_core-X.X.X-cp37m-cp37m-linux_x86_64.whl[gpu]
BASH

Offline Installation

In case you need to do an offline installation, you can download all required packages using the command python -m pip wheel inds_core_[...].whl and copy them to the offline system.
Install the package from the downloaded wheels via:
python -m pip install inds_core_[...].whl --no-index -f path/to/wheels

Both, the CPU and the GPU version, are based on inferencing using the ONNX Runtime. In order to use the GPU version, you need to install Nvidia CUDA 10.2 and cuDNN 8.

In case you get a warning that a certain version of a dependency such as onnxruntime-gpu is not available for your system on https://pypi.org, make sure that you have the latest version of pip installed. This can be assured via python3.7 -m pip install --upgrade pip

Once you have installed the Core and its dependencies, you can run it via:

python3.7 -m inds.app
BASH

It can be shut down using CTRL+C which can take a few seconds until the graceful shutdown finishes.

Nvidia Jetson Platform

You can use the Inference DS User Interface after downloading it (see releases below) to connect to a running Inference DS Core. The User Interface is provided as an App Image.

To run the user interface, make sure that the library libgconf is installed:

sudo apt update
sudo apt install libgconf-2-4
BASH

You can install the Inference DS Core on the Nvidia Jetson Platform (TX2, Nano, Xavier) by downloading the Python package and all required packages. Once downloaded, extract the required packages and install the package using the following command:

# exchange X.X.X with the correct version number of inds-core
python3 -m pip install inds-core[gpu] --no-index -f path/to/inds-core-X.X.X-jetson-dependencies
# install pypylon
python3 -m pip install pypylon --no-index -f path/to/inds-core-X.X.X-jetson-dependencies
BASH

Releases

All releases are available for download at https://www.data-spree.com/resources/releases