This repository contains the code for all experiments in the submitted manuscript. The dataset download will be handled automatically by the plant-celltype-graph-benchmark.
- Linux
- Anaconda / miniconda
- python >= 3.8
- ctg-benchmark
- tiffile
- scikit-image
- scikit-spatial
- python-elf
- pytorch-lightning
- for cuda 11.3
conda create -n pct -c rusty1s -c lcerrone -c pytorch -c conda-forge ctg-benchmark cudatoolkit=11.3 tifffile scikit-image scikit-spatial python-elf pytorch-lightning
- for cuda 10.2
conda create -n pct -c rusty1s -c lcerrone -c pytorch -c conda-forge ctg-benchmark cudatoolkit=10.2 tifffile scikit-image scikit-spatial python-elf pytorch-lightning
- for cpu only
conda create -n pct -c rusty1s -c lcerrone -c pytorch -c conda-forge ctg-benchmark cpuonly tifffile scikit-image scikit-spatial python-elf pytorch-lightning
Additional dependencies
pip install class_resolver
With the pct environment active, executed from the root directory:
pip install .
pip install 'napari[pyqt5]'
pip install plotly==5.0.0
All experiments reported in the manuscript are self-contained in experiments, please check the
README.md inside the experiment directory for additional instructions.
Features can be computed from segmentation by running:
python run_dataprocessing.py -c example_config/build_dataset/CONFIG-NAME.yaml
To run prediction on new segmentation data using a pretrained model
- Configure the pipeline by editing the prediction config.
- Run the pipeline by:
python run_dataprocessing.py -c example_config/node_predictions/predict_from_segmentation.yaml
@inproceedings{cerrone2022celltypegraph, title={CellTypeGraph: A New Geometric Computer Vision Benchmark}, author={Cerrone, Lorenzo and Vijayan, Athul and Mody, Tejasvinee and Schneitz, Kay and Hamprecht, Fred A}, booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition}, pages={20897--20907}, year={2022} }