Installation guide

This guide shows how to manually install the Toolbox and its requirements as a Python package.

Requirements:

• Git

• Python [3.9 - 3.10]

• Optional GPU support: CUDA >=10.1

1. Clone the repository and navigate to the project directory

   git clone https://github.com/CommuniCityProject/communicity_toolbox.git
   cd communicity_toolbox
   

2. Download the Toolbox’s data from the repository releases, which includes machine learning models and additional resources. Download and extract it manually or run the following script:

   python download_data.py
   

3. Install the Python requirements

   python -m pip install -r requirements.txt
   

4. Install Detectron2

   python -m pip install --user 'git+https://github.com/facebookresearch/fvcore'
   git clone https://github.com/facebookresearch/detectron2 detectron2
   python -m pip install --user -e detectron2
   

5. Install the Toolbox

   python -m pip install -e .
   

Optional: Install Orion-LD

Some Toolbox components require an NGSI-LD context broker to function. The recommended one is Orion-LD. You can launch an instance with Docker Compose using the provided Orion-LD.yaml docker compose file.

Requirements:

• docker-compose

docker compose -f Orion-LD.yaml up -d