A segmentation plugin to adapt Omnipose implementation to partial labelling.
A plugin to adapt the Omnipose implementation to frugal labeling. It aims to facilitate the training from scratch or the use of transfer learning with little data, by not needing to draw entire cells, but a few squiggles instead (see GIF below).
Image Credit: Eduard Muzhevskyi¶
First, we advise you to create a conda environment in Python 3.10, in which you will run Napari:
conda create -n sketchpose_env python=3.10 conda activate sketchpose_env conda install pip python -m pip install "napari[all]" --upgrade
You can install
napari_sketchpose via pip:
pip install napari_sketchpose
For Windows users, CUDA version of PyTorch may not be installed properly. When the plugin starts for the first time, it checks whether CUDA version is installed. If not, it tries to install it using light-the-torch library. If this does not work, you should re-install CUDA torch and torchvision versions manually, otherwise the plugin will not work properly.
We strongly recommend reading the documentation to get the most out of the plugin. A step-by-step tutorial illustrated with GIFs will guide you through the various stages.
Contributions are very welcome. Tests can be run with tox, please ensure the coverage at least stays the same before you submit a pull request.
Distributed under the terms of the GNU GPL v3.0 license, "napari-sketchpose" is free and open source software
If you encounter any problems, please [file an issue] along with a detailed description.
- 27 November 2023
- 07 November 2023
- Information not submitted
- Stars: 0
- Forks: 0
- Issues + PRs: 0