The code in this repository implements object detection with star-convex polygons as described in the paper:
Uwe Schmidt, Martin Weigert, Coleman Broaddus, and Gene Myers.
Cell Detection with Star-convex Polygons.
International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), Granada, Spain, September 2018.
Please cite the paper if you are using this code in your research.
This package requires Python 3.5 (or newer) and can be installed with pip
:
pip install stardist
- Depending on your Python installation, you may need to use
pip3
instead ofpip
. - Since this package relies on a C++ extension, you could run into compilation problems (see Troubleshooting below). We currently do not provide pre-compiled binaries.
- StarDist uses the deep learning library Keras, which requires a suitable backend (we only tested TensorFlow).
We provide several Jupyter notebooks that illustrate how this package can be used.
Installation requires Python 3.5 (or newer) and a working C++ compiler. We have only tested GCC (macOS, Linux), Clang (macOS), and Visual Studio (Windows 10). Please open an issue if you have problems that are not resolved by the information below.
If available, the C++ code will make use of OpenMP to exploit multiple CPU cores for substantially reduced runtime on modern CPUs. This can be important to prevent the function star_dist
(utils.py) from slowing down model training.
Although Apple provides the Clang C/C++ compiler via Xcode, it does not come with OpenMP support.
Hence, we suggest to install the OpenMP-enabled GCC compiler, e.g. via Homebrew with brew install gcc
. After that, you can install the package like this (adjust names/paths as necessary):
CC=/usr/local/bin/gcc-8 CXX=/usr/local/bin/g++-8 pip install stardist
Please install the Build Tools for Visual Studio 2017 from Microsoft to compile extensions for Python 3.5 and 3.6 (see this for further information). During installation, make sure to select the Visual C++ build tools. Note that the compiler comes with OpenMP support.