- Code for fast watersheds. Code is based around code from https://bitbucket.org/poozh/watershed described in http://arxiv.org/abs/1505.00249. For use in https://github.com/naibaf7/PyGreentea.
# building
### conda
- `conda install -c conda-forge zwatershed`
### pip [<img src="https://img.shields.io/pypi/v/zwatershed.svg?maxAge=2592000">](https://pypi.python.org/pypi/zwatershed/)
- `pip install zwatershed`
### from source
- clone the repository
- run ./make.sh
### requirements
- numpy, h5py, cython
- if using parallel watershed, also requires multiprocessing or pyspark
- in order to build the cython, requires a c++ compiler and boost
# function api
- `(segs, rand) = zwatershed_and_metrics(segTrue, aff_graph, eval_thresh_list, seg_save_thresh_list)`
- *returns segmentations and metrics*
- `segs`: list of segmentations
- `len(segs) == len(seg_save_thresh_list)`
- `rand`: dict
- `rand['V_Rand']`: V_Rand score (scalar)
- `rand['V_Rand_split']`: list of score values
- `len(rand['V_Rand_split']) == len(eval_thresh_list)`
- `rand['V_Rand_merge']`: list of score values,
- `len(rand['V_Rand_merge']) == len(eval_thresh_list)`
- `segs = zwatershed(aff_graph, seg_save_thresh_list)`
- *returns segmentations*
- `segs`: list of segmentations
- `len(segs) == len(seg_save_thresh_list)`
##### These methods have versions which save the segmentations to hdf5 files instead of returning them
- `rand = zwatershed_and_metrics_h5(segTrue, aff_graph, eval_thresh_list, seg_save_thresh_list, seg_save_path)`
- `zwatershed_h5(aff_graph, eval_thresh_list, seg_save_path)`
##### All 4 methods have versions which take an edgelist representation of the affinity graph
- `(segs, rand) = zwatershed_and_metrics_arb(segTrue, node1, node2, edgeWeight, eval_thresh_list, seg_save_thresh_list)`
- `segs = zwatershed_arb(seg_shape, node1, node2, edgeWeight, seg_save_thresh_list)`
- `rand = zwatershed_and_metrics_h5_arb(segTrue, node1, node2, edgeWeight, eval_thresh_list, seg_save_thresh_list, seg_save_path)`
- `zwatershed_h5_arb(seg_shape, node1, node2, edgeWeight, eval_thresh_list, seg_save_path)`
# parallel watershed - 4 steps
- *a full example is given in par_ex.ipynb*
1. Partition the subvolumes
- `partition_data = partition_subvols(pred_file,out_folder,max_len)`
- evenly divides the data in *pred_file* with the constraint that no dimension of any subvolume is longer than max_len
2. Zwatershed the subvolumes
1. `eval_with_spark(partition_data[0])`
- *with spark*
2. `eval_with_par_map(partition_data[0],NUM_WORKERS)`
- *with python multiprocessing map*
- after evaluating, subvolumes will be saved into the out\_folder directory named based on their smallest indices in each dimension (ex. path/to/out\_folder/0\_0\_0\_vol)
3. Stitch the subvolumes together
- `stitch_and_save(partition_data,outname)`
- stitch together the subvolumes in partition_data
- save to the hdf5 file outname
- outname['starts'] = list of min_indices of each subvolume
- outname['ends'] = list of max_indices of each subvolume
- outname['seg'] = full stitched segmentation
- outname['seg_sizes'] = array of size of each segmentation
- outname['rg_i'] = region graph for ith subvolume
4. Threshold individual subvolumes by merging
- `seg_merged = merge_by_thresh(seg,seg_sizes,rg,thresh)`
- load in these areguments from outname
Metadata
Release files for zwatershed 0.10
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| zwatershed-0.10.tar.gz | 29.6 kB | Without provenance | Details for zwatershed-0.10.tar.gz |
Release files / zwatershed-0.10.tar.gz
| Download URL | zwatershed-0.10.tar.gz |
|---|---|
| Size | 29.6 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
4892f63d5aefb7dd5ae199c4481e94c88ce7e398d30cd90fa45fa555587ff5c1
|
|
BLAKE2b-256 checksum How to use checksums |
fe4bfa666cc4c5921bb19250087ce200360923969aaaa902795f7a8794da41e9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
Security for zwatershed 0.10
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.