A deep learning based tool to segment epithelial tissues. The epyseg GUI can be uesd to build, train or run custom networks
Project description
EPySeg
EPySeg is a package for segmenting 2D epithelial tissues. EPySeg also ships with a graphical user interface that allows for building, training and running deep learning models. Training can be done with or without data augmentation (2D-xy and 3D-xyz data augmentation are supported). EPySeg relies on the segmentation_models library. EPySeg source code is available here. Cloud version available here.
Install
-
Install Miniconda (if not already present on your system)
-
Then in a command prompt type:
conda create -y -n TA python==3.10.12
conda activate TA
pip install -U pip
pip install -U epyseg
python -m epyseg
NB:
- To open a command prompt on Windows press 'Windows'+R then type 'cmd'
- To open a command prompt on MacOS press 'Command'+Space then type in 'Terminal'
Run
To open the graphical user interface, type the following in a command:
conda activate TA
python -m epyseg
Third party libraries
Below is a list of the 3rd party libraries used by EPySeg and/or pyTA.
IMPORTANTLY: if you disagree with any license below, please uninstall EPySeg.
| Library name | Use | Link | License |
|---|---|---|---|
| tensorflow | Deep learning library | https://pypi.org/project/tensorflow/ | Apache 2.0 |
| segmentation-models | Models | https://pypi.org/project/segmentation-models/ | MIT |
| czifile | Reads Zeiss .czi files | https://pypi.org/project/czifile/ | BSD (BSD-3-Clause) |
| Markdown | Python implementation of Markdown | https://pypi.org/project/Markdown/ | BSD |
| matplotlib | Plots images and graphs | https://pypi.org/project/matplotlib/ | PSF |
| numpy | Array/Image computing | https://pypi.org/project/numpy/ | BSD |
| Pillow | Reads 'basic' images (.bmp, .png, .pnm, ...) | https://pypi.org/project/Pillow/ | HPND |
| PyQt5 | Graphical user interface (GUI) | https://pypi.org/project/PyQt5/ | GPL v3 |
| PyQt6 | Graphical user interface (GUI) | https://pypi.org/project/PyQt6/ | GPL v3 |
| read-lif | Reads Leica .lif files | https://pypi.org/project/read-lif/ | GPL v3 |
| scikit-image | Image processing | https://pypi.org/project/scikit-image/ | BSD (Modified BSD) |
| scipy | Great library to work with numpy arrays | https://pypi.org/project/scipy/ | BSD |
| tifffile | Reads .tiff files (also reads Zeiss .lsm files) | https://pypi.org/project/tifffile/ | BSD |
| tqdm | Command line progress | https://pypi.org/project/tqdm/ | MIT, MPL 2.0 |
| natsort | 'Human' like sorting of strings | https://pypi.org/project/natsort/ | MIT |
| numexpr | Speeds up image math | https://pypi.org/project/numexpr/ | MIT |
| urllib3 | Model architecture and trained models download | https://pypi.org/project/urllib3/ | MIT |
| qtawesome | Elegant icons in pyTA | https://pypi.org/project/QtAwesome/ | MIT |
| pandas | Data analysis toolkit | https://pypi.org/project/pandas/ | BSD (BSD-3-Clause) |
| numba | GPU acceleration of numpy ops | https://pypi.org/project/numba/ | BSD |
| elasticdeform | Image deformation (data augmentation) | https://pypi.org/project/elasticdeform/ | BSD |
| CARE/csbdeep | pyTA uses custom trained derivatives of the CARE surface projection model to generate (denoised) surface projections | https://pypi.org/project/csbdeep/ | BSD (BSD-3-Clause) |
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
