Installation

August 4, 2025 ยท View on GitHub

For the installation, a version of Python >= 3.9 is required.

Dependencies

It is recommended to install the dependencies of mrSegmentation through a virtual environment.

Tip

Meshroom expects the virtual environment to be named "venv", otherwise it will not be recognised.

Note


By default, some of the dependencies in the requirements.txt file install a CPU-only version of PyTorch. The plugin will run smoothly on the CPU, but for better performances, it is advised to install the GPU-enabled version of PyTorch.

  • On Linux:
cd /path/to/mrSegmentation
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

# For GPU support (optional)
pip install -r torch-requirements.txt --upgrade --force-reinstall

deactivate
  • On Windows:
cd /path/to/mrSegmentation
python -m venv venv
.\venv\Scripts\activate.bat
pip install -r requirements.txt

# For GPU support (optional)
pip install -r torch-requirements.txt --upgrade --force-reinstall

deactivate

Models

Downloading models

Models are available on the following repository.

git clone -n --depth=1 --filter=tree:0 https://gitlab.com/alicevision/imagepromptnnmodels
cd imagepromptnnmodels
git sparse-checkout set --no-cone /models
git checkout

It is recommended to move the models folder to the mrSegmentation repository, at the top-level.

Setting up environment variables

The following environment variables need to be set for the nodes to find the appropriate models and use the correct settings:

TRANSFORMERS_OFFLINE=TRUE
RDS_DETECTION_MODEL_PATH=/path/to/models/groundingdino_swint_ogc.pth
RDS_RECOGNITION_MODEL_PATH=/path/to/models/ram_plus_swin_large_14m.pth
RDS_SEGMENTATION_MODEL_PATH=/path/to/models/sam_vit_h_4b8939.pth
RDS_TOKENIZER_PATH=/path/to/models/tokenizer

Setting up mrSegmentation for Meshroom

  • On Linux:
export MESHROOM_PLUGINS_PATH=/path/to/mrSegmentation:$MESHROOM_PLUGINS_PATH
  • On Windows:
set MESHROOM_PLUGINS_PATH=/path/to/mrSegmentation;%MESHROOM_PLUGINS_PATH%